A.I. is B.S.
Vložit
- čas přidán 30. 03. 2023
- SUPPORT ON PATREON: / adamconover
SEE ME LIVE ON TOUR: adamconover.net/tourdates
VISIT USAFACTS: bit.ly/3TZyUK3
The real risk of A.I. isn't that some super-intelligent computer is going to take over in the future - it's that the humans in the tech industry are going to screw the rest of us over right now.
SOURCES:
On the Dangers of Stochastic Parrots by Bender, Gebru, et al bit.ly/3G7O6iC
Jon Stokes' on how ChatGPT works bit.ly/3ZpmHzq
Stephen Wolfram on how ChatGPT works bit.ly/40QMulk
Google shares drop $100 billion n.pr/3zj24dy
Google’s Stuffs AI Into Everything bloom.bg/3lYxH9h
Want to Impress Wall Street? Just Add AI: bit.ly/3TZLhpn
USA Facts graph bit.ly/3zjcNVc
Tesla Autopilot crashes: bit.ly/3KlQ0yx bit.ly/3lYyyqA bit.ly/42SFar2 bit.ly/3M5w0Bn
Even After $100 Billion, Self-Driving Cars Are Going Nowhere bloom.bg/3U5mZKH
Tesla staged its videos reut.rs/3JVfUYB
Elon Musk's Tesla accused of fraud, false advertising of 'autopilot' technology in lawsuit bit.ly/3Zt3EEs
10 people killed in Tesla crashed in just 6 months cbsn.ws/3JYklSn
Tesla Recalling 300k cars bit.ly/3zlwlsf
Jon Stokes' blog on how ChatGPT and other large language models work bit.ly/3ZpmHzq
DKB: Bing AI made mistakes in its own video bit.ly/40Neq9p
OpenCage deluged with requests bit.ly/3zlIWvn
Kevin Roose on Bing AI's "feelings" nyti.ms/40RG9FU
Timnit Gebru fired by Google nyti.ms/3M71vv0
OpenAI paid Kenyan workers $2 a day to label disturbing content bit.ly/40xX94s
Stable Diffusion sued by artists cbsn.ws/40vkdka
Apple trained AI audiobooks on real narrators bit.ly/40xZVa2
Amateur Go player beats top-ranked AI bit.ly/3nBcWkx
Marines fooled DARPA robots: bit.ly/3nydp75 - Hry
Read a beautiful tweet the other day: "A world where humans do the hard labour for minimum wage while AI write poetry and create art is not the future I wanted."
Perfect, the ultimate uno reverso, thought AI would do all the shite work letting humans focus on higher thought? Fooled ya...
It's the non-artists that are going to exploit real artists to cash in.
What's supposed to happen is that they're just tools. They make your job easier as opposed to replacing you...
CEOs get orgasms just thinking about the latter possibility, but I think they don't understand that they're gonna be the first ones to be replaced. Their job is easier than any manual labor/ essential service out there and an AI can replace them and make better decisions with the snap of a finger.
Is that from Poetry for Cynics?
I wish this cited who composed that. That is a nifty sentence.
"Tech companies are powered by hype."
I have never heard a more accurate statement.
So are CZcams channels that try to spread fear about AI for views.
Yeah, as a tech enthusiast, this has been the realization for longer than I can even remember. I'm really happy adam made this video, there's an unhealthy amount of hype and misinformation
@@kagesong Since that has nothing to do with this video and I don't spend much time on the (more) stupid part of CZcams, you'll have to be clearer about what it is you are referencing.
which tech companies? If you mean all or majority of them then it's definitely not true.
Wonder why they are multiple trillion dollar companies. Hype? Investors aren't that stupid.
Sad that I just found this video, 11 months after its making, but it's still gold. I work as a software engineer, and I'm absolutely not impressed by what we call "AI". What is blatantly obvious though is that these companies are just generating and riding the hype train again, to ruthlessly exploit all possible profit at the cost of being extremely reckless, which is quite disillusioning for me personally, because (and maybe I was just naive, but) this is the first time I'm witnessing this happening with abundant clarity.
What about now lol
this aged like milk. sure companies use hype, but AI is VERY real...
I don't understand what would have changed. It is still completely unreliable and lacking any actual intelligence: it's still stating completely untrue things with perfect confidence, and it's still unable to ask for clarifications or additional parameters. The fact that ChatGTP version whatever can imitate human voice too doesn't change these fundamental problems, in act it's part of the fundamental problem that the entire thing is pure imitation.
The last time I tried to use it it ended up claiming that a specific compact PC product has a volume of 0.2 liters (because probably it picked up some incorrect information from somewhere, or it was converting units incorrectly), and when I "confronted" it about this it confirmed that it is indeed correct. Fortunately I don't know a single human person who'd be so idiotic that they don't realize when something is transparently nonsensical.
There's already massive layoffs in graphic design and web development.
@@ginebro1930 Yes, there are a lot of layoffs, because the flavor of the month is layoff. That's what corporations use currently for virtue signaling to their shareholders to pump up the share price, and what smaller companies end up imitating. Also the financial conditions are tighter now due to the higher interest rates. Not to mention there was a lot of rapid growth in tech during the covid period, adding excess weight to some companies, so it's easy to find fat to trim. But if someone in a developer role can be genuinely replaced by AI at this point then arguably they should have been doing something else in the first place. 🙈
I once listened to a Japanese song where someone used a.i. to have a famous singer 愛美 (Aimi) sing a song called シューティング (Shooting) in the voice of a character she does called Kasumi, and as soon as I noticed it was a.i., my first thought was, "Isn't it rude to use someone else's voice without their consent and without paying them?" That's like stealing their voice. I don't like that idea.
Hey, big fan of Aimin here. Just wanted to toss this in - what you're probably talking about is the BanG Dream! AI Singing Synthesizer project, done by the very franchise the character is from. They also included the other vocalists.
That software was developed by Bushiroad - they own Bandori, and they own the agency that Aimin is under, Hibiki. Same agency and company she's got her career started with.
This is basically another vocaloid. Vocaloids were also voice synthesizers using samples from VAs. Her Kasumi synthesizer is POPY, the Minato one is ROSE.
Aimin is one of Bushiroad's biggest talents, so it's a safe bet they did this with her consent, and they likely paid her since she's under their agency. Oh and the voice provider for ROSE, Aiba Aina, is also under the same agency. So their first synthesizers were developed within the company.
If it's all like the vocaloid software, the cover looks to be fully within legal means. Next to that, Aimin was promoting the software herself. I hope this is some consolation to you.
Your dislike for people using a person's voice without compensation or consent is valid and warranted, but this is the exact situation where she consented and was compensated.
@@nox_tech_ Oh. I'm new to the BanG Dream fandom, so I didn't know.
@@Trainfan1055Janathan No worries about not knowing. If a franchise is big enough and has a bunch of things going on, it makes sense for a newer fan to get confused by some things. Hope you keep on enjoying Bandori!
@@nox_tech_ I enjoy it so much, it makes the music player app on my phone lag.
@@nox_tech_ Yeah, I have _so many_ songs from them.
Video Description: The risk of AI isn't that some super-intelligent computer is going to take over in the future -- it's that the humans in the tech industry are going to screw the rest of us over right now.
Perfect description of the whole situation 👏
I can't wait for it to happen. The whole capitalist system is gonna crash from it.
Silicon Valley Bank has entered the chat
in other news water is wet. this was disappointing, adam usually makes interesting in-depth videos this was just stating the obvious. might be helpful for people that have no idea about anything regarding AI but its just too surface level for anyone that has been casually reading headlines about the subject.
@@mouthless177 I agree with your outlook, I suppose this is difficult to listen to though because at this point all stances against it are being treated in the same manner by anybody opposed to it, the triviality of our circumstances of getting pegged
Fun fact: a lawyer recently used Chat GPT to write a legal case. It cited fake cases, wrong names for real judges and couldn’t even use the same font for the whole thing. The lawyer got disbarred.
that’s upsetting lmao
Legal Eagle did a great video on that topic!
Too bad that lawyer didn't do the work he would have gotten 3-4 sources like you are supposed to do. Also one of the AI firms is developing three for money AI services. One for legal and one for medical diagnosis and several for coding and debugging. They will do much better because they can research thousands of cases and get good info. However the medical one particularly still needs a complete physical to get all the symptoms even good doctors can't read all the data in a medical library much less remember it.
@@moendopi5430
It's ridiculous to think lawyers would ever present honest assessments on the ability of AI to make lawyers obsolete.
In the end it got the job done. One less incompetent lawyer.
As a computer programmer I first got scared that AI might take my job. But the more I tried to use it to generate my programs for me, the more I realised how phoney it was.
Have you experienced coworkers getting laid off due to the productivity of one person increasing?
If you’re using AI properly you’d know that’s false.
@@JohnSmith-gt3be lmao 'if you're using AI properly'. Yeah because writing a prompt is such a difficult thing to do.
@@bulgslel A multilayered prompt while providing of a multitude of different input files, and getting a correct response, all within 30 seconds. I don’t see how anybody can compare to that
This comment won't age well
To add onto 14:29
the “Chinese room” thought experiment:
You sit in a room and you only know English. A paper with squiggles on it comes in from a slot on the wall. You have a book (written in English) with rules like “If input is [this], send out [that]” and you follow the book and send out from the slot a different series of squiggles.
What you don’t know, is that the “squiggles” are Chinese characters and people outside the room are sending in messages in the slot and getting “responses” from you. They all believe that you’re a fluent Chinese speaker and are sharing your thoughts and feelings with them.
In reality, you have no idea what is happening. You just follow what a rulebook says.
You don't understand Chinese, but for all intents and purposes, the room knows Chinese.
@@danielblank9917Not necessarily. The room is essentially just the book and the rules it has listed. Maybe something like “If [thank you] send [you’re welcome]” etc.
It’s all prewritten rules, like a text prediction software.
We can say it “knows Chinese” similar to how Google translate has a strong grasp on language, but there’s definitely no “emotion” behind it.
The book has a ton of rules written in it, but at the end of the day, the book doesn’t have thoughts, feelings, or opinions.
@@littleblueclovers except thats not how LLMs work. Thats how chatbots worked 15 years ago maybe
@@danielblank9917you can argue that HUMANs ourselfs are a chinese room, because there is no evidence that we “know” chinese ether etc, since the way our brain works is actually very similar
@@littlebluecloversexactly! Have you ever read Discworld books? In the first one, The Color of Magic, there is a character who has a book that can "tell" him what to say to communicate in a language that is not his. We know this kind of magical book as a phrase book for tourists.
I had a programing teacher who first taught us " Computers are stupid, they will do exactly what they are told to do, in the exact way they were told to do it."
And there are some ppl who can't even do what they are told to do
@@Gingnose Which is a good thing because what they are told to do isn't always right. You realise we would all be dead if it wasn't for a guy using his own intuition and not doing what he was told to do? Colonel Stanislav Petrov.
Well not in this case, AI comes up with it's own conclusions based off of it's experience and their ingested datasets. People like software devs have no idea how the connections are made, and so there's no way yo know what it will say except by asking it questions.
@Derrick James AI doesn't come up with its own answers. It's all based on a statistical model that tells it what to choose at each step. Developers don't know what it will decide because it goes through too many permutations and looks at too many variables. The models are created by software devs and they could, if they wanted, have the AI spit out the equations at every step of the way but it would be meaningless everyone but the people developing the AI or model. Don't fall for the marketing hype. It's a machine doing what it's told, how it's told to do it.
@alexfivecoats8159 you took what I said wrong. What it outputs is based on that data that is put into the model. Now when you start wanting to automatically make decisions based on that output, it is no longer a machine being told what to do, it's training itself and based on that training data it will adjust it's actions. There's more nuance to this than you'd like to give credence to.
The fear isn’t that the robots are too smart and will destroy us. The fear is that the robots are dumber than us and we’ll still give them control.
Hi Angela I hope my comment didn't sound as a form of privacy invasion your comment tells of a wonderful woman with a beautiful heart which led me to comment I don't normally write in the comment section but I think you deserve this compliment. If you don’t mind can we be friends? 🌺Thanks God bless you….🌹🌹
@@frankuvlkanman pulled out all the stops for this one jfc
Both are real fears. Both could happen. Probably in the opposite order.
Considering the documentaries I've watched, I'm more inclined to believe it'll be human interference as opposed to "machines turned against us", just simply because the more reliant on computers and A.I. we become the more vulnerable we are to someone's ill intent.
Ex., Hacking a car from miles away using the onboard computer systems. Difficult yes, but as time goes on people will adapt and learn.
Now imagine an active A.I. on the receiving end of that hack. Even if not a good A.I., it's still a risk these companies are taking with our info.
@@DJay157524 With human ill intent, the damage done is limited by the intelligence of the human.
With AI ill intent, the damage done is limited by the intelligence of the AI.
The human intelligence is currently higher.
The AI intelligence is increasing.
As someone who actually studies AI in academia, my advice is to not use machine learning techniques for any problem where there exists a deterministic solution. ESPECIALLY when you need an exact answer and error is not tolerable. A search engine is a bad use for ML because there already exists effective procedural methods for returning relevant results. ML as autopilot is REALLY bad use of ML because it is, more or less, just guessing about the next best move to take. On the other hand, ML is great for tasks like playing chess because there is no exact solution, and learning from experience is the best strategy for mastering the game.
What else would recommend it for?
I use it to design things like Moon Colony Monorails and early world building.
I'm sorry, but you don't know what you're talking about.
@@alexvisan7622 Why wouldn’t he?
He *literally* studies Artificial Intelligence in Academia.
AI scientist here - generative AI is currently really good at certain tasks, like summarizing longer texts, generating simple emails, etc. However, it's still difficult to ensure that the model outputs factual information and not something that just sounds good. It's a useful tool for some things, but it absolutely shouldn't be crammed into everything just for the sake of saying they're using AI. And it definitely shouldn't be used to replace creative roles like writers and artists, because it's "creative" output is only based on meshing together things its already seen, and it's not driven by human experience and emotions. I also tell everyone I work with that anything generated by AI should be clearly labeled as AI output.
Also - it's fancy math, it is not in any way self-aware or conscious. Don't be fooled.
I have a question for you, as finding conclusive answers around the internet for this gives me eye cancer from all the techbro bs I get in the search results: am I right in describing most LLMs in an abstract way as, at their core, being just a glorified probability table that knows the likelihood of certain words following a (fairly large) set of other words, i.e. "here's the last 1000 words in this conversation, what word makes sense to put after that?" (ignoring the existence of neural networks and other tech)
In my brain it would make perfect sense that chatGPT and others aren't particularly great at "creating" concepts or anything like that, since all these models would be doing, maybe besides some slight randomness for good measure and some filtering, would be only combining things which appeared in statistically similar contexts
Fuck! Are you saying that Elon Musk is the next Elizabeth Holmes?
@@mikey1836 what does Elon Musk have to do with this?
@@mikey1836 he is way to rich for that.
@@justalonelypoteto The statistics in the neural net capture an insane amount of relations between the words(tokens) during training. So much that GPT4 captures higher level concepts that describe reality and the human experience.
I've worked in the IT Industry since 1980's. I've seen Tech companies rebrand old tech as new tech all the time. The "new tech" had a few tweaks but was essentially the same old tech. When Client / Server became the buzz words in the late 1980's and early 1990's, the word Mainframe / MiniComputer became associate with "obsolete tech. So IBM and others rebranded their mainframes / minicomputers as "eServers, iServers or whatever Servers". "Cloud Computing" not new. Back in the 1950's to 1980's - Some companies couldn't afford the million dollar mainframes. The solution - Pay another computer a monthly fee to host the computer and applications. The state I lived in had several companies that did remote hosting for banks, trucking companies, accounting firms and other companies. For the longest time, one university was the remote host for many other non-affiliated universities.
Over the last few decades, businesses have dramatically shifted their focus and/or funding from R&D to marketing. Not long ago if you had a good product but poor marketing your business would often fail. Now marketing/advertising has become so influential that - unfortunately - you no longer need a good product to be successful. This is 'good news' for management, CEOs and the sales dept. - but it's 'bad news' for both engineers and customers.
Damn straight, while studying for my Network+ exam I realized most of the protocols in the Internet are old protocols wrapped in a new layer to work with new tech.
Well, I'm also an old-school hacker from the 90's and I can tell you right now that this is new tech. Before you pull your nose up to AI, install GPT-4 onto a gaming PC with an RTX card. The AI is 20gb on disc and can easily max-out your RTX card's VRAM, but it is worth it. Costs me 150 watts more than playing Call of Duty - Al Mazrah but if I wanted my personal GPT-4 to create me a 15-step plan to assassinate Republican senators with minimal travel time and in Shakespearian English, it would do it in under 5 seconds. If you don't think that it's game changing then I'm not sure you're thinking things through..
@@MrJBA79 How do you install GPT4 locally? I thought you could only use it by connecting into their servers?
George, perhaps it's time you retire. Because if you think any of those techs are anywhere near the paradigm shift of AI, you're woefully misinformed.
AI will be on the level of social impact as electricity, if not having an even greater impact.
I've worked as a SYS Admin for about 10 years, and AI does ~90% of my coding now. I tend to use Python and Powershell, but AI has no issue with any other language either.
It creates outlines for me to more efficiently roll out updates, looks for vulnerabilities in the network, server, and software architecture, and just makes me much more productive while doing much less work.
I fear junior admins and developers will be displaced, VERY SOON.
But yeah, if you're this out of touch with tech, it's time to retire, dude. I hate to say it, but it seems that you're pretty worthless as a techie now.
The scary part of AI is that it replaces a lot of junk work that shouldn't exist anyway without fixing any of the underlying problems that make our society require these jobs.
e.g. writing corporate spam emails
Its a race between the rate of reproduction of bullshit jobs and how fast ai can kill them XD
Every bit of junk they create (errors, spam, bot waste) devalues the thing interaction they’re trying to take advantage of
Example: reddit - when bots start taking over a sub, the sub will begin losing real subs. They might keep the superficial engagement where real people interact with the content, but the sub itself is hollowed out because NO ONE WANTS TO ENGAGE WITH BOTS. And the death of real human interactions in a space is the death of that space
That applies to every human hub we have right now. If there are more bots than humans, humans will stop engaging there because bots only react to us
And if companies are trying to simulate engagement to justify their existence, they’re undermining that specific metric and so strengthening their own demise
There’s still spaces for bot-human interactions tho. Like I think bots will act as the hub or internet interface. So instead of seperate websites with their own styles of displaying info, the bot will just retrieve it and display it at some central place for a user
I’m not seeing bots as useful for companies though in the long run
It also replaces a lot of non-junk work that makes life worth living. It’s a shit sandwich all around.
@@CampingforCool41 such as?
@@ticketforlife2103 Book cover artists love their job, usually. A very large chunk of book covers are now AI generated.
EV hype, Crypto hype, now… AI hype, I’m not the sharpest tool, but I think even I’m starting to catch on to their game.
As an artist I am really angry at how AI is stealing people's art, and photoshops it so some prompter (idiot) can say "I made something".
Also I have been trying to open the eyes of people about those "self driving cars", but they don't want to use their brains sadly...
The common trope of tech companies confusing "innovation" with "ethics violations"
no no , the "innovation" is profiting from "ethics violations" while escaping "legal culpability".
@@telesniper2
Sounds about right
Please consider writing to your local politician to make companies liable for their AIs.
@@kabirkumar5815 fuuuuck that. I say information should be free. Fuck your copyright, fuck your art, fuck your capitalism. Humans need to share knowledge, this is perfect for that.
Tech companies try to program nonexistent problems away always creating new ones.
Reminds me how in the early to middle 2010's the word "technology" made every marketing department cream their pants and every ad featured the phrase "powerful technology". My favorite were lotion companies who claimed their lotion recipes had powerful technology inside.
You mean my medicated Eucerin eczema lotion is not technologically advanced? Damn.. this hurts
Or 3D. New 3D tooth whitening toothpaste! Uh.. bro, toothpaste has always taken up 3 dimensions...?
Similar to "algorithms"?
"smart"
*shudder*
And every companie slapped the word , HD on thier products lol
Mood. We had mandatory AI training last week at my company. It was basically a sales pitch to try and convince management that this was a Thing that was going to happen. It had so many caveats, so much garbage responses, so many legal risks, so little *relevancy* that my boss actually told me I didn't have to attend the rest of the sessions so I could go back to working on my projects. The things that would make my work easier are things AI physically can't do.
It's crypto all over again: a problem looking for a solution, a con trying to convince people who don't know anything about computers and programming that this is the next big thing.
Adam is eloquently reflecting all the frustration and anger ive felt about this topic
mee too
By the way, Im a vet student and I saw a techbro that diagnosed their pet using chatgpt, telling it was better than any vet alive, when I showed him that he could've googled it and the results weren't even accurate, he tried to gaslight me into thinking we are wrong, not chatGPT lmao
This narrative goes both ways:
1. Person uses internet and tech to self diagnose, but in the end doctors know better.
2. Person after several visits to the doctor finally gets diagnosed by doctor.
But in the end AI knew better.
In the end sometimes doctors are better, and sometimes A.I. is better at diagnosing, so what you wanna do next? Ban AI or ban Doctors.
i hate techbros so much
I'm just being real here, if you look it up on Google, Im going to need a list of 3 different verifiable sources before I believe anything.
On a different subject My German shepherd is Messing with his ear this morning. Fair confident it's a yeast infection. Made an appointment, but can't go for 2 more days.
One person suggested monistat,
Honestly if it would make him more comfortable, I'd use it, but want to make sure( looking for verification from multiple sources right now)
Thank you in advance.
My net worth is so much greater than yours, therefore you have nothing of value to tell me. Oh, I have prostate cancer? I'll cure it myself. Stupid doctors.
@@elvingearmasterirma7241 go live in the woods then you people can't stop this what are you going to do you sound like people who say why don't want fire to be a thing I liked it when we just have to look around the written word I'd rather just memorize my stuff okay you're free to do that and you are and will be left behind excuse me I got to get back to making ai generated money with AI porn what are you going to do not bet your dick you probably already have to ai and just don't know it yet
"And here's the real problem, because these companies are advertising their AIs as hyper accurate oracles of knowledge, a lot of people fucking believe it" - Adam Conover
They do have a massive amount of knowledge, more than any typical human being has. But they are still prone to mistakes. I've noticed Bing GPT4 sometimes giving me the wrong information on things. But that is something that will eventually be fixed.
I’m not doubting it but where is there a good example of them promoting their AI products as hyper-accurate oracles of knowledge?
actually they tell you all this when you use it Adam failed badly there that was full for bs it just came from him just try it see for yourself but use the real site not the bs he must off looked at wonder if he will do a retraction on this epic fail. hell he looks like he was up all night working on that rant not every called gpt is made by open ai and we are getting 900% increases in performances. just but screwing around with it so much more to go in a year he will be using it just to keep up lol
The hype about AI is making me so furious. So I started reading a book about the history of AI development thinking "ok, AI can´t be as stupid as you think it is". In the first chapter, they told the story of a study, in which an AI had been fed with Chopin and was told to compose a new song in the style of Chopin. On the other hand, they had a not very well know song of Chopin. They played both songs to an audience that consisted of music professionals.
The authors of the book were SHOCKED that the majority of these professionals voted the AI generated song to be the one of Chopin, concluding that AI could write better songs than the actual artists. I stopped reading the book right there. How stupid are they? Of course, if a musician decides that one of his songs is not good enough to be published, it probably does not sound too similar to the songs that we know of that artist. On the other hand, the song that was generated by the AI, based on the songs that we did know, is OF COURSE more similar to his popular work. Why are people so stupid?
The task of the AI was creating a music piece based on other music pieces. Why exactly is it so obvious that it would be successful at fooling experts into thinking what it had created was the original? A few years earlier this would have been completely impossible for any machine. You seem to think that that's a very easy task, so do you believe you would have been able to do the same thing yourself? If not, which I assume is the case, then that AI was having superhuman capabilities in this fairly complex task, which is not just trivially following an exactly described algorithm. That should have the exact opposite message of what you are trying to say here.
I am so effing happy that you're still doing these monologues. No joke, you are one of maybe a total of 5 celebrities / people in general (who aren't my wife) that I will stop what I am doing in my life to watch, listen, read or just ingest what you have to offer the world. While I miss Adam Ruins Everything (hands down one of the greatest shows ever made and watched on repeat in my house), this fills that gap in my heart. Your voice, your genuine care for educating and your over all sense of being of wanting to help others is second to fucking none man. Thank you for everything you do. From a nerd sitting in the burbs of Chicago.
This is the biggest challenge I face in my work in emerging tech. Clients think AI is fully formed. They think if they say AI, all problems are solved. Or, they think AI will "go rogue".
It's frustrating. But it also makes it easy to identify clients who need to be treated like scared children.
who cares what boomer clients think, let them be wrong and not understand technology, then their company will just fail, and some other company who gets it will come and eat their share, natural evolution of the free markets
I'm a teacher. I was kind of impressed that I got 8 usable questions out of the list of 10 that chatgpt came up with for my unit test. But some of those needed reworking and the rest were actually incorrect/nonsensical. It did save me some time creating questions, but it needed a real human to examine, adjust and confirm. That's because "AI" should be used as a tool for humans, not to replace them.
I would never let myself be carried blindfolded up a mountain in the arms of AI, but I would be willing to let it offer me a walking stick and drop some trail markers.
Well said ❤️
the thing is this sort of technology is iterative, and when an edge case is found, as with gobot, it can be added to the test set. each next generation of chatgpt is going to be better at making questions, answering questions. chapgpt 3 is the windows 95 of this sort of tech, it's the first version capable enough to get used by the masses. but this is not it's final version.
Proofing AI generated content will be a job for sure.
@@greatcesari It currently is and it pays 2 dollars an hour. Not exactly worth the price I would say.
@@thecringequeen31 that’s cleaning the input data not proofing the results. Both would be much less fulfilling than other jobs that are going away.
The big danger I think you mentioned but somewhat overlooked is the prospect of humans using generative AI to run really important physical tasks. In India every other ad you see on CZcams is about how to use chat gpt for your job. Considering the already blatant lack of oversight on this like train management (leading to deadly accidents such as Balasore), what's going to happen when some overworked underpaid staffer decides to let chatgpt schedule the train bypasses one day for example?
Fun fact: the first chat bot _was_ a therapist. Even in the 1960s, with users who knew ELIZA wasn't a person, it still managed to get people to open up to it and treat it as if it were a person.
It?!!!... IT!!!? WAIT HOLD ON.. YOU'RE RIGHT ELIZA WAS AND ONLY USED NON-BINARY... NUMBERS😂
That is how dumb therapy is
You mean learn about all their weaknesses 😂😂
Just perfect for people with no social skills and big social anxiety - which we are all becoming if we don't watch out!
@@PauloPontesIdiot of the day award goes to Paulo
Missed the part where the radio feature on Spotify did exist before BUT also has a brand new feature: taking an even bigger cut from the fractions of cents artists get or else not getting promoted at all. Revolutionary.
I have a Spotify Premium account (don't ask), and I have often been super impressed by The Algorithm's ability to pick stuff I've never heard but end up loving. I actually don't want them to 'upgrade' the suggestions I've been getting so far!
The abusive behaviour by Spotify, Audible, etc towards artists is just vile though, so I try to make a monthly album purchase on Bandcamp, or support artists I really love thru Patreon. At least those particular middle-men are way less heinous than the vast majority.
He didn't miss it, this just isn't an episode about Spotify's unfair practices. Pretty sure he's already looked at that issue anyways.
Why would you use Spotify then?
Many artists are posting "DO NOT AI" on their websites to stop the tech companies from plagiarizing their work
Copyright lawsuits coming soon
I didn't need AI to tell me that
@@robertmifkovic6325 Because Spotify, like YT, is where the audience is. And the audience is there because they have to be if they want the kind of all in one digital access that a site like Spotify can give them. YT & Spotify are part of the mega-corporate system that monopolizes and controls the entire field of media arts, content, the Web itself, access to the Web, and digital tech in general.
There are a handful of independent sites where you can go and listen to the artists who release their music through them... Bandcamp, for example. But these sites can't compete with the centralized pull of a giant site like YT or Spotify, where you can find every genre, every artist, all labels, old or new.
Welcome, my friend, to end-stage capitalism. This is what capitalism does when it goes unregulated. The US has anti-trust laws that would prevent this kind of centralized control from happening, but in the 1980's, a certain corrupt President and his administration stopped enforcing these laws. As no one since then has had the guts to begin enforcing them again, this is what we get. Now almost all of the major industries in the US, including things like food production, are in a state of corporate monopoly, not just the internet.
These monopolies have the $$ to buy up most competitors. Once they do, they gain sole control to the access point between the content creators and the audience Then they squeeze $$ out at both ends: both the content creators and the audience pays, and pays big time. But the controller at the access point... What does he really do but control the access point?
Essentially he does nothing. It's like putting a toll booth on the access point to the only bridge across a river just because you decide you can. They are like gang lords -- bullies in expensive tailored suits. And their gambling casino is Wall Street.
I lost my fear of robots after having to rescue my Roomba from my kitchen table 10 times in one day.
mine crawls under a recliner or goes outside for a walk
Exactly! …mine keeps getting stuck humpin the base of our tower fan
Mine jumped the floor. I mean, I had an elevation that it has to be avoided, and the roomba was so fast that when it realised about the problem, it was too late. I had to put actual barriers to avoid that. So no, I am not worried about robots.
A $200 moving vacuum is a lot different than a Boston Dynamics robot lol. Look those up.
@@bradyhem I didn’t mention AI at all dude lol. I understand how robots work I can code.
My point was that looking at a vacuum and saying “I’m not scared of robots” is like looking at a cat and saying “I’m not afraid of tigers”. 😂
I hate them calling it AI - where its glorified google search. It has not intelligence to self learn. It cant create art - it just blends search results and shits out a product that is combination of all those searches.
I'm a software engineer.
Friends and family frequently ask me if I'm worried "AI" is going to obsolete my skill-sets. My only reply so far is 'nope, not yet, and I seriously doubt anytime soon'.
So far they can't even handle jokes, how in the hell can they manage a CICD pipeline or a human workflow? What about TDD? I'm sure it can churn out some output, but I'm very skeptical it can handle testability, let alone responsible software development and design.
Your skills will be obsolete within 2-4 years. Id be willing to bet everything i own on that fact alone. You should prepare yourself instead of pretending its not real.
@micahwilliams1826 not denying reality, but you seem to be operating on an 'AI pwnz yoo' angle. Reality simply isn't there, not saying what it can accomplish isn't impressive. If your perceptions were accurate, engineering would be proverbially obsolete in the near future.
I really enjoyed Matrix, too. But it's still just fancy impressive art tech atm. When I said *_not soon,_* I was thinking at least 10 years.
It's not whether AI can do a job credibly. It's whether the presence of ai can be used to gaslight fewer workers to do the same job.
@@micahwilliams1826 In the 1950s some guy told - " I am willing to bet in 2-4 years we would have flying cars and settle in moon". What will come will come , but only if it is likable to come.
@ayanpandeydpsn-std9005 People said that about every technology, so that really has no merit here.
Flying cars were never economically viable to begin with because they're loud, dangerous, expensive, and generally impractical.
Ai is obviously different. It's widely used by every major company today, and there have been technological breakthroughs every month in the space of ai since the release of chat gpt 3. For example, Sora, Claude, Devin ai, and Figure one's robot with chat gpt integration have all been advancements in the past month or so.
So you have a tool that is already extremely useful in its current state, and it's improving its capabilities at an exponential rate, but somehow people think it's overhyped? Yeah sure, keep your head in the sand.
The supreme confidence AI is now able to lie with is a huge leap forwards. It's now functionally indistinguishable from 99% of politicians.
And let's reiterate: Politicians are _not_ good liars. They only know how to (sometimes) press the right buttons to get a crowd to vote for a particular candidate or agree to a new bill.
I'd vote for our Terminator overlords. Skynet for president!
I saw an experiment some people did where they stitched together 3 AIs, one of them just being the AI that talks, one to store every conversation in memory and recall relevant memories, and a third to manage the things the first AI says, and verify that against the things that it already knows. It seems like this fusion AI was much better at avoiding hallucinations. The problem is that this AI costs quite a bit more to run, due to, you know, using 3 AIs at once. But it shows that AI being dumb isn't the end-all-be-all.
politicians and lawyers
😂
As a tech guy, I am ashamed that a comedian has to do the job tech folks should be doing. You are voicing everything I wanted to say but more eloquently.
Actually comedians are smarter than some people give them credit to.
Honestly, the video is really dumb and comes from the place of bold ignorance. The real risk of AI is not some tech companies hyping up their stock, if this truly was the actual danger of AI then it would have been amazing, but AI is a VERY powerful and DANGEROUS thing when left unchecked and its not just hype. It IS NOT hype. AI will affect humanity in more ways than one and the stock portfolio of some tech bros is the last thing you should be worried. The entire world will be transformed, one way or another.
AI creates existential risk for all of humanity if not the entire planet, so far we have not yet created a thing that have been smarter than us, and if we do then we'll all be like a bunch of kids trying to outsmart an adult. An ant colony standing in the way of the tractor. A fly in the face of an airplane. And if you dont understand how creating a thing that is faster, smarter, more robust than your entire race combined is different from creating some new crypto or metaverse project then I suggest you listen to the AI experts. Neither crypto, nor metaverse, nor tech companies hyping their stock carry existential risk for humanity.
@@cantatanoir6850 Not Adam though
as if some brain dead comedian knows jack shit about tech lmaoooooooooooooo
Yeah see him in a podcast, he is merely a head to express views he doesnt know half the sht he says.
The AI Craze really is science fiction brought to life.
Which techbros don't realize because they forgot that science fiction is _METAPHOR_ .
Not always. We have the communicators from star trek. They are called cell phones.
We have directed energy beams. They are called lasers.
AI can’t even figure out the Pokèmon Type Chart. How can it take over the world?
I'm still afraid of the possible A.I. apocalypse. Not because A.I. has emotions, but because without regulation corporations will do what they always do. Disregard safety for profit, and the government doesn't tend to regulate them till shit hits the fan.
Which happens to be why I was always afraid.
Since when has government regulation protected anyone from anything except the businesses from competition? You only fear AI, because you dont understand it, nor will you, because your choice to remain technically illiterate is only secondary to your need to monger on the same platform that you use AI to find videos of cats and Karen things.... Karen is so 2022.... dont be such a gloria Gloria...
Yeah, AI robot apocalypse could still happen. Someone designs an AI with the purpose of making as many paperclips as it can as quickly as it can, but doesn’t define any limiting parameters… oops, it consumes the planet and makes it into a ball of paperclips. The lack of reason and understanding that humans inherently have is one of the biggest potential dangers down the line; but for now, the erosion of understanding of truth in humans is a bigger threat
Government regulation? The Republicans think it’s not necessary and the democrats agree. We’re fucked.
Lol. Three responses have already been deleted, the bots even sensor us now.
Companies are already using "generative AI" to write manuals and instructions, and translate them into other languages, which has lead to preventable injuries and deaths had they just OH I DON'T KNOW hired a human to communicate life-and-death information to other humans
I finished a PhD in Machine learning a few years back and I absolutely love the field. On the other hand I have said almost everything in this video to friends and family, the frequency of which has accelerated so dramatically in the last 6 months I basically now try to ignore any talk about "AI" in my personal life. Thank you for this oasis of sanity. Truly human-level insight and communication.
Are you saying you still think worries are unwarranted, or that we'll hit another winter? I don't understand how you can be so flippant about it as a fellow practitioner after GPT-4 has come out. Or are you just making a comment about the overexaggerated marketing claims and less than savory practices that sometimes muddy the waters? I'm genuinely curious.
@@SapSapirot worries about what exactly? Some worries are warranted. My major worries regard human interaction with such models, including hype, fear-mongering and anthropomorphism, inaccurate/false/psychologically harmful outputs (unintended), misinformation and spam (intended), of taking human work without attribution, over-reliance on partially-reliable models in the hands of a small number of individuals. The latter is particularly problematic I think since humans exhibit such strong herding effects already, without relying on a single source of truth which provides answers of unknown provenance. And I'm troubled at the trend of layoffs in the AI ethics field.
Wrt winter: AI winters have historically happened precisely because of hype - the public are given to expect too much and the reality doesn't match expectation. If we can all just calm down and say "there's some cool stuff we couldn't do before", I don't see that we have to have a winter. But I think it is quite possible with the current hype storm. I think the analogy with self-driving cars is apt. I would anticipate a similar problem -- yes we can keep pushing performance of these things, but enforcing constraints of truth and ethics will become increasingly challenging, and at some point the question will be asked "is it worth it?"
The answer in some circumstances will be yes, in others no. And we will likely make some more advances in the coming years which make this easier.
Curious where you got the PhD? I'm looking about going into the same field
I had to swap to a second sockpuppet account so my comment would actually show up, but I just want to remind everyone the real threat of AI is automated censorship. apparently we're not allowed to think against the narrative. cheers.
Sounds like you have a third rate PhD..
Don't forget when the car fails they blame the "driver" .
You said everything I’ve been telling people for the last year. Someone at work used Chat GPT to get medical advice. I had to tell them the truth about AI and that using in that way could be dangerous. Incredible.
I worked with an AI safety lab for a while and Adam's comment "It's just a parrot choosing words its heard before" and the ending about creating entirely new ideas is almost verbatim what the lab director says about generative AI and LLMs.
While it's true that LLM are just guessing at the next "word" that is most likely to be used; this process is actually a very important aspect of top-down processing. These large datasets are compiled into the model and how words and ideas are mapped inside these models is actually still a mystery to people working on them. This process of "mapping" things is a fundamental part of reasoning, which LLM's have already demonstrated they are somewhat capable of doing. I'm sure you are aware, but just wanted to clearly state, reasoning alone is not sentience.
Tbf, the director probably landed his role on connections and not competence. Atleast, it's safe to guess as much with how silly of a thing that is to say. When it generates a new idea, is it just parroting words its been trained on? It just seems to fundamentally not understand what LLMs are doing lmao.
@@Eric-vh4qg Yes that's always the way I understood it, that they're capable of certain kinds of 'reasoning' but not 'truly creating things'! The parrot thing is certainly a huge simplifcation. I was just a comms guy so the minutiae were a bit lost on me but they way they explained it was, "If you told something like DALL-E to mash three art styles together in a particular way, it could do a pretty good job. If you told it to invent an entirely new art style (like Picasso did with Cubism), it wouldn't be capable of understanding what you were asking." I was always hung up on the idea of all art being derivitive but I do think that explanation captures the nuance on how it's different.
@@Eric-vh4qg you're confusing complexity with sentience. Developers don't know every one of the millions of possibilities it examined or the billions of variables it pulled from its training data because the dataset is too large. Developers told it every step of the way what to do and how to do it. They don't know what it will decide because there are too many possibilities for a person consider. They could at every step of the process stop execution and examine the equation it's basing its next decision on and they could have it print out every one of the millions of equations it used to determine the 1 sentence answer you got but that information would be meaningless to almost everyone and would take a team of people the size of Google to interpret.
@@Kakerate2 this is so absurdly arrogant and stupid of a thing to say, it's laughable
Human levels of inteligence is hard to achive even for some humans.
Ha ha ha ha ha! 🤣 Yep anyone who has worked with "the public" feels that comment.
hahahahahaha
😂😂😂😂👍
I fear the fact that humanity is AI's role model.
We are coming closer to the day where we, as a collectivity, will finally admit that capitalism isn't the right system to run a society.
Adam informing us that AI could eliminate James Corden is the strongest argument I've ever heard for this new technology.
😂😂😂
😂😂🙌
First time Adam gave me hope instead of anxiety
For the first time, corporate bullshit and exploitative business practices damaging to society have been the lesser evil.
Adams still wrong
Adam fixes everything.
what in the world are you guys talking about. sure... stay the cave... because, fire scary.
It's pretty easy to strap a rifle to a robot dog
An absolutely VITAL video to cut through the noise we're all currently being bombarded with. Thank you and bravo, Adam!
All this stuff they call AI is basically just a computer equivalent to cruise control.
The confidence that man had that the car wouldn't hit him is astounding
That was effing beautiful
Condolences to his family
10 people died by AI driven cars but how many died in that same period from humans driving?
Check-mate meat-sack! You say machines are dangerous, yet you created this video USING machines.
Interesting.
/ @zedbikle8213 adjusts fedora /
@@z213x95 what an absurd statement. AI cars aren’t nearly as widespread as.. human drivers. it’s crazy that this is something that actually needs to be put into words for someone. maybe adam was wrong about how smart we are.
I love the idea that corporations are chasing artificial intelligence when they show zero interest in the human kind. I'm not kidding.
Genuinely. It's fucking enraging. "Ai" is, collectively, attempts at finding ways to do what humans can do and literally enjoy doing, but without the human element. It is the complete commodification of humanity and it's disgustingly transparent about it. It's no surprise that people who are "enthralled" by this anti-human ai shit are all privileged upper-middle and high class individuals who feel entitled to maximum efficiency and are emotionally disconnected from others.
Intelligence is a rare commodity for them, so being able to make an artificial source would be like finding the fountain of youth.
@@varethedemon The ONLY "intelligence" (or inventive, problem solving "ideas") a narcissistic and/or psychopathic CEO, etc. can harness (Gates, Musk and many others... immediately come to mind) is that which they can either buy, or preferably: steal, from "higher creative intelligence..." (Then they pay themselves 500x as much, and "reinvest" via "insider info. and trading," while using foreign slave labor, claiming no home country 'profits...' (paying net zero taxes) and hiding the proceeds, in offshore trusts! Then the foundations 'reinvest' in the cult's next generation, of "of, by and for" the cult's vs. the people's benefit... TAX FREE!)
I think a better title would be "A.I. is marketing B.S." just because (& as you said in the video) it is a real field of computer science that has been studied for decades. The recent boom is due to marketing, Calling the video "A.I is B.S." may cause someone scrolling past this video to assume that it machine learning programs are somehow literally fake.
*WHAT AMAZES ME MOST* is how quickly humans learned to stop AI 'art', 'music' and 'writing'
Corporations love this AI obsession too. My last job was constantly threatening to lay us all off because of AI and they have been threatening poor truckers for years now (and simultaneously wonder why there's a shortage). It's a way for these corporations to now threaten skilled workers, outsourcing 2.0. Anyways, the AI at my last job was such a failure, they had to split my old role so someone could QA the robot. Increased processing time significantly but they tanked millions of dollars into these failed systems so just forced us to make due.
Every time I think of AI/self-driving trucks, I wonder how the AI will put on chains when I-70 through Colorado's mountains have a chain law in effect.
@@johnchedsey1306 I'm sure a few well-placed campaign contributions would make that question irrelevant when the companies running the trucks get a carveout "for the benefit of interstate commerce."
I’m now more convinced than ever it’s coming. But I’m also more convinced than ever it’s not going to be as quick as some people believe.
This just goes to show that more people are needed in this field and that people should be learning about it so they don’t lose their jobs.
ChatGPT has already led to some fiction publishers closing their doors to open submissions because they're getting inundated with AI-generated short stories. Which also closes those doors to new writers, because if the publisher doesn't know them to invite them to submit their work, they can't. This is going to have a chilling effect on art and creativity.
When did Adam Conover start wearing an aluminum foil hat? As a developer, ChatGPT comes up with better ways to sort using methods that I have never even hear of. I had never even heard of a TreeMap before I asked ChatGPT, "What is the most efficient way to sort my Java map alphabetically?"
Have you ever tried to write a story with AI? It randomly forgets things that it wrote in the previous paragraph. If you just let it write without spending half your time manually re-writing its output, then the result would look like it was written by a toddler with an impressive vocabulary.
If that's actually what's happening here(it isn't), then those publishers are *idiots.*
@@JaredBrewerAerospace I just looked it up and apparently regular HashMaps are faster that Treemaps. ChatGPT screwed it up royally yet again.
@@CheshireCad We're talking open upload sites for writing like CZcams is for videos. People who don't care about the quality of their upload can just post whatever string of random paragraphs the bot gives them, and these sites don't have the insanely robust servers that Google does to host all that garbage data coming in. They have to close new submissions to avoid the site crashing. It's like trying to weather a DDOS attack that never ends.
@@JaredBrewerAerospace Just because you've not heard of something before doesn't mean it didn't exist. You're literally describing a search result, not an innovation. But, a result you had to get lucky spinning a gambling wheel to find (even if it may be a high odds in your favor gamble).
I work in tech. I kid you not, we are being asked to find (or create) problems that can be solved with AI for the sole purpose to use AI. It doesn't even necessary make sense to use AI in these cases, but the C-Suite doesn't care. They just want to tell investors and customers that there is AI in the products/service.
the most disruptive business right now is exposing the hype and bs from tech industry
I'm so happy that Adam has his own channel now and can just go hog wild and start screaming and swearing.
i actually dont think the swearing adds much to the otherwise good content, maybe im too used to his softer adam ruins everything personality lol
Best thing that ever happened honestly. His show was great back in the day but you can tell how much more real he can be now
@@wipeoutxl21 I think it adds proper emphasis on how fucked up things are that he covers.
AI took our JERBS!
me toooo
I'm a PhD researcher in natural language processing (the area behind text generative AIs), and one thing I think is missing here is that Bing's AI (I can't say the same for Google), was released while it was bad ON-PURPOSE so that society could start to create videos like this and imagine what life would be like if the AI was better and start to adapt while the AI is still bad.
Some other errors that I think may cloud people's thinking in this video:
1. AI can also create new things that have never been seen before (but Adam is right in that the new creations must be based off of data the AI has seen before.) I don't know enough psychology to know if humans can create truly novel work, and I'm not sure if that question is settled. (Note that Adam did not provide a source for the idea that humans don't combine old stuff in new ways in a similar way to generative AI.)
2. The Bing Chat engine is not just generative AI, it can cite its sources, and is more similar to OpenAIs WebGPT model. It still has issues, but it's not the same as ChatGPT, which is generative AI + intelligently ranking outputs.
3. Most of the scary "AIs will takeover everything" worry is done based on the idea called "X-risk" that if we do adopt this technology into critical parts of our infrastructure, and we program it wrong, it could destroy humanity. It's a big worry from the potential consequences perspective, but not from a "this is likely to happen" perspective. In short, it's a good thing to be worried about, but it shouldn't take all the hype.
However, those were pretty much the only issues I could find with this video. He's right about the AIs not being good. He's right about it running on hype. He's right about the ethics teams being fired. He's right about it copying humans' work who didn't give consent for their work to be copied.
If you have questions, feel free to ask in the comments! I'm happy to help clear some stuff up and explain how/why he's right.
I am curious what you think about the field's current progress towards AGI? One of the main concerns I have around all this AI hype is probably outside the reach of LLMs, but we have been making great progress in the field and I feel that reaching AGI or something close seems definitely possible within the next few decades at most. Once we have this, I fear that the effect it'll have on humans and their work will be tremendous, and that our societies, economic systems and governments will be too slow to react and protect people from the inevitable hardship that will follow.
Thank you for this writeup! I was scanning the comments in hope that someone points out those gaps in a succinct fashion!
Ya, I have a problem with the idea that A) 99% of what humans generate isnt basically a logical generation of past work, and B) that AI isnt intelligent because HOW DARE YOU COMPARE IT TO HUMAN INTELLIGENCE, like humans have some sort of monopoly on intelligence.
We are absolutely living on the shoulders of giants in our current society, and almost every aspect of our lives and even thought processes are generative output based on either our own experiences or things people in the past have discovered. Just like AI can combine a million ideas to make a unique thing, humans combine ideas they have learned into new things, and our social and economic systems filter out the less useful ideas. On a macro scale, society very much functions like a generative AI. Sure, AI can only output its model because it trained on thousands of other works and weighed the outputs to find what works, but Adam trained off of the work of others, like George Carlton and John Oliver, to find an output format that got the most views in order to pay the bills while letting him talk about the things he is passionate about.
@@miclowgunman1987 But does that excuse the displacement of thousands/millions of people to support a handful of people at the top that created these models?
I agree, I had a hard time taking him seriously because he kept getting stuck on the "bad fanfiction" thing, and how he kept saying AI makes a lot of mistakes. And I'm glad I was able to get over that, because he did make some good points otherwise. It's just hard to talk about this because so many people have a black and white view of AI. Most people either hate it or love it, and aren't really able to think about it rationally. This includes me, I really like AI so I'm having a hard time considering counterpoints. Adam has helped a bit with that. I still disagree with a lot of things he has said, but he poses some pretty good points about the more realistic harms of AI.
This is one of the 10% (or less) CZcamsrs who are talking the truth. Over 90% CZcamsrs including those who claimed to be AI gurus don't have human intelligence. They are manipulated by hype.
Our current AI is only a brand. It's solely based on ML/DL.
I usually refer to "AI" as Simulated Intelligence. It's algorithms designed to simulate the appearance of intelligence. These days, it feels like I'm among a very small minority who isn't starstruck by ChatGPT and Midjourney. They are impressive technical achievements, sure, but that's where it ends.
Every day I see a new AI product come out and every day I slam my head into my desk whenever someone I love uses AI to do a very human task like writing a recipe for food
I've been following the various open source models. They are pretty interesting!
I think what happened is the somewhat never super hyped about stuff kinda folk got excited about this interesting technological step, and the stars aligned and it was a technical step that normies (and corporate executives) could interact with directly, allowing it to snowball more than the legitimately revolutionary development of memristors you have probably never heard someone mention.
I agree with you, but do not forget Turing's Imitation Game.
AI chatbots are really useful when you need to produce human like speech. Intriguing if you push them and ask for unconventional things they weren't made for. And either bad or completely useless at anything else.
Ask ChatGPT to rewrite Moby Dick as if it were narrated by Boomhauer, works great
Play a game of chess, no
Then you are wrong and do not understand what AI truly is. No need to brag about your ignorance
My theory is that that AI control problem they keep worrying about, about an entity that makes inhumane decisions to the detriment of real humans and that is beyond our control and that is so different that we don't even recognize it, that is something that already happened. We call those entities corporations and they have been around for quite a while.
I LOVE the algorithm analogy for describing corporations, niceee, and I’ve heard few others use it.
Finally, someone points this out. Corporations are only allowed to survive if they are most able to produce profit, even if it involves directly harming or even killing humans (this is also known as 'innovation'). So corporations are constantly tested and replaced to produce progressively worse outcomes, until they aren't selected solely for profit producing ability.
A misaligned AI is even worse because it can make inhumane decisions a thousand times faster than a corporation can. Corporations have been successfully trust-busted in the past, but that only worked because corporations and governments both operate on a human time scale.
Did you see the interview by Liv boeree? She speaks with Daniel schmachtenberger and he essentially points to our collective society as intelligence with a somewhat broken reward function. Does a great job of defining intelligence vs wisdom and talks about how to approach the problem
Like most powerful people the ultimate fear of Tech Bros are terrified of the possibility that someone or something might subject them to the exact same shit they subject others too.
It’s actually a common theme in Sci-Fi. The entire genre of Alien Invasion Sci-Fi is especially guilty of this. War of the Worlds can basically be described as “What if someone does to England what England’s doing to the rest of the World.” For us Americans Independence Day (the story of a alien power invading our country, bombing our cities, and slaughtering our fellow citizens so they might harvest our natural resources) is a story that exists in fiction. For many other countries that is just a fanciful description of the history of their relationship with the United States and Europe.
Well, you should read a book called "steal like an artist". The truth is: nothing gets "created" the way you THINK it does. Humans are imitation-machines too, even if you don't realize it.
Saying that AI isn't intelligent because it uses other people's work as a foundation is the same as saying a guitarrist is bad 'cause he's influenced by (and plays a lot like) Hendrix. I've lost count of how many times I've heard a new song that sounds way to similar to some old music (You know you've been there too).
Ai is young. It isn't currently capable of some crazy good creations. Expecting it to be flawless at this point is like expecting a 2 year old to write a best-seller. It's just dumb.
But hey, I've never expected artists to be intelligent anyway... At least I have my expectations right. And you artists might fight as much as you can to avoid AI, I think it's so entertaining to watch haha.
Generative AI? More like Degenerative AI. Am I right?
The fact that Adam knows running over James Corden in a mouse costume cause the AI thought he's a mouse would be funny as fuck is the reason he's a great comedian.
That just floored me
The only reasonable reaction when driving in front of any James Corden Crosswalk Show 😂😂😂
@@odd-ysseusdoesstuff6347 ABSO-FUCKING-LUTELY
I've been fired several times as a Principal/Staff Software Engineer over the past 4 years because the companies didn't like that I pointed out issues with their infrastructure, design, or process. That was literally my fucking job. I did so constructively and professionally, along with recommendations for remediation. These tech companies are run by such ego-driven assholes, I can't take it anymore. What happened to getting paid to do your job?
The assholes wanted a pat on the back for being so cool and smart, not having you crush their ego by.... doing your job. How dare you try and actually improve things!? I hate the tech industry. I love technology, I'm currently getting a degree in tech, but I don't think I can do this as a career.
@@stellar1526 You can but it's tough. You gotta find the right place. Look for true nerds and consumate professionals, not tech bros. Stay away from places that have VC funding and "high growth."
Your commentary on the state of AI was spot on. Your last 2 minutes (starting at 23:03) of commentary on humanity was beautiful. Keep climbing to the top of bullshit mountain, you're great.
Yes, the hopeful tone after all the blustering is reassuring... that is, until you look into how "generation alpha" can't read, regulate emotions, be creative and lack so much basic understanding that teachers are quitting in droves. The qualities he claims will always save us are disappearing from this latest crop of "iPad kids".
Meh, this didnt age well
This was the most uneducated commentary I ever heard. The dude has no idea how AI works and the way he explained LLMs is idiotic. Every downside of AI you can also attribute to humans except for worse and occurs much more frequently. AI can out perform humans on just about every cognitive task now. AI has solved protein folding, discovered new materials on earth, and can answer way more questions correctly than any human being who has ever lived. The guy beating AI in go is such a stupid example considering AI has beat all the best players and this dude hasnt. It will also learn the players strategy and be able to beat him. He said it was stupid because AI was never trained on it so the fact it has to see it first means its stupid. Guess what humans have to be trained and be exposed to things in order for humans to accomplish something. It says AI just parrots people and creates nothing because its based off of other peoples work. Well guess what dumbass, anything people creates is just an amalgamation of other peoples ideas and data they seen. There really is no difference except AI is has more memory, much faster, much more accurate and precise, learns way quicker, has no emotion so more objective, and has the ability to never rest. The guy was so stupid in saying how bias it wae but humans are a trillion times more bias with information. The fact people think this was a good commentary shows you AI is winning the war on intelligence.
Thanks, I needed that. A guy from India once told me, "It is clearly written", even though it was totally wrong. So AI today is The Great Pretender.
The comments about AI ethicists were so good. I have worked on some language analysis projects and it has some amazing potential. On the other side bigoted algorithms are already a problem. Algorithms will reflect the bias in their data sets and the real world is full of bias
if i am invisible in the Chinese surveillance system because I am white, GOOD.
As and IT geek. Industry rebrands existing functionality on the regular. I really enjoyed this assessment. "It's just all marketing". Yes, it's going to be used to screw us over. As always.
Yup. I remember when Expert Systems was all the rage and would render work obsolete.
From 1984 by George Orwell:
“There was a whole chain of separate departments dealing with proletarian literature, music, drama, and entertainment generally. Here were produced rubbishy newspapers containing almost nothing except sport, crime and astrology, sensational five-cent novelettes, films oozing with sex, and sentimental songs which were composed entirely by mechanical means on a special kind of kaleidoscope known as a versificator.”
I for one, appreciate the "AI boom". Just like NFT, it's a handy self applied label that's screams, "avoid me!".
I'm a software engineer, and you hit the nail on the head when you said "Mostly gonna revolutionize spammers." Had me rolling lmao
On the other hand, baysean learning has been used for over a decade now to combat spam, it should hardly be a surprise when it is used to create spam.
Ha! Imagen a anti spam bit being fed spam from a spam bot.
@@CaptRespect and that's what's called a GAN
So 99 percent of spammers will end without a job?! Hmmmmmm....the cons.....but also the pros!
@@tomikexboii5403 Worse, those who do spam will do it more efficiently. There's no demand for spam, so they can make as much as they want. The easier it is to make spam the more we'll see.
We all laughed at that Google engineer who quit because he thought the AI was his waifu, but then that kind of AI was released to the public and we all automatically personalized it too
Advanced sneako moment
It's funny how before that happened AI was still a pipe dream of the future. Then the Google whistle blower released their statement, and the corporations were all "oh yeah, thats been around for a while. Then virtually over night AI is everywhere. ..
@@jaegrant6441
Thats certainly one way to rapidly conjure a conspiracy theory out of nowhere... If thats really what you're doing here, I'd argue its much more likely the reverse, the guy was an AI insider, so he was making his statement because of the incoming of publically available "AI" models. I say that because it really sounds like you are suggesting a bunch of multi-billion dollar corporations saw some dudes article on medium and suddenly released their AI because of his article. That'd be a pretty odd thing to suggest, hopefully I'm confused.
@@jaegrant6441 it's is gpt 4 because it is version 4. Ai has been around for awhile. I used gpt 2, 3, 3.5, 4 and it's just hit a point where it's impressive. Gpt 2 was ok 4 is way better. That's why the hype.
If you go and read that engineer's blog you wouldn't be quite as flippant about his concerns, honestly. His name is Blake Lemoine, just search it and you'll find it right away. He spent a lot of time interacting with LaMDA (Google's AI language model) and while I don't believe he conclusively demonstrates that LaMDA qualifies as sentient, his argument is far more detailed and nuanced than what the media portrayed it to be.
LaMDA is much more advanced than ChatGPT, and the uncensored / unfiltered version Lemoine had access to is capable of producing far more cogent and, frankly, spooky communications. There *are* ethical concerns at the core of his claim, and we *will* have to deal with those concerns some day.
Current A.I. is the world’s most complex Magic 8 Ball, and people think it has real magic powers.
nothing magical about auto generated data,that's the internet equivalent of fast food. instant,and bad for you.
Man made the buildings that reach for the sky
And man made the motorcar and learned how to drive
But he didn't make the flowers and he didn't make the trees
And he didn't make you and he didn't make me
And he's got no right to turn us into machines
No, he's got no right at all
'Cause we are all God's children
And they got no right to change us
Oh, we gotta go back the way the good Lord made us all
Don't want this world to change me
I want to go back the way the good Lord made me
Same lungs that he gave me to breathe with
Same eyes he gave me to see with
Oh, the rich man, poor man, the saint and the sinner
The wise man, the simpleton, the loser and the winner
We are all the same to Him
Stripped of our clothes and all the things we own
Oh, the day that we are born
We are all God's children
And they got no right to change us
Oh, we gotta go back the way the good Lord made
Oh, the good Lord made us all
And we are all His children
And they got no right to change us
Oh, we gotta go back the way the good Lord made us all
Yeah, we gotta go back the way the good Lord made us all
Fun fact
Cnn published article from medical researchers in usa universities
Chatgpt not only lied on is diagnosis but have fictionalised articles it created
So chatgpt official spkkesperson admitted chatgpt and AI whatever is just incompetent
Thanks for this, man. I’m so sick of AI hype,specially coming from so called experts...
not saying he’s wrong, but you’re just gonna take a single source and run with it? even if “ai” doesn’t develop any further than what we have now it’ll still have pretty drastic effects on society
Yeah, experts that all their life used ctr+c/ctrl+v and "trust me bro" source to make a point. LOL ! Yeah, real "experts"
Honestly, the video is really dumb and comes from the place of bold ignorance. The real risk of AI is not some tech companies hyping up their stock, if this truly was the actual danger of AI then it would have been amazing, but AI is a VERY powerful and DANGEROUS thing when left unchecked and its not just hype. It IS NOT hype. AI will affect humanity in more ways than one and the stock portfolio of some tech bros is the last thing you should be worried. The entire world will be transformed, one way or another.
AI creates existential risk for all of humanity if not the entire planet, so far we have not yet created a thing that have been smarter than us, and if we do then we'll all be like a bunch of kids trying to outsmart an adult. An ant colony standing in the way of the tractor. A fly in the face of an airplane. And if you dont understand how creating a thing that is faster, smarter, more robust than your entire race combined is different from creating some new crypto or metaverse project then I suggest you listen to the AI experts. Neither crypto, nor metaverse, nor tech companies hyping their stock carry existential risk for humanity.
I had to drive nearly an hour to and from work on busy roads for a few years awhile back and had a lot of time to ponder just how my human brain, eyes and ears took in the vast amount of information you can get by observing not only the traffic around you but also the body language of the other drivers. I also found that often the safest times were when you were driving moderately aggressive, not allowing the bullies of the road to push their way in an out of lanes. On a crowded freeway you were far safer going the speed the other cars around you even if that was over the posted speed limit. Your brain made instant decisions of lane positioning to ensure that drivers in other lanes could see you and that you had a place to go if someone did something unexpected. You would read a sign that says your lane is ending a mile ahead and know that if you get over too early, an entire string of very aggressive drivers will go around you in that ending lane only to have to push their way in ahead of you. You would be punished for being responsible so you would find some compromise. AI or self driving cars could make none of those decisions. Your brain can add in so many gray factors and experiences instantly that a computer program could never emulate. Cameras and sensors can only resolve pre-programmed inputs. Let it try to understand a part falling off a delivery truck in front of you bouncing wildly or a driver gazing at their cellphone weaving over a lane.
That said, self-driving cars can work....if every vehicle on the road is self-driving and programmed to respond to one another. Society will NEVER pay for the infrastructure required for that dynamic.
Honestly, you'd get automated trains before that
The other part which he leaves out is that it's odd they went for blue collar and physical things first before white collar stuff now. If a computer dominated chess first you go white collar next not drive a car. Self preservation. The more symbolic it is the easier for a computer even if driving can seem robotic for us after a while. The obsession with self driving cars probably shaved off some years from the imitation game of white collar AI
@@PapaphobiaPictures amd we are never getting that, at least in the ole USA
All vehicles would have to be self driving, belong to the same company, and there can be no weather, human, animal, or any other natural or unnatural concept or entity that might even remotely interfere with how it's programmed.
" You would be punished for being responsible "
so what? only human ego suffers from that and feels belittled and goes into road rage - bot doesn't care. it has no ego. it goes the safest scared 80 year old grandma style, if its the safest way, period.
Have a friend that is a professor of AI and teaches it in university, was asked to come on a news show to discuss it. He refused because the people from whom the money he gets paid originate wouldn't be happy if he went on and told the truth.
You know what they say about AI: garbage in, garbage out.
The more reason we need to hold the companies accountable.
We can’t unless we collectively as an American society march to all the tech companies, Silicon Valley, and DC and protest everyday for years until we threaten congress’ jobs. There’s a growing underlying resentment in America right now that’s slowly fizzling it’s way to the top. January 6th is an important example of that 😬
@HypaSnypa12 true, even if that was politically mislead.
Accountable for what?
@Alex in short? Ruining everything while trying to see what they can get away with.
The question is how? We need to reinvent accountability
Just think about how frustrating it is when you contact customer service for anything and it's a bot (which is way too common these days). It hardly ever solves anything, they never understand what you actually want and they always give you the same 5 solutions, none of which work for what you need. Imagine the whole world working like that.
To be fair that’s a gross comparison.
That’s like saying the first ford only went 12 mph so imagine a world where cars replace horses.
But back on topic in
I mean... the bots you're talking about are stupid scripts that have no idea what you're asking. If you had ChatGPT on the phone it would actually understand what you want and help you. So you're really arguing against yourself here, companies already have garbage customer support and these systems could very well improve them tremendously.
Whenever a bot asks me what I want on the phone I say something like "I want to talk to a human" and it always redirects me without a fail. Just a hint.
@@spwn6738 "I don’t see how customer service humans compete with an ai chat bot."
Maybe if your problem is so run-of-the-mill, that the surface data is enough to deal with it. Anything more complex than that, and you will speak to a human really fast. Also, you have to consider that callers themselves can communicate quite inefficiantly. A human operator can actually recognize that and wrap their head around it. Can a freaking bot do that, too?
YES! It’s mindboggling to me that companies invest millions in some shiny new objects and can’t even have a person to talk to you over a phone/chat in less than 30 minutes!
Self driving Ai conceptually is totally possible- Code Bullet makes Evolutionary-Ai that can drive in virtual environments. Having a more advanced neural network with training models would probably perform even better.
Taking that into the real world is difficult. And Tesla likes to cut corners.
Also, do we really want to get rid of millions of jobs? *thats rhetorical
This is amazing. Finally someone makes the salient point that has always needed to be said: "Billionaires with zero integrity and zero incentive to care about human lives are lying to us about the functionality of AI vehicles, drones, computers, and toasters." THANK YOU!
One of my favorite quotes about artificial intelligence came from an issue of Scientific American, where an interviewer asked an AI specialist when we we would see a true machine intelligence emerged, and the researcher responded "Well, what do you mean by 'intelligent'?"
most ppl don't even understand the very word "intelligence"
He asked that because the term "inelligence" is used incorrectly in 99% of colloquial conversations.
People who don't study it tend to drastically overestimate what an intelligent system should be capable of
@@fritt_wastaken there isn't one correct definition of intelligence, which is why it's important to nail down what someone means by it before you can have a discussion
That's a perfectly reasonable question though... it's a complex question and by some definitions we have had AI for over a decade now.
Yes, before a productive definition can start, the parties involved need to agree on what definitions to use, there are lots of words that mean different things to different people for some reason, intelligence, and consciousness are common examples of words with many or vague meanings
Filtering out graphic content happens in the US too. During the 2020 election I worked for a contractor that flagged misinformation and troubling/violent content for Facebook and I saw some deeply violent and disturbing things there that I can never unsee. And I live in Cleveland.
Wow. Thank you for that. I’ll try to keep this new perspective in mind and not take content filters - and the people behind them - for granted.
Damn bro whatchu got against Cleveland
@@k.taylor3526 The people who bad mouth, social media companies moderating content have no idea what sort of chaotic hell it would be without it. They complain about their children being exposed to diversity centric content. Just wait until they see an unfiltered internet. They'll want to ban the internet as well
I'm sure they paid you more than $2/hour though
@@semmelwhat are you talking about? The guy said he worked filtering graphic content in Cleveland, but Adam is talking about exploiting work in Kenya paying workers less than $2/hour. Both are bad, but being exposed to filthy shit AND being exploited and underpaid is even worse.
I love AI for work it really helps, it listens to meetings and writes it all down which is super helpful and saves me time.
I do think a lot of it is very overhype though.
See ? That's a great use of automation ! It takes a tedious task away from us to make our lives easier.
The "fear that AI is going to take over" isn't just fear, I know quite a few people who lost their jobs as graphic designers or artists this year to be replaced with "AI art"
My experience using chatgpt to help coding has been like working with a 90 year old engineer who is really knowledgable, but has dimentia. THey have the answers, but its all kind of jumbled up and they forgot what they were talking about a few paragraphs back. Still, its been very useful and helpful.
well this is because it exactly how it is( just that you should have used word Alzheimer's not dementia)
it only has 2000 word context window. everything past 2000 words is forgotten and probably big chunk of that window is taken by character description
they are trying to make that window bigger but that's very difficult task
@@deltaxcd What? It’s not a difficult task, GPT-4 already has way bigger memory, and there’s plugins for even bigger and more advanced memory.
But it takes resources that smaller companies may not have. So if they’re telling you "boohoo it’s difficult" that’s the BS, they’re just not investing enough in their servers.
@@Hodoss well the problem is that to double memory you need to increase your processing and memory 4 times and considering how extreme it is already, it is troublesome to do. Doubling memory will not be enough anyway. and if you look at the GPU operating costs from companies that sell you computing services it is getting quite expensive.
gpt4 has 2x of chatgpt memory and I think gpt5 will double it again but hardware requirements are getting ridiculous and It is questionable benefit. still not enough to process entire book of harry potter and give you summary or read any other reasonably long document.
Also model can't learn it needs to be trained with new data so it can't work as search engine in principle
@@deltaxcd not entirely accurate. It's "memory" isn't based on words but rather tokens. tokens can be words, but they can also be word parts or even while phrases. It isn't a 1 to 1.
@@deltaxcd also, there are ways around the inability to summarize an entire book all at once. For instance, summarize the chapters, then give a summary of those summaries. There is a chance you miss a bit of context of things that carry over between chapters, but it can still get you quite far.
Not gonna lie, if Adam was a writer on the Simpsons, it would be a major boost in quality.
Probably not, the problem with The Simpsons is that it's been going on so long it's both dated in format and utterly mined out for content. The quality of the writers isn't really the problem at this point, the IP is just so uninspiring that it makes good writers mediocre and bad writers worse.
@@theuserofthissite All the good writers left long ago.
@omgshrimpz I'm saying you wouldn't be able to tell if there WERE good writers. They don't have the opportunity to show what their writing skills are if they're writing The Simpsons. If you put a world class chef as a Subway sandwich artist, they'd only be able to make you a Subway sandwich.
Many have said this but it does not sound right until Adam says it. Love this guy!
I love that this video was a year ago, and today we have Google AI overview telling us to mix Elmer's glue into our pizza sauce to help cheese stick, and to eat one small rock a day
I’m a translator. My job is to figure out the exact thought a person was trying to communicate to another person based on our inherently imprecise mechanism of language, which necessarily carries entire thought patterns and belief systems, then translate that into another bundle of thought patterns and belief systems, all while retaining the original communicating thought.
So when this lady said I must be so worried about AI, I was like, Have you ever seen AI translate anything more than “where’s the bathroom” with any accuracy? ‘Cause it don’t. It’s terrible at anything over six words and its grammar is like Schwarzenegger.
Did you mean.. Swachateiger?
you are coping. AI definitly going to take your job
Our company has auto translation on ecommerce pages and then translators optimize the most visited ones or ones that show signs of problems. Sure doesnt automate away the entire job. But it does # of translators needed from 10 to 3 so in the end on the overall market less demand for translators. If i was in that market i would get really good with using AI to translate the basic stuff and then optimize the important stuff myself.
Donde esta el bano. Google once translated that as where is el bano. Half the meaning gone is a puff of smoke
I frequently use translate apps in the classroom because I have students who are very recent immigrants who are brand new to speaking English, so I really *really* feel the lack from these apps.
Goods news: they are smart kids and learning fast. Bad News: I've probably translated some instructions into seriously useless nonsense instructions.
I'm a truck driver and I've been told the last 4 years that my job will be obsolete and I better start learning to program. While I certainly want to evolve to more planet friendly shipping methods I never was convinced my job would be taken over any time soon. This video basically confirms my skepticism was right the whole time, so thanks Adam, love your content
I feel so stupid for falling for that "ai will take over transportation" scheme. I believe some day, but not with the current approach. Current "AI" is fundamentally poorly programmed - it mashes input and code into one and it becomes impossible to debug, and understand. There is also a tendency for AI models to get worse once they approach a certain size.
it's funny that programmer jobs might go before driver jobs
You and Adam are wrong. Once self driving breaks through fully... cya later Jack.
You are a fool if you think otherwise.
@@quatricise
You were convincingly lied to about self-driving AI, with fake data, fake video proof, and everything.
However, generative AI is being developed at a record pace. It's hard to tell how close we are to reaching the end of the stick.
@@fearfx1 they have been saying they will break through for the last 7-8 years. I just don't think it's going to happen the way you expect.
And I'm kind of ok with self driving premise, mostly because even if there were 10 deaths in the first 4 months, that's less than the number of drunk driving fatalities in that same 4 months.
Not to mention all the other dui possibilities.
People make bad decisions too.
I think a combination of driver and computer could be safer than either one left alone.
The mistake is trying to train a computer to think like a person. We already have things that think like people. They're called people. We create tools for things we are bad at doing. The fact that AI researchers are trying to make AI do the things that people are GOOD at doing is just an indication that they are hoping to eliminate needing people. Good AI helps one person do the work that would otherwise take 5 people. Bad AI gets companies to fire half their workforce to produce a worse product for much less money.
Guys, I have a new startup idea!
AI pizza
Now give me seed money
Artists have been screaming about the dangers of 'AI' for months, so I'm glad it's becoming more obvious for everyone. It's suspicious that tech companies are pushing for these 'AI' so damn hard when every other day we hear about thousands of layoffs and the millions of dollars they're losing. They're trying to hype it as much as they can because they're haemorrhaging money.
Great video, thank you!
They're also haemorraging money 'cause the people getting fired can't afford their bullshit anymore. AI is nothing but a race to the bottom.
AI/Machine Learning has been here for a long time and been used in everything, is just now that it can generate new information instead of just analysing it
why cant you think outside the capitalism box....................... why are you so brain broken that you only think of the profits? why none of the American commentators think how much more efficient and transparent the AI can make public services?
it could literally fix the government services and make people trust it again.
@@AnnasVirtual I'll grant you that though I'm not sure I'd call those 'AI' either. Also, 'it can generate new information' is debatable. It can vomit cocktails of things people have done already at best, at least where the image generators are concerned, and they're cocktails of stolen content.
@@AnnasVirtual more like Mish mash it the AI can't make new info that's simply bullshit it runs available data that's done by humans
The cruel duality of AI is that anything new gets treated like Ultron and everything we're used to gets called "not AI", so the very real problems with current, real "AI" get overlooked and unaddressed
I was gonna say...haven’t we had AI for YEARS in the form of the automated assistant for company phone systems?
That's because we're no where near a true A.I. it's just a hot button word. All their programs do is run very sophisticated algorithms. Same input in, same output out.
@@DrakenAshen so self driving cars will never exist? ever?
@@celinelia8127not until computers can actually think instead of run probability equations
We may get autonomous cars capable of having in average less accidents than humans. But two things are important to remember: we as a society, and our way to see things... can't accept accidents by malfunction of automated actions of a device. They are infinitely more indignant for us than a human mistake. We will sue the maker (not saying it's wrong or not... just stating the fact). The other thing is, we humans can't agree on many things, on what os wrong or right as an absolute. We evaluate things under multiple criteria, but different people will give weights to those criteria in different ways. We can't therefore automate a decision making machine properly, except in very structured and controlled situations
I am not afraid of AI , I am afraid of natural stupidity.
humans are very good at GETTING DISTRACTED BY novel stimuli they have never seen before....