The Real Danger Of ChatGPT
Vložit
- čas přidán 29. 12. 2022
- Go to www.squarespace.com/nerdwriter for 10% off your first purchase.
GET MY BOOK HERE: amzn.to/3EPDQKt
Support Nerdwriter videos: / nerdwriter Subscribe: bit.ly/SubNerdwriter
Watch the most popular Nerdwriter episodes: • How Donald Trump Answe...
Facebook: / the-nerdwriter-3141415...
Twitter: / theenerdwriter
Patreon: / nerdwriter
MUSIC (via Epidemic Sound)
Stationary Sign, "Paint By Imagination"
Arthur Benson, "Sneaking Into The Kitchen"
Arthur Benson, "Mixed Emotions"
Sture Zetterberg "Merry Christmas Baby"
Watch More Nerdwriter:
Latest Uploads: • Video
Understanding Art: • What The Truman Show T...
Essays About Art: • What The Truman Show T...
Essays About Social Science: • How To Correct Donald ...
Popular Videos: • How Donald Trump Answe...
The Nerdwriter is a series of video essays about art, culture, politics, philosophy and more. - Zábava
"Not to write is to live according to the language of others."
That's a powerful statement.
Not to draw is to live according to the artstyle of others
@@stevealex6163 Writing is drawing your imagination
@@JishinimaTidehoshi isn't drawing is just drawing our imagination? (if that makes sense)
@@stevealex6163 That's not an accurate analogy. Language actually plays a major role in your intellectual develop.
nope. does choosing what books you read dictate what you know? that would imply reading makes you smarter. nope.
I really expected the twist to be that the whole script was written by ChatGPT 😁
I’ve already seen a few CZcams essays that played that trick haha
I think he hinted that at least certain parts of it were.
I was about to say the same thing.
That already got old about a year ago
Axshole stop giving me spoilers
"Not to write is to live according to the language of others" is such a powerful quote! As a creative, I've felt this abstractly, but the way you phrase it is so concise and perfect.
It is flawed though, because there is no divine inspiration. Every creation is based on the works of others, just like an AI.
@@TrepidFaceHugger If every creation is based on works that have come before, how was the first idea, the one that preceeded every other idea, created?
@@TrepidFaceHugger The difference between a human being and an algorithm is that we have ingenuity and creativity. We can look at existing ideas, and form connections between them nobody else has seen before. An a.i. is physically incapable of this. It can only mimic what has already been done before to create an illusion of originality, but it is incapable of genuinely original thought.
@@UnchainedEruption An ai isn't "physically incapable of this" that's just bollocks. There's already millions of examples of AI creating original art and it will only get better. Who are you to decide what's genuinely original or not?
@@gremlinchemist3621 That's like asking "how did the universe originate?". It's not something I know the answer to but ultimately not relevant to my objection.
As an IT professional (and former website builder) it's pretty ironic to see this video end with a commercial for Squarespace. 10 years ago we could have said that sites like Squarespace that only provide cookie-cutter templates that are far from unique don't let people show their identity through a truly hand made website.
The reality is, as with any new technology, that the new technology does not replace the old one. There are probably more professional website builders today than there were 10 years ago, and services like Squarespace, Themeforest, etc. allow people that would otherwise never have a permanent online presence to create a simple, basic useful page.
I expect the same for ChatGPT. I doubt that it will replace professional writers any time soon, but it can likely provide fodder for clickbait sites and other low quality content. As for language teachers all over the world; it's now their turn to get creative in finding ways for students to "show their work"
Based
I do wonder if ChatGPT produces the same essay each time you use the same prompt. I suppose I could test it
Edit: it did when asked to write an essay on the same thing was put in back to back. I'm going to check again in a couple of days
@@chestersnap It reproduces the same content with slight variations. Try for instance asking for 10 dad jokes. Once you have those ask for 10 new dad jokes. Then ask it to not recycle content across queries... It fails no matter how hard you try
love it, slipperydippery
best take
In university, one teacher's final exam revolved around a research paper... that you had to write by hand in the exam hall. He required no formal sourcing, but you had to base your arguments on what you'd learned in class. This is an excellent way to get around ChatGPT's capabilities. Require on-the-spot formulation of thoughts and then communication of such through written language. Making the process happen in front of the teacher means there can be no behind-the-scenes cheating.
This.
this is how all my essays were in highschool and its incredibly strange to me that other schools just let you do it at home. no shit every kid is gonna either copy, plagiarise or just get someone else to do it.
@@_holy__ghost I don't necessarily think it leads to more plagiarism or otherwise cheating. But just as bad, I think it leads to the same type of studying as when you study intensely 24 hours before a test. You don't actually memorize the things you write about - you only keep them in your short-term memory for as long as it takes you to finish the essay. You don't actually have to ingrain stuff into your memory and draw it out on the spot - you can continuously look up your notes and sources repeatedly whenever you need to, which means that this information never really lodges itself in your long-term memory. And then it's gone.
And the problem is that you can't really build a well thought-out, structural, complex understanding of the world or its thoughts if you don't actually memorize enough to compare and contrast and generally think on a greater scale. It leads to simplified worldviews specialized around one's personal interests, at least that's what it looks like to me. More fragmented individuals, because students don't get the full picture because they never learn to memorize properly.
Another idea is to interview the kid about the essay. It'll take less than two minutes to discern if the essay was ai generated or plagiarized.
@@_holy__ghost i disagree with the idea that everyone just cheats if they get to write essays at home. i personally love getting to take essays home because it gives me enough time to write a GOOD essay and actually edit. any good essay requires skill but it also requires time. judging students only on essays done under test conditions with a time limit is a surefire way of never getting a proper idea of their skill level. there are definitely people who will cheat whenever they're given the opportunity but loads of people i know don't cheat on these essays and try to do their best work instead. it's more important to have good teachers who are able to convey the importance of learning these skills for yourself than to start grading essays that were written in under an hour.
I'm a High School English teacher and I get asked that question all the time. I tell them that it's not the actual essay that they'll use in the future, but the many skills they used to create it.
Well said
But that is untrue. They will not *need* these skills in the future. They will not *need* any skills.
@@danielguy3581 yeah analysis of the topic and research materials, consolidation of knowledge, clearly articulating your thoughts, using evidence, being coherent, fact checking your claims are all things that no one needs to do ever, right? not like being clear and concise are things that people ever need to do, right? what kind of world are you living in? do you never have to communicate with people? do you never read a news article and think hey that doesn't sound right? these skills are literally the backbone of laying out any type of explanation or argument for literally anything, ever.
@@ishathakor I'm afraid you misapprehend. You ask 'what kind of world are you living in?', but as noted, I do not refer to the world in its current state, but the one we are rapidly hurtling towards.
@@danielguy3581 You gotta write more here to be saying anything meaningful.
I see a lot of people saying that AI generated text won't change the world, but what I'm worried about is education. I think now more than ever we see the importance of being able to think for yourself, generate your own opinions, and recognize when you are wrong or missing something. And with social media and content creation moving more and more towards endless consumption of their product I fear children will leave learning by the wayside, out source homework, and get lost in the terrible echo chambers of the virtual world. Hell, this is happening to adults as well. We need to make sure we preserve critical thinking.
Critical thought is helped by people with shared values for critical thought. When a culture says it's time to not talk about things, that's when a culture becomes dumb. And, frankly, that's every culture.
The truth is, I believe everyone is capable of critical thought. It's more about teaching people when to use critical thought. Schools need to be teaching people hardskills as soft skills come naturally as people grow through new environments. In fact, I'd go so far as to say you'd have to purposely try to not have soft skills to be missing them.
The fact students are cheating at all should immediately tell you they have the skill to critically think. They don't care because the schools don't give them a reason to care - we don't expose them to the world or philosophy, so they never learn why they should care.
Who is we in that. The world isn't a homogeneous group of people with common interests. It's a vast vast chaos of billions and billions of differences that expand and contract into each other. Where is the we in that?
Finally someone who 'GETS IT'
I will use ChatGPT to reply on your comment hahaha
I agree with you that it is important for people to be able to think for themselves and recognize when they are wrong or missing something. This is especially true in today's world, where there is an abundance of information available online and it can be easy to get caught in echo chambers. It is important for people to be able to critically evaluate information and come to their own conclusions.
At the same time, I do think that AI and machine learning can have a positive impact on education. For example, AI can be used to personalize learning and provide customized recommendations for students based on their unique strengths and weaknesses. It can also be used to help teachers grade assignments and provide feedback more efficiently.
Overall, it is important to use AI and other technologies in a way that enhances, rather than replaces, critical thinking and independent learning.
Really appreciate you mentioning the value of writing that is often not understood. I can't tell you how many high functioning, well paid, skilled professionals that I have met who didn't really have a voice of their own when given opportunity to express it. Most people don't take the time to explore the space of language on their own. Most don't understand that writing for yourself--as an exploratory tool--is possibly the most immediate ways to unlock the treasured trait of being a person who thinks for themselves. As you said, it is really the only way TO think for yourself.
I think conversation can act as a similar canvas, though I don't know if conversation is as readily available, nor as open-ended. In conversation I have certainly explored the world through language, and come to important realizations and defining philosophies and ideas. But only in the presence of people who are calmly and effortfully engaged in the same kind of conversation. Unfortunately, not many folks value this sort of activity; even less could put to words what the activity is at all... an arcane behavior in that sense. I think technology has made it much easier to all together avoid individualistic thought and exploration, which is not all bad, but honestly, I do see there being some interesting partitions being generated by the status inherently granted to those who CAN think for themselves and practice it regularly, and those who can't for any given reason.
Thinking for yourself is like a super power. We all look up to those who do it as leaders. That is why anyone watching this video IS watching this video. It's built into your biology to respect such activity. Don't take the idea lightly y'all.
Wow, what a superb, inspiring comment...you just made my New Year, thanks!
@@brindlekintales yeah this person writes
@@brindlekintales :) happy new year!
@@MrChaluliss 🐣
This might be one of my favorite comments I've read on CZcams, kudos to you. Very inspiring and ellucidating, writerly in the best sense of the word.
Do you have any published works to your name?
I outsourced my life to AI. Now I have the time to float weightlessly submerged in an oxygenated gelatinous goo
finally
Some people find that thought really attractive. I could go with it probably...
my brain wants me to inplant a ai be chip into my brain and let the ai decide what i want to do.
@@sownheard Living life optimally as per emerging data that is refreshed regularly.
The Matrix, but less evil, though potentially as malicious.
"Are we going to use this in the real world?" No, that's not what you're learning it for haha
Do athletes lift weights during a game? Of course not - so why do they even do it?
It's training that improves your muscles. The brain is no different!
That’s a pretty good comparison.
that's a pretty straightforward answer. I like it
I agree with the sentiment, a lot of modern education was devised as a way to make a well-rounded citizen and exercise a variety of different brain functions, but it's slow to adapt and doesn't teach enough practical skills along with the general knowledge. it's like teaching pro athletes table manners as a replacement for doing drills and strength training
@@xXx_Regulus_xXx I agree with you completely
Wow
As a teacher and fellow writer, this almost moved me to tears. It's a shame even less people in the future will experience the benefits of finding who they are and what they believe through writing.
Writing can be a powerful tool for self-discovery and expression, and it's important for people to have the opportunity to explore their thoughts and ideas through this medium. While technology like AI language models can assist with the writing process, it's important to remember that the heart and soul of writing comes from the writer's unique perspective and voice. Hopefully, even with the evolution of technology, people will continue to appreciate the value of writing as a means of personal growth and creative expression
Ill still hire writers and teachers. We have to support our brothers and sisters instead of killing creativity. A creative person will always have the most haters whose end goal is to dumb him or her down or make them the same/robotic as the rest.
I’m not a professional or trained writer, yet I love to write. I find myself through putting my thoughts to paper (or computer)
People don’t think anymore.
I understand your point it’s very valid , but before chat gpt lots off us had already used Siri or Google assistant etc so if I want to know something Siri makes life x100 easier
To me, this has the classic feel of a nerdwriter episode. It's informative, thoughtful, and slightly contrarian/unorthodox/unconventional. It's also pithy and engaging.
Great job, my dude. Keep beltin out those winners.
did you use chatGPT to write this masterful comment
@@MrJoelerjoeler if not now when
bingo! totally correct dude
soy
.... and a little verbose and over-earnest
I have bought so many stocks in individual companies. There are so many stocks going to rocket in the long run, right now safe to invest in an ETF that tracks the S&P500 and ride it out. I put $130k into some growth stocks with a Financial advisor handling my portfolio. some of my picks are, NVR, LISP.SW, SEB, VOO,DOWJ, BRK-A, AAPL,IVV, NYSE, NASDAQ, TSLA, I've gotten 82% return so far this year I'll see where it goes.
that's quite impressive, you surely made a good bit of money. I myself invested in warren's BRK-A stock quite pricey but totally worth it.
this is more like it, I have always said if you are going to refer your advisor at least let one know specific plays he or she uses to make gains in the market, nice picks. you've got great stocks, just started a few weeks back, I'm going for long term, I'm still trying to wrap my head around it, can I get the number of your F.A ?
@@AshtonGrace AUTUMN LYNZI SMITH
@@AshtonGrace search about her on your browser
Terry Eagleton wrote, “by having to grapple with language in a more strenuous, self-conscious way than usual, the world which that language contains is vividly renewed” and I find that to be one of the most compelling arguments for what the humanities accomplishes that is constantly overshadowed by the glitz and glamour of the development of new technologies or AI
In high school, I learned more about writing and thinking from debate than I ever did from an English class. Getting feedback at a tournament (i.e., lossing), going back, editing, then going back the next week and getting more feedback (i.e., losing again) taught me how to really think. I think that type of work is what, at least at the high school level, teacher are going to need to be assigning and teaching. You will need to be grading students based on logic and rhetoric, often displayed in real time in front of you.
@@hashimrahman51 Not if wars happen.
Civil War happened because of people's beliefs and anger.
You really think billions and billions are gonna sit in poverty and problems whilst the world robs them off their jobs? No.
There'll be fire and violence, demanding.
sure, but when ai does all work for us we will free ourselves in many ways. humanity will find a way, in fact we’ll probably have ai in our own bodies. perhaps we’ll have ubi. we could solve world hunger, cure diseases. sure this looks scary if applied incorrectly, but imagine a world where nobody is required to work for 12 hours every day to provide for their family, where every person has a comfortable home.. where everyone is fed , addictions are solved. imagine if the tedious activities in life could be eliminated. sure work is a good part of life, but with the work done for us what work we do wil be up to choice. and that’s freeing. hell, rich people already get to decide what the hell they wanna do. pretty soon everyone will. to follow their passions regardless of economic gain.
czcams.com/video/1O8efHrxmy8/video.html&ab_channel=Authenticinquiry
Came for the ChatGPT content, got hit with a mildly existential crisis regarding the concept of understanding uniquely through the process of writing. Loved it. Thank you.
These AI tools remind me of the Library of Babel. Where all text (with some constraints) is already written. The only problem is, it's just as difficult to find any specific text as it is to just write it yourself. I don't think these tools can kill artists and writers because of how personal art is.
The problem with these tools is that people have to work to survive. Companies can use these tools because they are cheap and way more efficient than a human. There will always be value in creating art, but I definitely see it becoming even more difficult to succeed financially than it already was.
Not only that, but the inherent non-trustworthiness of this incarnation of "AI" makes it that much more useless to anyone with a critical mind. I cannot believe people are letting themselves be misled by its auto-complete responses. OpenAI must be the most unethical company that has ever existed, to have enabled that on a massive scale.
As a math teacher, I experience that very question way too often! "Why do I need to learn this, when I can just use a calculator?". It's so difficult to help kids understand that without this knowledge, they won't understand the subtle nuances in mathematical operations, or how to rearrange math in a smarter way - or indeed understand what they are actually calculating! Only through an insight into the details and operations of the calculations will they gain an understanding that allows them to see the underlying patterns and meaning in mathematics. At that point, the calculator becomes a useful tool for routine operations.
Ultimately, tools that perform operations for us, should - in my opinion - only be used when we have learned the patterns themselves; and they should still be trained regularly. as someone else wrote, the brain needs exercise as well. Having electronic brains perform the work for us, lessens human value and usefulness in our materialistic world, and in a sense diminishes human meaning while having nothing to replace it with.
Creative thought, whatever form of expression or thought pattern it takes - written, pattern, numerical, visual, auditive, tactile, etc. - is something that should be nutured, and it grants us a strong skillset that can be used in a variety of areas in life. We should teach kids not just to be productive citizens, but also to be humans - and to be able to carry that torch of humanity forward.
AI is inevitable - but how we use it, and how we include it into our society, is critical for the future of our species. This bot might not be the great breakthrough - but AI is developing rapidly, and if we're not careful, we won't have a place for ourselves in the world we're creating.
how do you explain to a student why they need to learn to see the underlying patterns and meaning in mathematics?
@@Borat69able It's difficult, and relies a lot on you getting to know and relating to your students, and vice versa. The relation you have with your students, and the insight you have into their mathematical knowledge and understanding, and their conditions going into the subject, are both key to reaching them.
Investigative mathematics and mathematical modelling are both important tools to use, since it encourages mathematical thinking. Investigative mathematics encourages you to reflect upon mathematical patterns, alone and in groups; and mathematical models relates mathematics to the world at large, and encourages you to see that relation and the underlying mathematical patterns present in the world at large.
Both of these require a foundation of knowledge to be present, mainly basic mathematical operations. That's what's so difficult about learning: Sometimes you can't get the full picture until you've gone through the learning activity. You might not understand why you need to learn fractions, until you realize later on how it gives you another way of grasping and thinking about numbers, division, and relations. Fractions are actually one of the most fascinating and useful things in mathematics, and it makes everything else much easier!
So to answer your question: Since it's almost entirely impossible to grasp exactly why they need such understanding until they actually start working with it, you'll usually simply go through the necessary routine mathematics, and ensure lots of reflection in groups and in plenum (or individually) on the subjects they discuss. Ensuring clear communication as to the purpose of the subject you're approaching, and how it relates to other subjects or to the world at large, also helps create relevance.
My personal arguments (from the viewpoint of someone who understands the value) is that such an understanding makes all mathematics much easier, and it helps relate mathematics to the world we live in. An understanding of statistics and the underlying patterns and mathematical behaviours, will help you evaluate numerous developments in human society, and help you not being misled by political spin, for instance - but this requires you to have the indepth understanding that allows you to transcend mere numbers, and instead grasp meaning.
I don't think there's a simple answer to the question - schools around the world do things differently. Here in Denmark, our schools are very different from those in the US, for instance.
I hope this gave you some of the answer you were looking for!
The question is why would these nuances be relevant to that student?
They might be, they might be not, no one really knows, the best shot of knowing that has that student.
And you can argue the student can't know that because he is too young... But then how do you know that you aren't too young either?
The truth is that school is still a tool to enforce conformity. And that's understandable, it makes life easier at first. But that's also why people will always try to break out of that conformity or be bogged down by it.
That's why the older ones are weary of change as they have carved out there place in life and don't want to loose it, while the younger ones crave change as they are still up to carving out their place. And within the two, the wrestling evolution occurs. =)
U could just restrict the use of a calculator and let them figure it out for themselves. And honestly, just like working out, I think we need similar work out for adults because using calculators all day has made me reliant on them to a frightening degree.
Bad math classes are about plugging numbers into formulas.
Good math classes are about puzzles - what do these equations have in common, how can we shape one into another, how can we take advantage of these similarities to identify their unique essence?
As someone who has worked with IT for 30 years I immediately recognized chatgpt as next level and I knew it would be very disruptive. However I know it’s next iterations (coming soon) will be much better. So this version writes like a high school kid and still needs a human to edit the work, and it doesn’t do math well, and makes factual errors-do you think that it’s not going to be 10x better in a year or two? Don’t get complacent about its weaknesses now and think this doesn’t change anything.
As people volunteer their data and time to train this it's just going to get better and better.
I completely understand where you're coming from but this document changed my perspective on AI's within the art field
docs.google.com/document/d/128yey0VfYhM9eUdvkvCpk5zvvoIkqXfI4hEPAYeJCHU/edit (PS: later found out that it's also on youtube as a video)
@@asbjrnfihl1039 As an artist I will tell you its far worst. Currently the young generation are already quitting to persuade the art carrier because of the AI. Nobody wants to compete with a machine, and trying to live from art have always been a challenge. So there will be massive drop in people with actual skills to be able paint for themselves and from imagination in the next 20-30 years. People will rely on machines to be creative instead of them. And in few years from now humans wouldn't be required anymore for full animations, movies, full games and so on to be created. All automatic and automated. And this is not all. Art have always serve as the soul of the humanity. Art held in check the psyche of the society, the sanitary. Replacing it with a machinery is a nightmarish dystopian stuff. I hope general public will wake up, but little hope. Too many people see all novelty technology as the future, as in a state of a trance, they don't accept critiques or pessimism about it - no matter how harmful it is. Because all pessimism now is backwards for them they will never listen. They want to create their digital deity.
@@LyubomirIko it will push artists to new arts and better expression of their soul.
That it is difficult to earn your money through art has a good reason. If it becomes too generic, which is too similar to regular jobs, it loses most of its value. The point of art is to elevate the mundane and we have plenty of artists that just recreate the mundane because it feeds their family. And those will not make it though.
It's just the nature of it, if you want the guarantee of food on the table for your kids then you work in a bank. If you want to transcend mundane reality and pull the world with you, then you need the risk of failure. And as unfortunate this is for the individual, if you want guarantees you need to get a different job. If you want to be at the forefront of evolution then you can fall off the cliff, that's just how it is.
And as AI will churn the wheels of the economy, the pressure will spew out new artists that will transform our understanding of the world. And for any one such artist plenty of others will just not succeed. Just like the countless of athletes that do their best and still have to watch another on the throne.
That's the nature of risk, that there is the chance to really fail.
@@sonkeschmidt2027 Art isn't and shouldn't be about blood and clifts.
Human condition is just that - human condition. No matter how hard you push yourself off a cliff you won't be able to win against chess AI. Or against biomachine 1:1 in the Olympics. And so on.
I'll say this, every ChatGPT text I've read has felt like the literary equivalent of the Uncanny Valley. There's something distinctly... off about the way it writes that betrays a lack of personality behind the writing.
The problem is that most current mass-produced written works already lack personality, having to be generic (a.k.a. inoffensive/apolitical/unopinionated) enough to appeal to a mass audience that is decreasingly interested to read long texts. A text written by ChatGPT can sometimes disguise itself well enough to look authentic, and a less discerning audience won't know how or care enough to tell the difference.
But then again, AI still hasn't completely replaced humans in a lot of aspects yet. For example, AI art has been around for slightly longer and is still never not creepy. Who knows where this technology will lead to or how society will react to it.
I feel the same. It can write. But it cannot write well, like a writer. It is great for transporting info that is already repeated a lot. But that's it. No soul or personality obviously
I think chatGPT will go on the same path as AI art in that we will use it to form a "generic foundation" of sorts, then edit it to inject personality and our own understanding, thereby increasing efficiency. Some artists are already using AI generated images as a base for their painting, the end product doesn't look soul-less like a pure AI generated piece but the artist was able to save hours of work drawing the background.
Noticed that too it's a bit too sterile. But what if you just ask it to rewrite the comment with more personality? Or in the style of someone specific? Might already be done
You are totally right about ChatGPT not feeling quite 100% authentic when you read it. But the thing about technology is it keeps improving itself year after year. Humans can only improve so much. Evolution is very very slow, and a single persons lifespan is limited. Technology like this will no doubt be better 5 years from now than it is today. What about 25 years? 50 years? We're about to witness endeavors that were uniquely human since we started painting cave walls and chiseling into stone tablets be replaced by AI. Maybe not today, or tomorrow, but likely within the lifespan of younger people alive now. It is a big deal I think.
ChatGPT wrote this comment; it is already self aware.
The one thing ChatGPT cannot do is care. Care about the outcomes of its processes, care about the damage it can do to education, care about the connection between disparate subjects that result in creative progress.
Nothing about “rote memorization” in essay writing.
What writing - and the entire process - does for your mind is equivalent to what weight training does for the body. You cannot just carry a gym membership card in your wallet and call yourself fit. Carrying you education in your pocket is the same thing on a monumental scale. If you cannot write, you cannot think. Period.
Similar analogy is for drawing. AI can paint within seconds what a human would do in hours but the process of drawing makes you analyze what you are looking at, strategize your next strokes and reflect.
This is why actual general AI equal to the human mind is a long ways off. It can do what various parts of the brain can do, alone, but not the entire whole. Actual general intelligence will be required to utilize the full potential of these nueral networks, though by then it'll likely have even better language, art, etc capabilities built in.
@@hashimrahman51 This isn't a problem. While creativity being eliminated can be argued as an issue as it is not solving any actual problems, malpractice is among the leading causes of in hospital deaths, and eliminating doctors and replacing then with a better alternative, as harsh as it sounds, is only good for humanity.
@@cortster12 When we finally achieve AGI and later ASI, what will be the point of humans? Is there any point in keeping them around when so much of the data the AGI/ASI will inevitably find suggests that humans cause a majority of the problems in the world. AGI/ASI will theoretically be able to do everything so much better than their creators. Why not just remove humans from the equation entirely? Really hoping that future AI doesn't come to that conclusion.
@@omnipenne9101 That's the question, isn't it?
In high school, I often found myself doing badly in both math and english, because they failed to provide context for how I was going to use them. I actually do enjoy writing, but my talents are more focused towards crafting tutorials and doing tech writing than it was towards narrative storytelling. I've also had to teach myself more advanced mathematics in my adult life. I certainly would have paid more attention in class when I was younger, if I had been inspired about the tangible applications for it in the future.
A numberphile video had an excellent analogy with this on why most people hate math. It's like if you had an art school that never let students look at the masterworks, but instead forced them to constantly draw squares and circles
An interesting thing about ChatGPT I didn't hear mentioned, is that you can actually feed it examples (as many as you have the patience for) of a thing, and then ask it to do something with those examples; including, asking it to write an essay in the style of sample essays you provide. You could, of course, mimic other people's styles - but, to me more interestingly, you could feed it your own essays, and ask it to write you a new one in your own style. There's not really more to this thought, I just found it interesting!
shoot, here I was , thinking about how we could use previous students essays to test similarity vs machine made-ness. I suppose not.. and that gptzero would be made useless by this? yes?
Yeah seems like in person exams and interviews/presentations at the only way round this problem
Man, I love this channel. It always comes across as boastful to say that laymen don't really understand what it takes to produce a piece of literature, film, or art. The very process is intended to make the finished product look or feel 'effortless'. ChatGPT can only produce text based on what other human beings have already written. It is profoundly adept at synthesis, but that is precisely because it is as synthetic as you can get. ChatGPT cannot write your wedding vows, as you would write them, to your unique spouse. It cannot write a poem about that one perfect Autumn day you spent with your family on a holiday that one time. We should be thankful we have access to such astounding technology, but we should also realise that as we cede further and further ground, we lose our humanity, if only an inch at a time. Thank you, Nerdwriter, for this one. Right now, the world needs good words, and better people to write them. -JC
What about fact checking essays (generated by AI or others) as a new form of exercise in school? It is, and is becoming, an increasingly important skill in the modern world and also requieres knowledge of the sourcematerial. Of course it doesn‘t help with the outsourcing of writing and ideas themselves but it might still be a way to adapt
You bring up an important - vital - point. I have tested chatGPT on some factual items. In one case I asked it to abstract a particular research paper. It dutifully wrote something, but I was stunned when the text it output was 180 degrees away from the actual conclusion of the research paper. So - would you want chatGPT to fact check anything? (And I see that is not what you are saying).
I have also input a question that bore upon a particularly controvesial point in medicine. Interestingly, chatGPT took one side of the argument, and promoted that, ignoring arguments on the other side. Even when prompted to "Show that the contention that XXX leads to YYY is wrong" chatGPT took sides in the argument and produced text saying that XXX leads to YYY was correct".
This brings up another important point. Humans are shockingly willing to believe the output of a computer instead of their own conclusions. The reason this is so would make an interesting discussion, but I have seen this multiple times. So - my great concern is there will be 100s of thousands of content creators who now can generate unlimited amounts of text that contains errors or, if not errors, a misleading conclusion. People will believe these errors rather than question the output of the computer.
It's going to be a rough ride
@@Rolfrad
Current technology is more form over substance. I heard it summarised as implementing an intuitive type I system (in Kahneman's terms) without analytical type II system.
That should be a required class on any core curriculum nowadays, really. And ChatGPT is a great example to use. It creates such plausible sounding BS that it's hard for people to immediately see that it's BS. And that's the danger.
In a test we ran in class, it invented citations. They all sounded extremely believable, all titles and names and journal titles sounded great. Some had DOIs (and some of those DOIs even lead to actual papers, albeit not the "cited" ones).
There is so much actual, willful BS on the internet, where people insert citations but actually misquote the papers, that this next step seems extra dangerous. So teaching how to deal with that is important.
Been using ChatGPT and Sudowrite for the past few weeks. Super cool tools. ChatGPT helps me brainstorm characters and ideas much faster. I take what I need and leave the rest. Sudowrite helps me see my own writing in new light and helps me with revisions. I still do the bulk of my writing longhand and typing. But these two tools have helped me improve my own work at the brainstorming and editing stages.
If you were to know an actual human, that is smart enough to write to you the answers to your questions - would you call him a tool?
@@LyubomirIko yes. Editors are considered tools too. It's not that far fetched for authors to be considered tools lol.
@@danielhobbyist The author is not a tool, what are you smoking? Cow's poo?
So you think you are a writer?? No way.. it is ridiculous
As a human adult with ADHD, I've been using chatgpt as an assistive tool to help me figure out what people are trying to say when they're vague in emails and chat. I've also been using it to help me get started on tasks that are otherwise tedious. It's a neat tool.
When I have some kind of creative or expressive work, I'll use it to edit, because I enjoy writing.
How did you use it on vague emails? Did you just copy pasted the contents into the Chat input and asked it to rephrase it?
@@defdac you can literally ask it: what does this email mean? (And then you just copy paste the email).
I’ve done this quite a bit with song lyrics, since I’m a songwriter myself I was curious to see what it would come up with. Sometimes it can be quite accurate but most of the time it is super surface level and generic. Also it really struggles with irony and sarcasm since a lot of the songs I asked it to interpret the ironic parts completely flew over it’s “head”.
@@lacadam nice, thanks!
@@defdac "What is this person trying to tell me?" Or "Summarize this in short bullet points"
thought you were an adult dog there for a sec, thanks for specifying
That "high school student happy with a B-" analogy is great. This is essentially the experience I had. I did an experiment with my class (who are at least sophomores or even further in their studies). They had to write their own assignment and then give chatGPT a prompt and try to get it to write a similar essay. Then we tried to see if they can tell them apart. It was fairly easy to tell them apart. ChatGPT delivered what a lazy high school student would deliver to get a C on an assignment, essentially, even down to the pattern of which references it used and how. That, in particular, was quite interesting.
What a beautiful video! The way you introduce the topic, tackle it from both sides and loop it back around to the conclusion was so incredibly smooth! Its like I listened to the audible version of wrapping a bow.
that "memory to internet" thing is terrifying because i used to do exactly that, but moving to a small town with weak internet signal and sometimes blackouts, i begin to dedicate portions of my time to writing on books again. written media survived to this day for a reason
High school teacher here. It really is a problem at this level and it's driving teachers to advocate for serious change but no one listens to us anyway.
Try to embrace the change instead of work against it! The use of AI is an inevitable change you cannot avoid.
@@niek7422 the problem is not that they don’t want the use of AI. the problem is that they’re trying to get students to learn, to think uniquely, to come up with their own arguments, etc, and that’s the challenge that an AI writing FOR u creates. Nerdwriter says it better than I do 😅
@@niek7422 you're preaching to a dedicated pc builder and gamer, I know. Hence why I said teachers are advocating for change. Project based learning that is contextual to real world communities and intrinsic motivation for reading is what we've fighting for and it's never a winning battle for us.
If I was a teacher I’d probably dedicate more class time to traditional homework assignments to ensure students leave school with some degree of writing ability.
@@jaewon4316 I totally agree with you and the entire point of the video. However, teachers should see this as an opportunity to implement AI into the writing process rather than a mere problem. Teachers should know the limits of stuff like chatGPT and teach students about it. At the same time teachers can try and find ways to implement AI into learning exercises rather than completely avoiding it. The tool will be there (and will surely improve). It is up to teachers how to teach students the real value of writing uniquely, even with AI tools available.
I used ChatGPT for a uni essay. Partially because I had only a couple of hours to send a 6 page essay I forgot about it, and partially because I wanted to test it.
All I did was ask "What is XYZ?" "What the advantages of XYZ" "What are the downsides?" "How to implement XYZ" And other probing questions... get all the answers and format it.
I changed very little, besides changing some similar lines between answers and also changing bullet points into paragraphs.
Got an 8/10. For 30 minutes of work... on what should've been an 6-8 hours work.
Frightening.
If it finally tears down the broken and corrupting education system for real, that'd be one actual positive.
Plus you learnt how to use modern tools that are gonna be valuable knowledge in the future. Congrats!
I just failed a writing student for doing what you describe, as in getting a zero one one assignment tanked his semester grade. And I didn't need an AI detector to identify the moment he leaped from writing on his own to using ChatGPT, though the detector is handy to point to as a backup.
I also feel like I failed him (as in the sense that I, as an instructor, failed in my job) because in a pinch he turned to ChatGPT to complete the assignment.
I did catch him earlier in the semester using ChatGPT and gave him a warning, so to see him turn to it again is sad. I'm sorry your instructor taught you the lesson that using ChatGPT is acceptable.
@@briandavidson3390 It's useless try and prevent people from using chatgpt. It is like when teachers 50 years ago prohibited using calculators in class. My uncle couldn't use calculators in his calculus class in uni. When I took calculus it was mandatory the use of a calculator.
If you don't want your students to use ChatGPT or other AI tools, which only will get more powerful and useful, you need to redesign your assignments so their use doesn't interferer with their learning.
Old school essay style is dead... trying to prevent it's death by prohibiting AI tools is a losing battle.
Really appreciated the reference to inherited rules of language and the baggage (for better or worse) that comes with it. Drives home the deeper value of knowing how to write - as a way to make sense of the world, its limitations and biases on thinking, and opportunities to break through those limitations in your own unique way. Or to put it more bluntly for the kiddos: If you don't learn how to master language, someone (or something) else will decide what the world means for you.
What a timely video essay! As a high school English teacher, I have spent a good portion of my winter break considering how I will introduce or acknowledge ChatGPT in my classroom. Tomorrow, I will spent to my principal seeking permission to include this new technology through a curriculum focused on argumentation and fallacious reasoning. I stoked to see how this will change my profession, my students' learning, and our creative world. Wish me luck.
What a lot of people forget when they make these kind of hot takes that "humanity and our passion will prevail" is that this technology is only in it's infancy. It took approximately 7 years (around the same time as a undergrad + a masters degree) since OpenAI was founded for this tech to get to this point and it's only speeding up exponentally from here. So in the time it takes for some above teenager in a classroom with a below average teacher (and lets be honest here how many of us actually had good teachers, teaching has been a downhill career for a long long time here in America and it's not going to change any time soon) to get their degree they're more than likely going to be eclipsed by technology.
Even if a smal handful of consumers stand their ground and reject AI and all that it stands for the large majority of the consumers will not care. If an AI is able to write with as much (if not more passion) as one of us flesh bags, you can bet companies will be pushing this tech hard (so they don't need to pay said flesh bag) and consumer will not care either as long as they get their content.
TLDR: Aritsts and creatives speak from a place of passion, but what they don't understand is that what they do is just a lower level of what AI is capable of in the long term (in a few years), i.e. they take in content, synthesise, embelish, hit publish. They take 20 years to learn this whereas AI is going to soon do this in an afternoon.
No. Artists collect life experiences, which form their taste, personality, beliefs and how they interact with their surroundings. If they make art, even if it is low quality and utterly generic, these things flow into it, in a way they can control what experiences their using. They also make decisions with every line, the art is created through the process.
AI has instructions that tell it how a thing looks on average, derived from millions of images. If you tell it to draw that thing it tries to stick to those instructions. There is no process, you demand, the AI gives, it's not creativity but consumption.
So, no, artists aren't doing a lower level version of what the AI is doing, they're doing something completely different.
Obviously, a Buzzfeed article can be done by an AI (and better!).
Lord of the Rings though? That takes "life" experience
Erik Hoel and L. M. Sacasas, from their respective Substack platforms and perspectives, have been writing some fantastic essays on what AI technologies like ChatGPT mean for our society and culture. They're both worth checking out.
How would they know what it means for society? What's their background?
@@lorenzomizushal3980 What background do you need to commentate on society?Anybody can do that. The AI can do that too.
Great video, it's important for people to understand the potential dangers and limitations of language models like GPT-3 and ChatGPT. While they have the ability to generate human-like text, they are not sentient and should not be treated as such. It's important to use these models responsibly and with caution, especially when it comes to sensitive topics like politics and personal information.
ChatGPT and Wall-E are very dangerous. Their mechanisms for preventing harmful material are astonishingly weak, and they can be used to disseminate vast amounts of disinformation, create harmful pornography, create plagiarism, uproot entire industries by displacing workers - the list goes on. Also can be used to impersonate humans, which is what you're doing right now and I think it's completely unacceptable TBH. Quit it.
Writing as part of the process of reflection and learning is invaluable, and you've captured that well here. As an instructional designer for a university, helping students connect to the reason why they should do an unmarked, formative learning activity is at the heart of what I do. If there isn't a solid reason why (feedback and practice in preparation for a higher stakes summative assessment, contributing to a group project/body of knowledge/learning artefact they can later use in a portfolio or reinforcement/self-testing on foundational skills and knowledge) - students will either ignore or outsource their effort. Academics know that students are typically strategic about what they do - whether they are teenagers or in their 60s,
Honestly, I don't blame teenagers for using it on generic essays written a million times before them. It may push teachers to personalize learning a bit more and to teach the students how to find material that's interesting to them and then synthesize that knowledge instead of just assigning the same book to read and write on for twenty years.
I also worry that it will make a lot of people overly lazy to find new learning material and trust whatever the AI is happy to churn out.
How would personalizing solve the problem? The reason that students write on the same books and topics year after year is to make them culturally literate. If I write a personalized question for each student or have them write about their own lives they can still use a chatbot. Personalization doesn't stop anything. I'm sorry you didn't understand when you were in school while you were being asked the same question that other people before you had also been asked. It was to bring you into society and so that you would be able to have a conversation with other people BECAUSE you had before considered the same topics.
@@oliviabasham Getting some general culture and having to write an essay on a topic the teacher is going to feed you the explanation anyway are not the same. You don't need to synthesize a topic that doesn't matter that much to you in order to get culture.
@@oliviabasham This is the question that Montessori educators have asked and answered. Teaching from a child's natural interests isn't a silver bullet on it's own, but it's something important that's vastly overlooked in the American educational system
@@oliviabasham personalization isn't making students only write about things that are immediately relevant to their personal life. it's about letting students have a say in what they get to learn. imagine if instead of supplying a specific topic that you spend several classes covering which your students just have to regurgitate knowledge about, you introduce a more general topic and allow students to formulate their own specific essay questions about it and explore. students will be a lot more interested if they get to have a say in what they learn, and having to conduct their own research and pick their own research topic is going to develop even more of their critical thinking skills.
@@ishathakor You aren't understanding my point which is that what you yourself describe doesn't solve the problem of using a chatbot in anyway at all
Love how deep you considered this topic. Keep writing
This channel and Every Frame a Painting were my introduction to the world of video essays. Haven’t looked back ever since!
Great video as always! I listened to your audiobook and really enjoyed it. I am hoping you’ll do a video based on the urban planning chapter. I was especially inspired by that one!
I used ChatGPT to write an analysis of an album that I know well and it was full of errors. It claimed certain songs are on the album that aren't, made up subsequent albums, and even provided fake urls as references. Beyond all of that, it gave me the most superficial analysis possible. It may have been that the data it was trained on didn't include a lot of text about this album in which case training data is of utmost importance. I worry about not just the errors that could potentially go unnoticed but also the way it could work as an amplification engine for existing perspectives from the training sets.
It's a language model. It predicts the upcoming text character by character, so that the words go together as they should according to the countless examples. This implicitly encodes a lot of real world information, but the program doesn't explicitly look up data. Expect future versions to do this (as well as run things through an explicit calculation math module when needed).
ChatGPT doesn't have access to the web. So the amount of in-depth information it can gather is not there yet
@@nostrace How was it not trained on the web?
How to get around chat gpt.
Super simply:
Do a draft submission, that counts into the final grade of the essay.
Basically one week after reading the book you write the first draft. Two weeks later final draft.
Even if someone used chat gpt. If the teacher said use the drafting days as reference material to alter the essay.
You would still have to write and tweak it in such a way that showed both changes and didn't score over a certain percent of plagiarism checks on both.
Only works for literary analysis. What about a history course? What about creative writing? Simple solutions are comforting but they don't solve the problem at all.
@@oliviabasham
The basic solution is the do a draft, bring it in, have 2-3 peers review it and then bring in a final draft.
Works for every humanities class you have in high school.
That method was used in English, history, creative writing all of them.
I went to hs in 2010-2014.
And that's all of them.
That's basically all the essay based projects you did.
Even history fair was done that way.
Teachers didn't give weekend essays that we never not came back too.
Like those weekend essays, became essays we redid, over the up coming week.
But like a 1 done essay, those were still hand written. Those were the essays you did in class as like test prep.
You know for AP test or the state writing or sat essay prep.
Like this doesn't work for science and math essays. But the ones you do in high school are basically experiment reports, and chatgpt would need your raw data and synthesis that into something.
The other alternatives are the ones that were being used for years, plagiarism checkers and formating and quoting guidelines, aka MLA, APA & Chicago and forcing kids to cite text.
Which was something used in all my humanities courses except creative writing.
However Creative Writing used only the peer review method I explained earlier.
That was well thought out. I really attest that writing, whether it be for university essays, journaling, or planning out something, really helps with my understanding of my changing self. Basically, writing might just be the main reason why I can have independent thoughts
This video is really impressive on a technical level. It's so well edited and well made in general
Another BRILLIANT video essay. Just so you know, I checked out your book at a LIBRARY. If I could afford it, I would have bought it. Thank you for all the beautiful thinking you're done, and continue to. Mark
If you were at the library why didn't simply borrow it?
Surely you don't have to pay to join a library!🤔
@@andiemorgan961 our library had to buy the book to provide it, which they did.
Got your book for Christmas and just started reading it last night, so it was interesting to see this video through the lens of the first essay in the book. Congrats by the way!
I used GPT-3 to help me write a cover letter for engineering internships I'll be applying to. It wasn't that I had it wite the entire thing, I just needed help getting started, wrapping it up, and proofreading. I gave it some information about myself and asked for a generic introduction. In a snap, it had it ready, and then I did the rest. I used it mostly to make sure my tone was appropriate and that I using concise language. There were times I modified or rejected its revisions. It's good, but you still need to be a proficient writer for tasks such as a cover letter.
So you couldn't do something as simple as a cover letter...?
@@fistofram5526 read my comment again.
In an ideal world, I think the relationship between people and AI is less one of an editor and more one of a director
seems the person directs the ai (or whoever trains it) and ai directs the person
Woah I watch your stuff dude! Big fan!
but for that to happen, the director needs to be more competent than the AI, sadly not everyone is
@Casual Take In some ways yes, in some ways no. A movie director doesn't need to be a good actor
There can only be so many directors.
Great video! I completely agree with your point about the danger of relying too heavily on artificial intelligence to write texts and essays. While AI can certainly be a useful tool for proofreading and suggesting edits, it can never replace the critical thinking and analysis that comes from actually learning and understanding the material. Plus, using AI to do the work for us robs us of the opportunity to improve our own writing skills and develop our own unique voice and style. Thank you for bringing this important issue to light!
---Unironically written by ChatGPT :P
I do sometimes use AI as somewhat of a tool to inspire but never to do something for me, same with AI image generation, as a writer and filmmaker I think it's great to visualize something the way you envision it for those who have a hard time doing that just from text and I think AI like this is something that should continue to be used to support but not be the one to do something for you.
You know what's kind-of funny? I found your channel just now after asking ChatGPT for recommendations of CZcams channels I might like based on a list of channels I enjoy that I gave as input. Since I had not watched any video of yours yet (or so I think), this one seemed fitting. While this video was playing, I continued to go through the list of suggestions, and out of interest, I started to look up the subscriber counts, so in a different tab I was searching for one channel after another. And one channel that came only 3 suggestions after your channel was Mark Rober's - who has just released a video called "Amazing Invention- This Drone Will Change Everything". Reading that title about 20 seconds after you stated "Human beings love to write the words 'This will change everything'." made me chuckle for a bit; it's just coming all together too perfectly.
Best wishes for the new year, Nerdwriter crew! :)
It's tempting to use chatgpt to write for me. However when i look at the whole writing process, i tend to rant for pages on end and I really enjoy that. Editing after the fact to play with voice and synthesis many options, not so much. To retain my voice but be efficient, I'll likely outsource more editing to ai than writing. Linguistics wise, I'll stay ahead of the curve by not giving into easy "writing" with chat gpt, but i will still gain the efficiency i desire when i can ask the bot to make 10 versions of MY paragraph while fixing all of my awful grammar. Hopefully reading dozens of versions of my own work will also let me be more creative and expirimental over time too.
Edit: after reading other comments, i was reminded how chatgpt sucks at making connections at times (the whole point of an essay). Often these corrections are false or it can't be creative. I've asked chatgpt to find similarities between vastly different subjects, like a biological organism and an atom, and it often states there is none. This happens over and over. My niche in writing is finding absurd, universal abstractions for everything. If i want those thoughts out there, as for now, I am still the one who must write. But since chatgpt is so good with language, making it asmr to bullshit so well, then my editing idea checks out. Chatgpt is a language calculator, so it won't make any proofs, but it can reduce the nitty gritty computation needed when just trying to find a good replacement word in a paragraph or a different order of sentences.
Small and medium-sized businesses will adopt ChatGPT because it's cheaper. I doubt artists will use it; it makes no sense for them to outsource what they wanna say.
Applies to AI image generators too.
Happy new year bud! And everybody 🎉
a small note from an engineer doing a lot of calculations: we outsourced *calculating* to calculators, actual *math* is still very much a human endevour, we just eased the boring and tedious parts.
*I asked ChatGPT which jobs it will replace in the future*:
🤖 Bookkeeping, accounting, and auditing clerks: AI could potentially be used to perform tasks such as reconciling accounts and generating financial reports
🤖 Data analysts and scientists: AI could be used to analyze and interpret large amounts of data more quickly than a human
🤖 Medical and health services managers: AI could potentially be used to assist with tasks such as scheduling and managing patient records
🤖 Insurance underwriters: AI could be used to analyze and evaluate insurance applications
🤖 Bank tellers: AI could potentially be used to perform tasks such as processing transactions and assisting customers with basic banking inquiries
🤖 Accountants and auditors: AI could be used to analyze financial data and identify patterns and trends
🤖 Data entry clerks: AI could potentially be used to process and organize large amounts of data more quickly and accurately than a human.
🤖 Customer service representatives: AI chatbots and virtual assistants could handle simple customer inquiries and complaints
🤖 Transcriptionists: AI could be used to transcribe audio and video recordings
🤖 Retail salespersons: AI could be used to track inventory and assist with product recommendations
🤖 Food service workers: AI could potentially be used to assist with tasks such as taking orders, preparing food, and cleaning up in a restaurant or food service setting.
🤖 Delivery drivers: Self-driving vehicles could potentially be used to make deliveries, reducing the need for human drivers.
🤖 Postal service workers: Self-driving delivery vehicles could potentially be used to deliver packages and mail
Market research analysts: AI could potentially be used to analyze market trends and consumer behavior
🤖 Medical records and health information technicians: AI could be used to organize and analyze electronic health record
🤖 Construction workers: AI-powered robots could potentially be used to perform tasks such as heavy lifting and welding on construction sites
🤖 Electrical and electronics installers and repairers: AI could potentially be used to assist with tasks such as diagnosing and repairing malfunctioning equipment
🤖 HVAC technicians: AI could potentially be used to diagnose and troubleshoot issues with heating, ventilation, and air conditioning systems
🤖 Plumbers, pipefitters, and steamfitters: AI could potentially be used to assist with tasks such as diagnosing and repairing
🤖 Manufacturing and assembly line workers : AI-powered robots could potentially be used to perform tasks such as welding, painting, and assembly on a production line
I wonder what humans will be after that lol
you must've surely heard how your teenage self looks a lot like john krasinski. great stuff as always
What do you think about ChatGPT as way to help bridge the gap between those with parents who can help develop their child’s essay writing style and those whose parents are less able to support them in this way?
When I was young, I struggled a lot with essay writing and my parents would essentially re-write parts of my essay or restructure it and explain to me why they did it like this. I’d then be tasked with applying this to the rest of the essay. This improved my essay writing exponentially to the point where I was achieving top marks in all my essays with no input from my parents. Without that fast feedback loop, it would have taken me far longer to learn proper essay writing skills, since the feedback loop for school essays is measured in weeks not hours and there’s no opportunity for iterative feedback.
I recently ran an old essay through ChatGPT and it seemed to fulfil a very similar role. It kept the content the same but improved the structure and style of the essay. When asked, it was able (to some extent) explain the changes it had made and give general advice on good essay writing practice.
One of your best videos in a while imo, and a great example of the kind of video that you specifically can do better than just about anybody else I've seen on CZcams
Recently playing with ChatGPT, I had it edit a paragraph I wrote. On the one hand, it streamlined and even clarified the content, but it also used words with slightly different connotations and took away the quirkiness in the way I write. It didn't change the overall content but modified the feeling and aspects that were uniquely me. I agree with some of the pervious posters in that it is a new tool and how we use it, if we do at all, will determine to what extent its effects are negative. I'm reminded of Socrates who is supposed to have believed that the written word was a poor way for learning and communication. His arguments had merit, but we can see where writing has taken us.
I’ve been messing around with it to see if it can replace my job. I write instruction manuals. It does a good job of writing quickly. But when you actually read it, the information is usually wrong. A project manager sent me a doc ChatGP wrote like this is great! And wanted to publish it. After myself and a few others read it, we were like this is pretty wrong. I think I read it described as “confidently wrong”. Which just reminds me how much incorrect information there is out there. It can only reproduce what it was trained on. I think my job is safe for now. It doesn’t correct errors in existing content, so those errors keep getting repeated. Not to say it won’t be able to eventually, but it’s just regurgitating for now.
It also struggles to create new information(we built something new and it needs a chapter in the manual)and organize information so people can confidently read the manual and have it make sense. Maybe in the future, but I’m safe for now.
That's the key weakness of ChatGPT right now: It cannot create.
John Green recently did an experiment where he started off a video by reading off a script by ChatGPT, and it was just a string of generic, meaningless platitudes that sounded coherent, yet was nowhere close to as authentic and vibrant as John Green speaking extemporaneously.
An AI does not have its own opinions or values, it cannot evaluate new information and compare it to past ones. I don't know how it could possibly do so in the future, but we've been living with AI for quite a while now and we still don't trust it to drive cars or replace visual artists.
Do you think it's realistic that in the not very far future, a better trained bot would write an instruction manual and you or someone like you would just QA it before publishing?
@@ashikmohds I think it’s far off because of how wrong the information is. If you could have ChatGP somehow localized and trained on the data at my job, then maybe. But again, it can’t create anything new. My job focuses mostly on writing new stuff and revising the old based on user feedback. An AI could get feedback and implement it automatically, but not all feedback is good feedback. There needs to be some evaluation based on the content of the feedback, will it serve the majority of the users or just this one. Is it even possible to fix this user’s problem?
@@nousersnamesleft I was thinking about a bot that your company would train. But you're right. Your job is to look at an object, understand how it works, and write about it specifically (oversimplification). That's brand new information that the bot could never have access to.
I think the vast majority of people see ChatGPT as a final product, but it’s really just a language model with many limitations. It’s true power comes when you build a system that also includes an API ,database, and codified set of behaviors for it. It’s capable of simulating characters, worlds, interactions, etc.
We’ve heard a lot about prompt engineering (a term that most users of ChatGPT aren’t even aware of), but less about language model “programming”.
I feel like people are extremely flexible. To answer Who am I? What do I believe? there are near infinite ways to explore this question. Writing is how Nerdwriter found the answer to the question (and how I try to answer this question), but talking, singing, dancing, drawing, creating businesses, and thinking are all possible ways to explore identity and truth.
I do agree with that exploring life questions is easier if teenagers learn the basics of the exploration tool.
Your videos are so well made. Hope you had a good holiday season. Best of luck into the New Year!
In the real world, the skills you learn in secondary education will only be required of you by an authority like a boss if you choose to go into a career which needs them. That's your choice. If you learn this stuff, and find that you enjoy it, or are able to do it well, you might choose that type of work. You might be paid well for it and be glad to have the money.
Or you might want to use these skills for yourself. You might want to understand something to do with science or economics which is affecting your life, understand what other people believe, or express your ideas to others. You won't be required to do those things, but someday future you might be very glad to have these abilities.
Being a teenager is about becoming an independant person who has their own ideas and makes their own decisions. You have the oportunity in school to give your future self more options. You don't have to take all the opportunities you're given. You do have limited time. Choose where to put it.
Ah, an AI-generated text in the wild.
@@starryburst863 lol I am an actual human. I am autistic though, maybe that's what you're picking up on?
1:29 Well, in America you don't call anything cheating, even your aggressive capitalist model, here in Europe we call it cheating, because the lack of one's own thinking leads to laziness and lack of self-reflection.
I love your videos! As a history teacher for kids aged 12 to 18 years old I struggle with chat gpt too and this video helped me understand how important it is to keep teaching children how to write properly.
I think even more deeply, the potential problem is that of being reflective. outsourcing the creation of an act that can help one be very reflective (as your formulate and refine thoughts) could stunt the growth of that ability, unless it's cultivated elsewhere in life.
"I really appreciate the critical view taken by Nerdwriter1 in this video! It was a great reminder to think twice before relying too heavily on AI tools to do the writing for us."
man good I'm not living in the stone age anymore where I have to write my own comments
Yeah but it’s alway the most tepid, basic bitch vanilla perspectives on anything. Chatbot is perfectly adequate for people who were NPCs all along because comments weren’t worth reading when they were _organically sourced_ either.
@@gavranarh Well, that's not true, it can produce some pretty dope stuff that you would accidentally attribute to an actual person that has a way with words. Always depends on what you give it, plus some luck.
@@kareemjonson a way with words as in style wise, sure, though it's more in the uncanny valley territory from what I've seen so far - literally _generic_. but as far as content goes it's pretty light on creativity of ideas, just rehashing platitudes. it's basically the kind of content you'd get as a top comment on twitter on a default sub. it's nothing but a technology driven vox populi so far, and how can it be otherwise since it's been literally trained on datasets of people's actual interactions on social media and whatnot (AFAIK): not exactly cutting edge thought. not saying it can't be, though, if you feed it the right stuff, who knows, might outperform our leading thinkers.
It won't be long before there'll be a feature on one of these commercial LLM tools that allows you to fine-tune it to your style of writing, perhaps using a corpus (even just a few pages) of your work. If you include specific instructions on style of writing in the prompt then you might be able to get some-way there already with ChatGPT, although you might need fine-tuning to truly pick up on subtleties of how you write. Indeed, for example, you are already able to instruct ChatGPT to write in the style of specific writers it recognises (e.g Shakespeare), with varying results. If you specifically fine-tuned an LLM to write in a particular writer's style with examples (such as your own) then I think you'd find the results somewhat uncanny. Perhaps you would still need to do some editing to add more substance, and if you wanted to try something new you'd need to write some more, but I think this would largely address the main concern in the video.
I think you missed the entire point of the video... The goal is not to generate things that sound like what you write. There is more to writing, and more to life, than just producing a product
@@NihongoWakannai also that it takes a lot of writing to discover a unique voice!
What if you don't have a style yet though? It's not like you are born with one. It develops over writing a lot.
@@NihongoWakannai > There is more to writing, and more to life, than just producing a product
Yet our system works to either teach you how to produce a product, determine the worth of that product, and determine your status in society based on the value and quantity of products you produce. If you are not productive, its as if you almost don't have a right to live.
Even this video is a product, designed to keep attention of its viewers enough to show an ad for a product with the goal being some of the people watching will be interested in the advertised product and perhaps even buy it.
@@click.advance I can see a similar unique style being possible to develop through working with a LLM like ChatGPT. You can litterally discuss the style you would like the chatbot to write in. A natural language interfact would probably be easier than the sliders you see in the generative imaging applications.
Great video. Totally agree on how abuse of this technology could be an existential threat to our ability to express ourselves as individuals.
I was already weirded out when Gmail started suggesting auto-generated replies to me, even ones as simple as "thanks!" or "will do!" We express ourselves and build relationships with others through language, and the idea of outsourcing any part of that process to an algorithm is legitimately scary to me.
ChatGPT just blatantly lied to me. I told it to make a note of a particular issue it was having. And it clearly told me that it would .... exact quote ..."I'll certainly pass along your feedback to OpenAI. They value user input and use it to help improve and refine their models." Then I asked it if it would actually do that or if it was simply indulging me. Here's what it responded with... "When I mentioned passing along feedback, I was simply expressing that I would communicate your sentiments to OpenAI if I were able to do so. However, in reality, this is a hypothetical statement as I do not have the capacity to actually do this." WTF..... I mean WTF!! OpenAI is a joke and a scam and I'm furious about this and whole list of other bullshit I've had to deal with.
Writing is editing.
So, this basically means everything we write is just a remix of alphabets. Can't stop AI from that. The best counter we can find against AI is to be more human. AI can provide a rhetoric of anything but it can't diminish the emotional connection between writer and the reader. The best example is to ask chatgpt to write one of your videos without telling us. See how good it executes and viewers will still find flaws in the connection they generally get with your original scripts.
sculpture is just like... a busted up rock, man.
As much as I want to believe this, I see no reason why you couldn’t feed it the psychographic profile of your audience to write to. Won’t be perfect of course and will require editing but I’d be surprised if it couldn’t create some of the key moments that make a piece of writing ‘connect’. Especially if there was source material to feed it as there is for nerdwriter.
This holds up perfectly when it comes to AI art too.
Those who use AI for art aren't artists, they are editors.
Yes, exactly, call them Image editors instead of Artists
if they want to be called an Artists, then they should show their brush work, and paint in real time,
in front of an audience.
if they're really an Artists, this will be no problem.
most of the people that use Stable diffusion call themself an Artists, yet
they don't know how to draw in perspective, painting an accurate anatomy,
paint in multiple light condition, composition, and color mixing
When it gets adjustable enough so you can change any area of the image with the ai effectively , I dont see a big difference between the editing artists do and the editing ai artists do ,
They both recognise the parts that would work better with changes
It's close to artists using references . Instead you get an ai created reference closer to the idea you want from the start by using highly specific prompts
But the traditional artist can get it exactly as they see it in their minds eye where the ai artist will have more difficulty getting it close .
If that even matters though as the ai output can be good to replace the original vision anyway with some adjustments to that instead
getting exact details changed is the hard part that traditional artists can do readily
@@ayy2193 drop the Artist from the editing artist, the only title that they deserve is "Editor".
don't add any unncecessary title that are way above your skill.
don't fool yourself and calling yourself an artists when the only thing you do is writing a description for a prompt, and push a button.
Everyone can do that, even the Artist can do that, and can do it way better than you ever could.
having the ability to wite a prompt is not special, it doesn't give you any special title.
Art direction is so much more than that,
if you're clueless about Perspective, anatomy, Architecture, Color design, composition, Character design,lightning & shadow.
Then what value can you offer to the AI?
the AI can generate well enough without your intervention, your help is not needed
You can't just write a Prompt and call it a Movie and a Game.
AI art is an Incomplete Art.
one single Art can't be sold, and unuseable.
if someone use AI art in a real Production, i bet it will become a meme and ridiculed in no time.
because if you zoom on it, it's just a bunch of incoherent gibberish.
Yeah, saying something in your own words and then teaching someone else in your own way is really one of the most valid ways to confirm you’ve learned something. But beyond that, writing is like talking, it’s much more organized, thoughtful and articulate, but it is really just an extension of talking. Talking is insanely important for conscious beings including humans. We are extremely social and must constantly be advertising our conscious state to others. More than that, talking is our conscience. We are always talking to ourselves. Our ability to talk is how we understand ourselves. It shapes our thoughts and how we express our self to ourselves… This is the pitch for learning how to write. If you can’t express yourself in an articulate way, you can’t think deeper, communicate better or convince anyone else that you or your ideas are meaningful. It’s worth the effort to learn to write, and I hope that this doesn’t take that away and leave us essentially mute in the sense that we don’t have access to language that accurately conveys what and who we are to ourselves or anyone else.
I have always loved your videos. It feels almost weird to actually learn from an internet video, but you never disappoint in doing so. When I found out you had a book out I immediately opened Amazon and bought it. Thank you for these amazing videos! Hoping your book is just as good :)
As a highschool AP teacher, I was annoyed when I had to start checking everything for plagiarism. This emergence of machine writing has made me choose to leave teaching (among other things). Impressed but not surprised that you got straight to the heart of the issue to society. You always do though.
lmao that is such a weird takeaway
@@PinkMonkeyBird I didn't share a take away... You seem confused.
@@oliviabasham Weird takeaway to leave teaching because of emergent technology. Hope that cleared up your confusion
I’ve used it for 2 weeks now and agree it always reads like a high school essay. It doesn’t hold a candle to a real writer (yet anyway)
It arranges information cleverly but no original thoughts
I think it's important to remember that this is just a 'feeling lucky' feature on Google. All the information is already out there and there are already services that can help with homework or editing text professionally. Instead of worrying about this, educational institutions should be encouraging self-learning and providing students with the proper resources and learning outlines. With the right outline, students can research and learn the course on their own. Schools only provide a course outline divided into chapters and subchapters, but the real learning happens when students take the initiative to explore and understand the information for themselves
Plot twist, this entire video's script was written by ChatGPT
Double plot twist. This videos transcription and CZcams comments are used as a data sets
Our existence was written by something like ChatGPT
2:25 - I asked that question in English class over 15 years ago. I later learned the reason is the same as in math. Well, and the same as in gym. No, you may never have to bench press a tree that had fallen on you. But doing the exercise is good for your body, and will help you do other difficult tasks since they are easier than what you've practiced.
Great video. I’d also add that a lot of the contextual information pulled and used in the output is from blind sources which we cannot verify the truthfulness of.
This is currently its biggest deficiency. If they could deploy an algorithm that could prioritise authoritative sources then it would improve the quality of its output massively.
What music is playing at 4:00 and where did you get it? It inspired me to want to write an essay of my own about how single clarinet accompanied with piano always brings a to mind a feeling of intelligence and discovery.
It's Mixed Emotions by Arthur Benson; it was credited in the description.
Lol, it stood out immediately to me too
Also commented it, then realized it's in the description
And this video is written by chatgpt. 💀
I struggle with focus and attention. I some level of dyslexic too. Writing has for long been my weakness and I failed school because of it. I’ve been using Chat GPT lately and being an “Editor” has been the best help for me. I still appreciate anyone writing from their mind, but having a Chatbot begin for me, then Edit things later is really helpful for me at least. i can see this tool help a lot of others that struggle with focusing.
this is a great use-case!!! i talk about this in my video i uploaded yesterday, would love to hear your opinion markus!
Nerdwriter: ...without compunction.
ChatGPT: *raises an eyebrow*
This is also the way I feel about ai art, it just lacks the context about things
Not necessarily and as it gets smarter and more advanced training more creativity will come about not that its already creative!
@HardKore5250 Yeah, one day maybe, but it is not conscious now haha, don't underestimate human intelligence. There's nuance about these things. Like I can tell you the solution to the formulas, but you got to figure that out yourself, you know? Like I can be an expert in math but you won't get my probable years of expertise by me just telling you, you know? What informs a artist is their entire human experience and mind at hand. Of course some people aren't very creative and go "visual candy" and copy things like they are ai themselves. Art is very hard to master, it's a study of life, of everything visual and of storytelling. Try to take that in and the fact that we don't know everything about the brain either, art is a very physical thing too, you can see it in each stroke how the artist felt and saw things, but with ai art it still feels uncanny valley and patchworky, like I can feel it has a mental issue
It exists, it should be incorporated into the learning process. Heck maybe the assignment is to take a promt and figure out what's wrong with it, or make a rebuttle(I assume that's harder to generate). Write a paper on how you think the bot picked what it did.
There has to be a creative way to adapt to the fact it exists.
My last few college essays have been written by GPT and then edited by myself for half an hour or so afterwards
That was an amazing analogy to mathematics, computations outsourced to the calculator, soon enough conversation outsourced to the communicator.
Expand this text into a essay: ''His mother's addiction eventually led to her death when Franklin was a little boy''
Grand Theft Auto 5, or GTA 5, is a highly popular and critically acclaimed video game that has sparked widespread debate and controversy over its depiction of violence and criminal activity. At the heart of the game is the character Franklin, a young man who finds himself caught up in the criminal underworld of the fictional city of Los Santos.
One aspect of Franklin's character that is explored in GTA 5 is the impact of his mother's addiction on his life. As the game reveals, Franklin's mother's addiction eventually led to her death when he was a young boy, leaving him to navigate the world without the guidance and support of a parent. This tragic event serves as a poignant reminder of the destructive power of addiction and the ways in which it can ripple outward, affecting not only the individual struggling with it but also those around them.
As a philosopher, one could delve into the underlying themes and social commentary of GTA 5 by examining the portrayal of addiction in the game and its broader implications. On a deeper level, the mother's addiction and its consequences can be seen as a commentary on the ways in which societal issues, such as poverty and lack of access to resources, can contribute to the prevalence and severity of addiction. It can also be seen as a commentary on the destructive power of addiction and the ways in which it can consume and ultimately destroy an individual.
Additionally, the depiction of Franklin's mother's addiction in GTA 5 can be seen as a commentary on the ways in which society often stigmatizes and marginalizes those struggling with addiction, rather than providing them with the support and resources they need to overcome it. This can be seen in the game through the portrayal of Franklin's mother as a deeply flawed and tragic figure, rather than a complex and fully realized character with agency and dignity.
Overall, the portrayal of Franklin's mother's addiction in GTA 5 serves as a thought-provoking and poignant example of the complexity and gravity of this issue, one that requires careful examination and contemplation in order to truly understand and address its effects.
I'm interested to see how universities will deal with this technology in the future. I have just used ChatGTP to write a 5000 word essay. It only took a handful of creatively worded prompts, and minimal editing once everything I needed was generated. Ran it through a plagiarism checker, and it came out as 100% original. I am expecting to graduate with a first, even without getting full marks on this essay, so it doesn't matter too much. But at least now I don't have to waste my precious time doing something I hate, and can focus more on my main project in the lead up to my final hand in. Nearing the end of my degree, I've long since become disillusioned with higher education and see the whole system as a scam.
How were you able to make it write a 5000 word essay?
@@infinitebrian9178
First I got it to generate a numbered outline for each section of the essay. And then worked on each individual section one by one. You have to learn the basics of "prompt engineering" and then you can generate anything. I was even able to generate Harvard referencing.
I’m a neuroscientist by education and my biggest worry is that, is using bots going to become addictive just as social media did. Just think about it: Button pressed -> Job done -> instant gratification. PING! The same dopamine loop as social media notifications. How can we prevent this convenience and addiction?
3:16 the picture on the right was taken in Barcelona. Passeig de Gracia to be specific.
This is a really nice defense of the value of writing for young people. I'll just add my usual comment to the view that early life is a process of self-discovery, as is nicely presented at 5:20: it is also a process of self-creation. We don't just discover ourselves in writing; we, at least in part, make ourselves as well.
I'm the only person who doesn't used chat gbt in my school
What if it helped you gain a new understanding of something as you're writing about it? How would you feel about it then? Because i haven't been in a classroom for over a decade but i have never stopped learning, and for me, ChatGPT is an amazing learning tool. it's a tutor with infinite patience for all of my most detailed and ridiculous questions. It can be a Socrat and help me understand my own beliefs better. Like any tool it has uses that are both good and bad. Pocket calculators may have made us unable to do long division in our heads but hey at least we invented microwaves right?