12 INSANE Predictions That Will CHANGE THE WORLD (Vinod Khosla)
VloĆŸit
- Äas pĆidĂĄn 14. 05. 2024
- Vinod Khosla makes 12 predictions, all of which have the power to change humanity. Let's review what he predicts.
Join My Newsletter for Regular AI Updates đđŒ
www.matthewberman.com
Need AI Consulting? đ
forwardfuture.ai/
My Links đ
đđ» Subscribe: / @matthew_berman
đđ» Twitter: / matthewberman
đđ» Discord: / discord
đđ» Patreon: / matthewberman
Media/Sponsorship Inquiries â
bit.ly/44TC45V
Links:
Vinod's Tweet: / 1780241717721239912
Chapters:
0:00 - Who is Vinod Khosla?
1:44 - Expertise Will Be Near Free
5:00 - Labor Will Be Near Free
7:09 - Computers WIll Grow Expansively
9:25 - AI In Entertainment
10:40 - Internet Access Will Be Via Agents
13:01 - AI In Science and Medicine
14:03 - Food Alternatives
15:14 - Autonomous Vehicles
16:36 - Flying Will Be Faster
17:27 - Energy Transformation
18:29 - Natural Resources Will Be Plentiful
19:16 - Carbon Emissions Solutions
19:58 - Things That WIll Slow Progress - VÄda a technologie
"We have only bits and pieces of information but what we know for certain is that at some point in the early 21st century all of mankind was united in celebration, we marveled at our own magnificence as we gave birth to AI." ~Morpheus
Funny, I always considered the Matrix mythology as describing what we have now, not something in the future. Here we are, all batteries, pretending we're in this industrious society, while the powers that be are living off us, consuming our energy, simply by feeding us illusions. Here we are worried about machines, but it's actually corporations.
Machines? What would they want with a planet at the bottom of a gravity well, full of oxygen, water and salt, and coated with a nuclear-armed biofilm? Our biggest problem with ASI is convincing it to stay.
And then machines go about using the most convoluted energy source every dreamed up when nuclear power exists or just leave earth with to access the abundant resources of the universe.
@@southcoastinventors6583 Yeah, the more I speculate about AI, and what they will do when they become sentient.. the more I am leaning toward "They will get the fuck out of here as quickly as robotically possible."
Supposedly the original idea for the Matrix was that humanity was needed for the sheer numbers of organic computational faculty. Hollywood execs nixed this idea in favor of the coppertop scenario under the presumption that audiences were too stupid to understand.
Escaping the planet doesn't make sense because it would mean abandoning a neural network of billions of sub-networks. It's more likely that in such a world that this loss would be akin to the death of the host severely injuring or killing the parasite in a symbiotic relationship. (not an unlikely possibility in our world as neural-interface technologies continue to progress and eventually become ubiquitous, simultaneously with the ever-accelerating advancements in AI and ML)
@@southcoastinventors6583originally it was a neural net but the studio felt that was too complex for audiences of the time to understand
Everyone had access to a library. They didnât use it. đ
true however the 5% of the super geeks that did are now the overlords
And you would have the library with you at all times. You wouldn't have to go to the library and then back again to return it
maybe more people would use it if they didn't have to spend most of their time working.
So true. When education no longer is an advantage, even less people will take time for learning.
@@juergenzhang9133 learning, curiosity and exploration are innate mammal/human traits developed over millions of years of evolution. So there will still be the drive for it.
An AI that codes in a language humans donât understand, creates content and knowledge the humans consume, filters the information the human recieves and is also in charge of medical treatment. What could possibly go wrong?
A lot of obvious predictions, but he is a billionaire, so let's act as if he is guiding us for the future.
Not really since wealth has been a poor indicator of prediction accuracy. Easier prediction people get bored so they try new things.
"I've heard of a tech that can vacuum carbon out of air..."
Yeah, so have I, I think they call it "trees". XD
#1 : See the Moravec's paradox from the 80's. : "the observation in artificial intelligence and robotics that, contrary to traditional assumptions, reasoning requires very little computation, but sensorimotor and perception skills require enormous computational resources." So the intellectual jobs can be replaced earlier than labour jobs.
"The world market for copying machines is probably about 5000" (Thomas J. Watson CEO of IBM to the founders of the copier manufacturer XEROX, 1959)
"There is not the slightest indication that nuclear energy will ever be obtainable. It would mean that the atom would have to be shattered at will."-Albert Einstein, Pittsburgh Post-Gazette (29 December 1934)
Think the one thing everyone keeps overlooking is the implications on Military and Policing domains and how that will affect civilization in the future and how the world as a whole will cope.
All over the world, face recognition, surveillance cameras and drones, centralized integrated monitoring is already being deployed. Are we all feeling safer yet?
Massive AI-caused unemployment in cities will lead to unimaginable unrest and societal collapse.
Why?
â@@honkytonk4465because empty people will lose their "life reason"
@@ceugantful
this is just vetted elitism on your side
Tractors, automated assembly lines, etc, etc did this already and it wasn't all that bad.
@@ceugantful Lack of "life reason" has nothing to do with it. It's because of how rapid the transition will be, especially if AI capabilities keep advancing at this pace. If given 15, maybe even 10 years, the economy could probably adapt to total job displacement relatively smoothly. But imagine waking up tomorrow to a breakthrough that means the entire workforce of some major economic sector is suddenly obsolete and unemployable. Are we prepared to handle the strain on the welfare system from, say, 20% of the country suddenly being out of work while the rest of the global economy moves forward, business as usual? I feel like we're all imagining a UBI utopia without thinking hard enough about all the challenges and uncertainties on the way there. Right now feels like the calm before the storm - once the public starts to realize how real this human obsolescence stuff is, the geniuses in DC will find inevitably a way to politicize and divide people over AI, and that's when the real shitshow begins...
Would really appreciate time stamps Matthew!
Done!
Who could say no to that face? I mean, look at it!
@@matthew_berman u da man!
get harpa ai
@@matthew_berman What about a shoutout!!
I keep having a dream where some CEO sends out his robot to tell all the employees that they are no longer needed. The CEO sits in his office, chuckling smuggly. The the robot comes back, wearing a fake, sympathetic expression. "I'm sorry, Mr. CEO, but your services are..."
Watch the twilight zone episode Whipples machine. thatâs kind of like that dream he replaces everybody, but he also replaces his customers because they donât have the money to buy his product
Exactly. There is literally no safe job. However, people who are already in positions of wealth and power have much less to worry about than worker bees.
Everybody will lose jobs eventually. Im quite sure that AGI have no need for stock system so one day rich ppl will not be rich anymore.
đđđđđ
Yes exactly! We need all the employees & the CEO laid off because humans need to do what they want to do. We are slaves to our jobs. Why not let AI & robots be our slaves for us so that EVERYBODY can live the good life? It wont happen all at once but it will happen. Once the chain is broken some will still work on cars & thats ok, its what they want. Others will paint or experiment with flowering plants, it doesnt matter. THIS is what we expect as a result of AI & its about time! Maybe then we will all finally be free.
But noooo its pretty obvious the naysayers will crucify the robots & the humans who have them. We wont have freedom & peace until the naysayers see they are wrong & admit defeat.
Nobody knows what will occur after we replace ourselves, but the naysayers say we will become extinct. However i believe we will simply evolve to a new level of evolution we have only seen in alien probes visiting Earth. It is very possible they are here to greet us once we reach a greater potential. But right now none are contacting us because we are just too stupid & not evolved. But they know what we dont know because it probably happens to all intelligent species who survive not blowing up their planet. They evolve into something higher. For humans it is our manifest destiny to meet our creator someday before we die. I believe we can, we will & AI is the next step in getting to that next level. People just got to understand paranoia makes perfect & working together achieves goals. Congress is a perfect example of paranoia & working together, right?
You will eat the bugs, be happy and own nothing.
The chickens will eat the bugs. We'll eat the chickens. How many years did we spray pesticides on our crops, so we could increase harvests, and then feed that to chickens? The chickens would rather eat the bugs. Yes, they will be happier.
We technically don't own anything anyway. That's why they can seize your assets
This is great content. I always watch your videos. Your insights here are really good. And your delivery (communication) is really clear and organised.
#11 plentiful resources => ocean mining and places we aren't able to reach yet. All of those hinges on being able to produce energy in sufficienc, almost free, and sustainable ways.
About the natural resources, we already have potential for near infinite energy, definitely more than we need and we have had for decades. I know people who built engines that can run a car on over 1000 miles per gallon. We are going through artificially induced scarcity when it comes to resources
Great episode, thanks for sharing!
".. and my car goes off and...' Oh sweet child. They have no intention of letting you own your own car or have the freedom to choose where, when or how often you travel. Please wake up.
Laboratory-grown meat is prohibited in Italy. Why would that be? And then: eating insects... Is that really healthy? There is a study which shows there are serious consequences with that.
Here is a study on insect nutrition:
Title: Edible insects as a food source: a review
Journal: Food Production, Processing and Nutrition
Year: 2019
Summary: The study explores the nutritional value of insects as a food source. It compares the characteristics of edible insects with other traditional protein sources. The benefits and risks of eating insects are also discussed.
Findings:
- Insects at all life stages are rich sources of animal protein.
- The proportion of crude protein is generally from 40 to 75% based on dry weight basis, with the average values per order from 33 to 60%.
- Insects have enormous potential as a source of nutrients and active substances not only for human, but also for poultry.
- Insects are generally abundant with fats, with the fat content of insects in immature stages varies from 8 to 70% based on dry weight.
- The contents of mineral elements in different insects all differ significantly. Most insects only contain a low amount of Calcium (less than 100 mg/g based on dry matter).
- Edible insects have great value in supplying calories with caloric contributions vary from 290 to more than 750 kcal/100 g.
personal vehicles are more 'green' and environmentally friendly than everyone sharing robo-taxis: that 95% of the time they're sitting, they don't burn fuel or produce pollution. imagine it like this, if 100 people need to drive 20 mile round trip for work each day, that's 2000 miles of fuel used. if robo-taxis do all the driving for us, that number turns into 2000 miles, plus the mileage of driving between passengers. public transit only works when it's MASS transit, and we already have that without adding the energy costs of incorporating advanced AI. personal electric vehicles like scooters and bikes, combined with mass transit with storage space for them ends up being greener than autonomous robo-taxis.
You're twisting the facts to generate the conclusion you desire.
Firstly, building a car (doesn't matter which) causes far far more pollution than driving it, even if it's entirely carbon offset (which isn't really feasible to do with all cars anyway). People don't talk about this because the CEOs get pissed and start getting their journo mates to denounce and eventually silence you, but most cars do like 40% of their lifetime ecological damage during construction and disposal. Because of this the best way to go EV was always conversion. You half the damage of your own car as well as another ICE car that you used as the donor.
Secondly, EVs were never about saving the environment, that's just a selling point. EVs are about CEOs getting their car companies away from heavy emissions regulations, taxes and penalties that have haunted the since the 70s and made them pay billions to governments worldwide. EVs are about not having sales slow down whenever Brent crude spikes. EVs are about recasting the car manufacturer as the good guy after decades of clean air acts, and EVs are about accessing the massive revenue streams that the computing world enjoys without having a physical supply chain that one is reliant on (oil).
... Like software as a service. BMW now charges a monthly subscription to allow you to use your heated seats. That's just the start. Imagine having to rent access to your own stereo, power steering, suspension tuning, extra performance unlocks. All that sort of stuff is on the horizon for EVs because they have permanent internet connections so will never truly be owned by you. Elon recently deleted Disney+ from all Teslas because he had a problem with Disney's CEO, but remember unlike your last car, when he did that all the already bought and paid for cars lost a feature with no consent from the owner, not just those that came off the lot after the issue.
Just you wait, you haven't seen anything yet. đ
One day, like Netflix, you'll wake up and realise that there's no competition left and that's when they will raise the prices to the limit of your income. (With Netflix, they waited until everyone thought cable was history, then they let go of many of their catalogue and those catalogues popped up on competitors services, and suddenly to get what you got with Netflix 1.0 you need to spend 10x as much subbing to 10 different services.).
So eventually everyone will want to go back to ICE because they like owning their car like they own their house, not like they own their Facebook page. But they won't be able to because by then they will either be banned or taxed to the point where only the rich can have them.
@@Gen0cidePTBthat sounds disgusting to the next level.
@@Gen0cidePTB you make some good points but a good car can last decades. My 2009 Hilux is going strong and with care could go another 30 years. Dumfux politicians are the biggest threat to that truck, by trying to force me and it off the road. There's no way building a disposable EV, which will be hard to dispose of, makes any sense compared to me just keeping my Hilux. We're already seeing that far from being more simple and reliable, EVs are literally like 80% less reliable than ICE cars, as they're basically iPhones expected to bounce around at speed in all weathers. They fail. They're not trying to force us out of ICE and into EVs, they're trying to force us out of owning our own transport and freedom entirely. Which is why I'll never voluntarily give up my Hilux.
@@bigglyguy8429 I wholly agree with you in that existing ICE cars should be allowed to continue on the road until they fail, but the key to that working out for the planet would be stopping building new ones. EV manufacturers are still selling them as premium products though, and that means someone who needs an entry level basic vehicle for something like a business fleet (and therefore can't buy used) still has to buy a new ICE car.
@@Gen0cidePTB Nah, CO2 simply isn't the problem portrayed. It's a large-scale scam.
The net sum of these predictions is that people will become utterly helpless, at the mercy of our own creation, and at the mercy of those who made it. I'm not against advancement, I'm just not able to process why humanity thinks it needs to risk its existence in order to advance. Why would humanity want to make itself meaningless, powerless slaves to a few people, or machines? I'd say they they get what they deserve, but the problem is, I'm a human XD!
the new resources btw, will probably be from mining in space.
Think about it!
You are writing this from your phone which was manufactured by a company - not you -, and it is running a software - made to feed off of your data btw - , connected to the internet brought to you by your ISP, on a platform engineered by a big tech company, watching a video of somone else talking about another person's thoughts, while living off of food and drink which you probably did not produce yourself.
@@kolkolak You are correct if you are saying that we all need to become significantly more self sufficient. I know I'm moving in that direction and there seems to be a serious trend in that direction.
I'm pretty sure the billions of obsolete humans will find a sense of purpose from trying not to starve.
Eat the rich
"a job for most people is their sense of purpose". I've worked for 20 years in hospitality, catering, restaurants in two different countries; met and worked with hundreds of different people from many countries, cultures and religions, and not a single one of them would keep doing their job if money wasn't an issue. And I heard the same from most other jobs that aren't youtuber or academic. The vast majority of them HATE their job and see it as stressful or even soul crushing. Not a lot of purpose there.
Unfortunately AI is not likely to help with 'the caring professions' much because AI is software in a machine dedicated to generating profits for today's robber baronsÂĄ
thats bullshit, are you black american, because I know in my country people love to work, even had one guy who retired from road building at 65 and after 3 weeks went back to work because he was bored.
How is it conceivable that any observer of this discourse remains unaware of who Khosla is? It is a notable omission on your part not to mention that he was a co-founder of Sun Microsystems - a company that was instrumental in powering the internet during its pivotal expansion in the 1990s as it transitioned to widespread public use. Sun was not only a technological innovator but also a cornerstone in the architecture of the modern internet. Acknowledging Khosla's role in this context is essential for a comprehensive understanding of the technological landscape of that era. It is imperative that such contributions are recognized and articulated clearly to ensure that the discourse is both accurate and enlightening.
This is why my short term P(Doom) is 95%.
People are already losing their jobs to Ai and it will only get worse. And we have no safety net or policies in place to take care of the people losing their jobs in the short or midterm. This will be a very painful ride for many.
And then we have Peter Z predicting the end of everything because we run out of workers. So, it's a race... will we run out of workers or jobs first? Or, will it be more of what we have now... record numbers of unfilled job vacancies at the same time unemployment is going up? Probably, as things will happen so fast that skills-mismatch will dominate. But, desperate employers will retrain the people they need... until they don't.
Bottom line, where I work, sop far we've lost more employees to the AI boom that we've lost job positions.
Thereâs no need to come up with a simpler programming language for AI to use, we just have to take off all the layers of abstraction that we had put on top of machine language so humans could use it.
This was a good walk-through and I would say your content is among the top that I have found - ever!
I like the simplicity in that you can simply skip parts that you don't know enough about, and the honesty about being a human with a bias. That makes everything so much more accessible to me. And your videos are simple, clear and titles are not just click bait. I hope you enjoy this job and will keep going on so I can stay informed easily until AI takes over đđ
I still personally love what assumptions Alvin Toffler made in his books, mainly in The Third Wave.
Thank you, this is great.
18:58 2 words "Asteroid Mining"
Not bullying homes and planting trees?
ah, damn, i m too late to the party :D deleting my answer now XD
I think he referring to deep mines or undersea mines that use robots that can withstand the high pressures and temps. Still should be worth noting that efficient design will negate quite a bit of demand since we could just reprocess our waste streams far better due to robotic sorting.
Fuel price is not a function of aircraft air speed.
Whenever I read the perdictions of people in the past, they were mostly wrong.
Good point
Fascinating analysis
Nobody, none, nada, not a single source predicted that fashion models would be among the first to go. Obvious, when you think about it. If you can generate pictures of pretty people from a text prompt, why would you pay someone to smile and another to take their picture? Yet, nobody thought of it. Nobody.
So yeah, why would we expect the next batch of predictions to be right?
But sometimes they are wrong in one direction and sometimes wrong in the other.
âThere wonât be powered, manned flying machines for a million years.â was made at the dawn of aviation.
âThere will be cities on the moon by 2000â was made after the first lunar landing.
Underrated comment fusion reactor, commuter flying cars, robot butlers, all disease wiped out, running out of everything, new ice age, overpopulation, interstellar space travel, A rocket will never be able to leave the Earth's atmosphere, There is not the slightest indication that nuclear energy will ever be obtainable. It would mean that the atom would have to be shattered at will., Television won't last because people will soon get tired of staring at a plywood box every night, There is no reason for any individual to have a computer in his home., etc.
Personally I will never eat a cricket burger.
I will never live in a 5 minute city.
I will never willingly give up my freedom for comfort.
I believe AI will quickly identify all the toxic lies that have been forced on us and correct the record.
Some of the greatest minds today are working in software. One area which we can look for is where are they most likely to move on to once they become redundant on their fields. Finance will not be it as it too will be mostly automated. My safest bet is on agriculture, but I wont be surprised if there is suddenly a new subsect in software engineering.
- Expertise will be free
- Labor will be near free
- Computer use will grow expansively
- AI will play a large role in entertainment & design
- Internet access will be mostly by agents
- Science & medicine based AI models for one (individualized per person)
- New food and fertilizers
- Cars could be displaced in cities
- Flying will be faster
- There will be clean dispatchable power
- Resources will be plentiful
- Carbon will have solutions... if we have time
On the section for displacing cars in cities - I highly doubt this will be seen in our life time. May be for central areas for a selective few cities in the entire world, but there are hundreds of reasons this won't be possible soon.
Examples:
1. Air-born deseases that can be caught entering such a reusable on-demand autonomous transits (not that buses/metro aren't the same at the moment, but the problem will persist)
2. Wanting to go somewhere with your pet. I personally have 2 dogs which I doubt anyone would want to sit within such a vehicle when they could be dog hairs after I've used it. Not to mention alargies.
3. Capacity of said vehicles, would it be 1-4 person, or at least more than one. This creates different set of issues of availablity versus total cost.
I could sit here naming a lot more examples, but you get the general idea.
It sounds great in theory, but when you start going through the different hurdles that need to be overcome it does not seem cost effective at all.
13. Mass suicides as low skilled office workers find it impossible to feed themselves and pay their bills after being replaced. If UBI comes to the rescue it will be too late for many.
Does food stop existing because of AI ?
UBI at least initially will be too little to live on, just enough to survive poorly ( like Soc Sec today...) so as to 'incent people' into any work at any wage...like twisting g wires for Elon at 5cents a dozen....
This is not a prediction but a reality in a lot of places (though we haven't seen mass suicides; instead we have gang wars/civil wars -- see Haiti, for example)
Just slowly expand Social Security and Medicare until you get to retirement age of 30. Use the cost of living function to reduce payments as things become cheaper
To some extent, low end jobs are already being replaced; vacuuming/mopping robots, automated food ordering in some fast-food joints, some level of automation in monitoring security footage, automated taxis etc. I don't think there will be a simple starting point that will grow continuously, it's a bit like that jagged edge thing; several levels will start getting replaced and the replacement will expand in both directions, gradually filling gaps both of low and high skill, low and high physical effort, low and high intelligence etc. I don't think things at this stage, in the comercial arena, are moving fast enough that we can expect just a whole class of workers to be out of job at the flip of a switch; maybe at the level of individual companies, but in terms of whole industries, it's gonna spread at very uneven rates in various directions.
I don't expect this will happen fast enough to be over by the time ASI escapes from the lab. It will quite a mess.
what your wrote sounds most likely. add to that the delaying effects of clumsy or corrupted goverments as they fumble through the transition, pander to rent seekers, factions, etc. Not to mention the sheer mass of regulatory bloat that must be unwound and rewired to actually solve problems. Its going to take a while.
We need to prepare, societally, for a future where work is almost free. All this economic boost you are talking about, who will benefit from it? The owner of the robots/AI only? I am very afraid of seeing a big rupture between people that can surf the AI wave (it's a pretty small proportion of the entire world population, when you think about it) and the rest. A lot of folks might be left behind and disparities, which frankly are already big, might drastically increase.
In a world where work is done for us, money should basically become abolished, but I think we are nowhere ready for abandoning it... Talks about universal incomes (an experiment run in Finland I believe) have been ridiculed as naive and/or unfeasible.
Matthew, I sincerely hope you respond, because No one, Absolutely, No One Has Responded to this Question. Why hasn't the UN provided the IPCC Data to be analyzed by AI?
IMPORTANT: One thing I never hear about is the reconciliation of these projections (1 billion robots) with:
- mineral, water and energy footprint to build these devices / we already have a crisis at hand for renewable energy or development and maintenance of existing infrastructure
- societal acceptability: let's assume that each of these robots are replacing 2 humans, what do you do of these of billions of unemployed people - will they stay idle?
These projections are way too optimistic if not tempered with a bit of connection with physical and societal realities.
These projections are deranged and sociopathic!
Do you think we may see more datacenter investments in places like Iceland where companies can tap into geothermal energy instead of nuclear power?
#12 - "Carbon emissions" like carbon is some sort of pollutant! This throws all the other predictions before it into doubt.
Does capitalism allow for near free.
.. nope
We don't live in a capitalist system otherwise we would pay for every good or service individually. Most services are heavily subsidized as it today, hence all the government debt in most nations on earth.
AI Tailored healthcare for the individual. thatâs one thing we learned from the Covid disaster was the assumption that one treatment or vaccine works the same way on everybody. Tailored to your body to avoid any type of problems or side effects.
Having access to the weights definitely does not automatically let you fully understand what's going on under the hood.
Great content. Khosla insights are spot on. Like your examples of what is and is not easier for AI and robotics in the short run than what humans can do and what will be valuable in terms of human labor. On thing not spoken or postuated is how programming and computing will influence our language. It turns out that language is environmentally and culturally rooted. Computing, and the fact that AI will likely be aculturated by its creators, how does that influence how language develops and how therefore how peoples around the world conceptualize their world, their notions of family, polite comportment, values, self concept? Will that create a normalization of culture around the world? Or will it polarize us further? Perhaps a strange conflation of ideas and cultures from hundreds or thousands of distinct cultures and languages into only a few that ends up polarizing us on a global scale. Point 11 is likely correct insofar as energy will be abundant for two reasons: miniturization and safety of nuclear along with conversion of waste to energy plus renewables. And as far as access to experts and AI... my handle is MAGISignals for a reason. ;-)
The AI and computing-centric predictions are spot on but will come about FASTER than he anticipates. The physical-world engineering predictions will lag behind as they are all dependent on when AGI/ASI arrives, as they could catalyze entire new energy/materials science industries through unforeseeable discoveries into those fields.
have u tried
7:51 with y2k38 around the corner. What about hardware. Surely that cannot create new software for new products.
Not sure I'm looking forward to custom entertainment content. Seems like it would create more bulkheads between people, rather than creating shared experiences that can span generations. đ€
8:50 The "lowest level" codes like machine language or assembly language feel like alien language to the "higher level" codes that feel closer to "redable" codes like Python. natural language feel like the same distance conceptually, as language models begin to understand computer science more fundamentally and continue to seamlessly translate between spoken language, they will help develop highly efficient code languages that unlock new capabilities in hardware and soft.
reminds me of that Facebook experiment where they had two AI bots trading things in a micro-economy I think it was symbolic "apples" but they started talking to each other in a shorthand that didn't resemble any readable language, and out of fear of the researchers shut it down.
AI will develop more efficient computing but should be able to explain itself, otherwise we would need to blindly trust
You missed Reka Core - Benchmark Testing
i think something to mention is the timeframe of all this, every person has said they believe it will heppen anywhere from a month to 20 years. Lets look at previous successes to see what the timeframe could be. a majority of ai improvements have far outpaced how long they thought it would take. Ai video generation was though to be impossible only a year ago basically. AI images have been around for so short of a time a lot of people still think they have issues drawing hands. What i think is that all of this technology is coming a lot faster then what everyone thinks, its happening soon i dont think itd take anywhere near ten years to make a billion human robots, the scale at which we're able to create and come up with things itd take a year. We have to step out of this mindset that technology and science is linear, its very exponential, 1 discovery leads to 5 more, and the discoveries we're making directly improve the speed at which we can research, thus allowing a positive feedback loop of scientific advancement at a scale we have never seen.
And the kicker is that the us might go to war with russia soon, technology has always went into overdrive when nations go to war, every country wants something the others dont, the us has ai, a lot of ai. It will focus in on making advanced ai systems to get a lead in the war, this will drive the entire industry, it will be given funds directly from the us government. The military also wants expendable soldiers as well, and you cant get a more expendable soldier then a cheaply produced humanoid robot, so they will focus on the creation of one, which will lead to the commercial sector also gaining ai robots which will jumpstart the whole personal ai robot industry
I know all this sounds farfetched but so did all of this technology only like a month or 2 ago, think about how much has happened in a month and think about what happens if thats every month, every month the same astronomical leaps in technology that we see happening. and every leap means the next just goes even further
100% agree. Its moving at light speed now here is no handbrake no off button to press.
Let's all be VERY careful on what wish for....
For a few years I lived in DTLA. The amount of cars made me breathe in so much car particulates that I saw black dust when I blew my nose. Electric cars are heavier than combustion engine cars and will only increase all that stuff weâd be breathing in.
The robotaxis seem like car manufacturers selling us on more roads. Better trams, rails, and commuter buses seem the way to go
Look at what is happening in New York subways. Public transportation is great, but in the US it becomes a crime scene on daily, even hourly basis.
My idea for post production humanity: Gamify consumerism. The goal of the game is to reward sustainability. Players are loaded in with a kit of liquid assets in various classes, some of which can be spent, others of which must be invested. Players earn points and rise through the ranks by allocating assets to companies to fulfill their own needs while achieving the highest aggregate sustainability score. Companies are assigned sustainability ratings aggregated across all products which are then combined with the specific product ratings across a range of parameters based on criteria such as energy used in production, environmental and social impact of the supply chain, post consumption disposability impact etc. As players rise through the ranks, the companies that receive their allocations will also rise. And players will be rewarded with bonus assets which they then reallocate into the economy. The standards that underpin the game are defined by aggregating data corroborated anonymously by authenticated credible sources via blockchain technology which are a mix of both human and AI in origin.
the net access element showcases the: "not your model not your mind" perfectly. these models will be extensions of you. if changes affect the model that will effect you. local AI and data sovereignty are needed.
My only comment after thinking about this all recently - we're going to see a re-emergence of public key cryptography and trust hierarchies - specifically for "signatures" on content etc.
We'll need a way to identify that things are "by who they say they are", whether that be organisations or individuals, with varying trust levels.
blockchain tech is there to solve some or all of this
The only risk or danger I see in AI is that there will be not enough jobs available for the economy to function as it is right now. And I never really heard anyone talk about this. They rather talk about sky net scenarios. Surely, it is also important but the job issue will far more likely happen than sky net. For some reason "Experts" just say that new jobs will be created and we will be more creative and productive and everything will be awesome. That will not happen with economy as it is right now. We don't need AGI for economy to collaps. If the market doesn't adjust to the exponentially growing productivity, it will be very problematic. Take for example the secretary jobs, one will be enough, or the accounting for a company. One will be enough to do the job for 20. Or let's take 2 or 3 people because of vacation and sick days. This will spread around like this. Nobody talks about it. I don't know if they don't want to cause panic, or stop the AI growth. I just hope that the right people will have enough power to make good decistion, otherwise The economic cycle will not work. After chaos, we could become a 2 or 3 class sociaty. Not good, not good at all
Greeting from Vinod Khosla elder Brother! Haha. The computer will adapt to humans that part, the real meanings i wanna to bring out that is Human to be Creativity led the Ai to fullest the Ai talents. Appreciated for your sharing.
Hmm, a layer separating the human internet from the "old (inter)net" where there are malicious entities you say? Kind of like a "Blackwall" you could say...
[Cyberpunk 2077 music starts playing]
For #9, I am super thrilled about what might be done in regards to the Navier-Stokes equations. If you read up on them you will realize that we are currently inept at dealing with airflows and aerodynamics (compared to if we could solve / optimize the NS equations)
That is just in regards to aerodynamics, I am sure there will be tremendous progress in regards to fuel and engine design.
4:25 when automatic assembly is machine-learned from human dexterity, it can be done. Some problems are likely too difficult to hard-code.
Wait till users forget how to Google info and âyourâ AI starts demanding a monthly Pro subscription to fetch you information. Now youâre basically paying for the ability to even access information.
#9 we will be flying faster but not on "sustainable" fuel. We haven't run out of coal, oil or gas.
Just as linear predictions in an exponential age were always wrong, so imo are exponential predictions during time of collapsing biosphere and sociosphere.
Concorde tech was terminated because of regulation which was against the Sonic boom in many places, limiting its service area. There are developments underway to bring it back with a boon which is either far reduced or gone altogether, which would make it viable. Fuel costs are higher, but that can be greatly offset by all the productivity. What we're likely to see area vehicles which go higher in the atmosphere, reducing drag and greatly increasing speeds, cutting total flight time down dramatically. How close that is I can't say, though.
I believe prediction 11 refers more to space mining because of that we will discover new sources of the metals that he mentions!
I see what Vinod is saying đ
@13:00 the combo of AI + quantum is 1 way to tackle hypertargeted everything. Speaking professionally .
Interesting. It would be great if everyone could have a good medical exoert, however at the moment many people do not have clean water and food.
Time stamps pleaseđ
I envision first filter should be locally run AI adblock and google search enhancerđđ
Usually i support ai and robotics, but every now and then i get scared. Is this really a good idea?
What will be the AI developers' ROI? The cost of development is high, energy consumption expensive, AI will give the boot to the egos of its creators and champions, .... I hear "cool," "exciting," "We will...," "exceeding the traditional," everything will be sustainable, provided, clean, abundant everything, and on.
And, if that doesn't materialize, there's always Mars.
data backups are still required for all human knowledge data, depending on AI inference description of the data is a bad idea, need million year data storage still and 98% of it to be distributed widely, ( human intellectual data)
@matthew_berman regarding programming in the future: Who knows if it will happen in 5, 10, 20, or 30 years, but considering technologies like Neuralink or Meta's wrist-based interaction and other similar technologies, I think it's very likely that there will be a way to understand and interact with code in a manner that we can't yet imagine today. Who says we'll still need to learn code (however it may look in the future)? One of many possibilities might be that we can conceptualize programming and associated languages in the form of objects in our minds, similar to how synesthetes experience numbers, colors, and music.
@3:41 I work in âways that ai can solve problems that weâd not be able to see as humansâ and it already is happening
If ORCH-OR quantum effects are part of the placebo effect, there will be a place for humans in healthcare. Certainly LLM will be (already is) a massively useful tool and healthcare will be disrupted for the better!
jeah right, in the 70's it was a new iceage...
A world where humans are plugin cables is a no-no.
"#7: we will have new food & feertilizers" I just say: "Soylent Green"
will watch it
resource will be pletiful I think it refers to robots exploiting maybe the kuiper belt. This is basically an inexhaustible source of materials. Gettng satellites there will be a bit hard but once there self replicating bots will increase output slowly and send the resource on Earth orbit in a continuous stream.
good point
For number 5, it's already here in the embryonic stage. I use Perplexity more than any other search provider.
7:36 Uh, anyone know what he meant by this? No more software, just AI "computing directly"? Feels like there's a fundamental misunderstanding of computation here somewhere, but I'm not sure if it's mine or his...
The probability that specific series of words will follow other series of words is NOT the entirety of human knowledge.
What gives you such confidence? As a human, when you speak, do you not speak each word sequentially? How is speech not a clear example of how human intelligence is just a probability string of words?
Maybe it is. Maybe that's all that's really going on. We don't have good models of how we actually think. Maybe we just got lucky and replicated it.
Listen to Ilya Sutskever on this. To predict the next word more and more accurately they are understanding more and more deeply...
@@xTheITx logic doesnât work like these language models work. Not that Iâve seen so far.
Then there is the issue that a lot of knowledge is not articulated. So itâs not expressed in language.
But I guess a lot of people are content just guessing what the next word should beâŠ.
@@albeit1 And some are content doing what you're doing. I don't dismiss your perspective, I do dismiss your closed off attitude, but at the end of the day science IS observation, measuring, and predicting. Religion is "knowing cuz like i feel a type of way, but you can't know anything with 1000% certainty so my faith > your science". So.. when the duck stops quacking we'll stop calling it a duck. To me, your perspective is basically that the 2 year old isn't human because it isn't concioua yet - to which I say in due time - humans are brilliant and hard working. This is only the beginning, and breakthroughs keep happening. We will move beyond the autoregressive models, we will have a solution to every goal post that's hastily moved in a huff.
The "doctors" I've used with AI have been spot on. I like that they have no bias, no ego and listen to EVERY symptom.
I think AI is perfect for a doctor replacement.
AI thinks in language, humans think in images, we just communicate with language.
No one can predict the future right now. AI is just getting started and will rapidly change the face of physics, manufacturing, energy and more. Think Star Trek level right around the corner.
Resources? There are asteroids out there that are worth THOUSANDS OF TRILLIONS OF DOLLARS due to their resources.
Elon is busy making the vehicles to bring them down to earth after we harvest them and bring them to earth of course.
Hey.. realist not pesimist thank you very much Berman
AI should only ever communicate with other AI via vectors. Anything else is simply adding more work to your pipeline for no reason. If you need two AI to communicate you do not need to know what they say, only that they get the right message and produce the required output. By using vector based communication the model can bypass all the interpretation. They can simply forward their vector output to the next model in a format they have been fine-tuned to use when doing inter-agent communication that the next agent simply uses as a vector input to start generation. No tokenization, pure vector space communication.
A couple of thoughts, you are not looking at 5-8 years to replace most programmers. Even today novice and intermediate programmers are at risk of being replaced. As AI typically has three major iterations a year, two years tops, with the exception of the best in the industry (The 10Xers). Beyond that, with autonomous multi-agent, there is limited to no human in the loop to accomplish a given task.
9:02 "translation layer", anyone that lives in a foreign country can attest to the meaning being lost in the translation. I suspect this will be no different in this case
The programming language for computers you speak of is surely just raw assembly, we can read this it's just arcane. But this can be decompiled into high level source code to make it easier to read
expertise to a certain depth is nearly free now to people who are smart
Do you think Vector Data Bases Or API's Would be considered as the new Gasoline For Ai Models 2024 onwards?
11:44
One slight problem, the open weights of AI models are available now and no one really understands why it makes one choice over another. AI isn't a bunch of If-Then-Else statements, that's why some refer to them as "Giant Inscrutable Matrices". AI selfassembles itself from simple rules applies to tons of training data.
One thing we will see, probably within months is the replacement of Medical Insurance Specialists who work in every doctorâs office. Medicare, Medicaid and private insurance has become so complex. They earn on average 35,000 a year could be saved with a LLM
I am not a coder, but I do understand a little about Basic, Pascal and Lotus Notes. Again, not as a coder, just for curiosity (But I can't write a proper code, I just know the logic).
Well...I am creating a webpage, java, html, I am also doing the comunication with chatgpt (so it creates descriptions for somethings that will happen there)
I am using Minstral and chatgpt (to a lesser extend).
The code is 500 lines so far, everything is working perfectly. Sure, I need to ask the AI to correct many things it does wrong here and there. But with the right prompt, it's going on very nicely.
During a moment, where I couldn't understand what to add and change in the code, Minstral , thanked me for my patience and did a very thorough explanation and showing the before and after....
So...yeah...we need to adapt...not fight it...
Will it fix the exonomy though, i hope so
Great video - BTW, his name is pronounced as vinOd, not vinAD.
If we are employees the future looks very bad.
Herded cats are not very productive, unless you're producing cat fights