In Davos bubble, AI leaders say the AI bubble isn't real
 |
| Credit: Harun Ozalp/Anadolu via Getty Images |
In Jan. 2026, the gap between the haves and the have-nots could hardly be more stark — even in AI world.
On the have-not side, there's OpenAI CEO Sam Altman. Long the hype man for imminent digital superintelligence (aka AGI), Altman backtracked and said we're in an AI bubble last year. He's made odd deals to feed the money hog known as ChatGPT; subscriptions ain't paying the bills. Last week, Altman rolled out what in 2024 he called "the last resort" for the company — selling ads on ChatGPT. A popular LinkedIn post this week used an AI image of Altman out in the cold, begging for change with a self-mocking cardboard sign: "AGI = Ads Generate Income."
And then there are the haves in warm rooms in snowy Davos, Switzerland — where the weather report for the AI economy, from the annual World Economic Forum, is just peachy.
Jensen Huang, Nvidia CEO, who rode the company to a $4 trillion valuation on the back of powerful GPU chip sales growth that hasn't quite slowed yet, flipped the script when asked about the AI bubble. "This is the largest infrastructure build-out in human history," Huang said of active and promised data center projects. "And so the AI bubble is, comes about because the investments are large. And the investments are large, because we have to build the infrastructure necessary for all of the layers of AI above it."
Huang wasn't denying the existence of a bubble, exactly — he just suggested it was a misreading of our current economic outlook. Just look at all that infrastructure coming down the pipe! Left unsaid was whether those massive projects would dry up if AI itself continues to not show returns on investment for regular businesses (especially if AI models like DeepSeek can be built with minimal data center usage, the opposite of what made Nvidia successful).
In Nvidia's telling, the tail is now wagging the dog. And the rest of Davos, a popular confab for billionaires, pretty much wagged along in agreement.
"I think there will be big failures, but I don't think we are in a bubble," said Larry Fink, CEO of Blackrock, a top tech investment firm. (Blackrock holds more than $200 billion in Nvidia stock.)
Also at Davos, one Nobel Prize-winning economist described what was currently happening in AI as a "rational bubble", comparing it favorably to the infamous tulip bulb panic of the 17th century. Why? Because with AI, insisted Peter Howitt, economics professor at Brown University, "there's something real out there."
Howitt didn't suggest what that real thing was, but insisted there would be a winner — and their arrival would herald the bursting of the bubble. "At some point, when it becomes a little clearer who the winners are going to be, the values of the other firms are going to start to fall, and that's when the crash will take place."
So who is that winner, other than Nvidia and its circular OpenAI deals? Microsoft, with its own favorable OpenAI deal, might be in a position to "win" the AI economy. Microsoft currently owns a 27 percent stake in the ChatGPT maker. If Altman can't keep the lights on long-term, Microsoft CEO Satya Nadella could well ride in to the rescue and snap up OpenAI itself.
Nadella, on the surface, seemed just as optimistic about AI's future as the rest of the titans. And yet in his appearance, the Microsoft CEO also sounded a note of alarm. If the AI economy doesn't root itself in non-tech sectors, it will be a bubble — and soon.
"The real question in front of all of us is how do you ensure that the diffusion of AI happens, and happens fast," Nadella said. "For this not to be a bubble by definition, it requires that the benefits of this are much more evenly spread." (Nadella has been criticized by investors for too much AI infrastructure spending.)
So to summarize the view from the Davos bubble: We're not in an AI bubble, we're in a massive infrastructure investment that will benefit workers. Well, maybe we're in a rational bubble. But we will be in an irrational bubble if everyone doesn't get on board with the AI economy real soon, presumably by sending poor Sam Altman some ad dollars. Got that?
AI bubble is fear mongering propaganda by shorter
ReplyDeleteJust keep the earnings reports legitimate. Thats a fair ask. Investors can make their own decisions based on accurate data.
ReplyDeleteI'm not allowed to say anything. Community Guidelines.
ReplyDeletePeople should really watch Sneakers with Robert Redford to understand what "A.I." is.
ReplyDeleteThe bet is AGI. It's not a bubble if they are able to advance the tech such that it is doing its own AI research and advancing itself into super intelligence. Then if they're able to control it--that company will break out at the bubble and own everything.
ReplyDeleteThe alternative of course is a bubble.
AI generates nothing but entropy.
ReplyDeleteThe sooner the bubble - and all idiots betting on it - crash and burn, the sooner we can get computer components back to levels where they're not insane.
ReplyDeleteMost of the time market prices never go back down once the public gets used to a new price point.
DeleteIf the data shows only 10% of Americans Businesses are actually using AI in production then what’s with all the layoffs? Amazon just reported they plan to lay off 14,000 corporate workers in January due to AI.
ReplyDeleteBlended into that is rebalancing after giddy amounts of hiring across tech over the last decade
DeleteThey are getting short of money so they have to let people go and make the remainder work harder
DeleteAnd funneling that into AI… with the hopes that it pays off.
DeleteOnly 10% use in production?
ReplyDeleteI don’t know a single business where ten employees don’t rely on some form of LLM on a daily basis.
But are they or their company paying for it?
DeleteI am thinking that it may come in handy for autonomous aircraft traffic when these evtols are finally in numbers that overwhelm the faa.
ReplyDeleteAI is just another overhyped product that will eventually go the way of the 3D tv.
ReplyDeleteNo way. AI now is where internet was in early 90s. It’s going to revolutionise every little aspect of your life for better or worse. 30 years ago internet was a novelty and now villagers with no indoor plumbing or toilets have a device in their pocket that has access to world knowledge. AI wearables will be the next cellphone. Eventually everyone will have one.
DeleteAnd maybe you don’t believe in the tech or whatever but companies and govts are pouring so much money into it that it will be shoved down your throat in every way possible whether we like it or not.
AI wearables?
DeletePeople need cheaper groceries and universal healthcare not some fad for out of touch celebrities.
I wasn’t talking about groceries and healthcare (and neither were you), I was talking about AI
DeleteI know what people need, but people will get whatever makes the companies that run this country more money.
So I’m telling you, maybe you want to educate yourself about what’s coming rather than reduce the next technological revolution to “3D TVs”.
You know if you are a contributing member of the majority of society you can’t really survive without a smartphone? Well that’s what AI wearables will be like in the near future. That’s low hanging fruit though. AI will be incorporated into every business and system they can think of because they will get desperate after sinking billions and billions. And if you think it’s a a bubble that pops, well it’s going to drag everything down with it. If you think groceries are expensive now, wait during a depression.
Good day to you.
You are indian
DeleteEw
DeletePart of how this happened is that corporations like stress and others have been flush with extra cash for a while. Believe it or not, we had a pretty good economy for a few years not too long ago.
ReplyDeleteBut with that extra cash, they found little to do. Major corporations often already have all the funding they need to expand operations when needed. So they did stock-buybacks, because that would please the more diehard shareholders and they could see no other profitable ventures to invest in at the time.
But buybacks aren't popular with everyone, and nobody really likes it when corporations have a pile of excess money sitting around. Employees wonder why they aren't paid more, shareholders start eying the cash too because if it isn't going to be spent in the company they want it paid out to them, the government could always start taxing too, there's just no good angle for a corporation to hoard this cash.
Then along came AI and suddenly they have the perfect place to channel this extra funding into. Not to say AI is perfect, it might be a disaster, but it meets the need to find something to send this extra capital into.
That's one contributing part of this.
its equivalent of free energy hoax
ReplyDeleteI'm covering my ears ..
ReplyDeleteonce the paywall is up, its dead. they should be making it up in all those jobs it’s replaced. where is that cash going? there also should be a tax burden for companies to replace employees but companies won’t be honest. its just making rich richer. no regulation. we were not prepared for this.
ReplyDeleteclaude code max 20x costs $300 usd a month and i don't know a single developer personally who isn't paying for that. it's 100% worth it to me.
ReplyDeletei think it's still a bubble, but the paywall = 0 adoption thing depends greatly on the industry.
I’m not paying 300 for that shit.
ReplyDeletewell i don't, my business does because it does make me about 300% faster. but again outside of programming, i see 0 value to any LLM product.
ReplyDeleteDoes it make you 300% faster? Like people review your PRs 300% faster? Your stakeholders get 300% faster? Your PM writes tickets 300% faster?
ReplyDeleteYou starting to see the problem here? If anything this just means dev's get to spend a lot more time not working waiting for everyone else to do their jobs
ReplyDeletei'm not in that type of company but I can see how it'd work like that in a typical agile workplace.
for me, it makes the me receiving basic feature request all the way through to passing CI and me merging 300% faster, there's nobody else in the loop. the backlog of feature requests is long and evergrowing so there's no downtime for me. i get to spend more time on the architecture/planning side of things as the actual raw programming time is very fast, especially when AI agents are doing TDD.
the people i know whom i mentioned earlier are actually all in relatively similar positions, companies that massively downsized after covid boom collapsed, have taken on many more responsibilities. in my case the company went from ~8 local developers + product manager + designer -> me + ~6 outsource developers -> me
Aha, no I work in Fintech for a big enough tech company. There's about 200 dev's in my product group alone, every PR has to be approved by another member of your team.
ReplyDeleteWe don't use agile at all tbh, everyone hates it - so every team just kind of figures out their own system. Mostly you manage your own projects and you work on smallish teams with focused goals - eg I've worked on Subscriptions, Quoting, Approvals teams.
I agree that AI definitely lets you spend less time on the code in fact I think the way you are specifically using it - yourself in an IDE is the most successful way of using it. And that's pretty key to this whole argument.
Like we have slack set up so you can open an issue based on the conversation and the AI agent starts to code it and then you can just go into GitHub, leave comments and get it to update it. Except that's the way I think we keep creating very difficult to notice bugs.
Tbh I enjoy it most when I'm working in an IDE, we have full subscriptions for Cursor, Claude IntelliJ, OpenAI - we also have a partnership with OpenAI. Basically we have loads of options - I see a lot of people on here talking about models and I'm like - I don't even notice which one I'm using half the time because - if your coding java its Claude, if your coding FE its whatever Cursor happens to pick.
I think it works a lot better when I'm working in an IDE like you - it definitely speeds me up, I defs agree you spend more time on architecture - but you have to know what architecture you should be targeting. Not to mention how thats going to integrate into whatever DB,Kafka,Redis,Hadoop, etc other systems your using.
Like theres choices you have to make and tbh the AI will just suggest all of them randomly and if you built a system like that - your just creating a giant mess.
Which is why I think this whole Devs are all gonna get fired shit is a bit bullshit.
Totally understand your position tho - I worked jobs like that earlier in my career - I think it's just the bigger the company, the more people are involved, the bigger problems you have to work on - it just gets more complex.
Tho at least in big tech companies the kind of tech solutions you can use are usually fairly locked down and highly regulated. When I was working more your style company earlier in my career I could practically use any technology I wanted - I feel that could be an even bigger mess if you just let the AI run the AI so to speak.
ReplyDelete"Which is why I think this whole Devs are all gonna get fired shit is a bit bullshit."
I think we are protected as senior devs, I do not see junior or even early-mid developers surviving. Obviously that means there's no new talent coming in which is incredibly short-sighted, but that seems to be the way things are going.
I also feel that even if companies do keep hiring juniors, it'll be very demotivating as a student to stay motivated while learning, knowing they won't outpace AI for probably the first few years (with that threshold always moving away from them as times goes on)
ReplyDeleteI don't think this is true, because there's no way say a BSA would be able to go into the codebase and actually understand it. Think of all the different technologies they need to know about and understand how to put that architecture together.
And AI makes mistakes all the time, like you tell it to fix tests and it fixes them by removing all the assertions 😂
Like your job is to build the architecture of the company, AI is a great tool for helping you do it - but you still need to understand that architecture.
Like you can't say hey AI run my company, I've seen people saying oh just run it in a constant loop improving and that works for green field projects - but it doesn't work when you have 60,000 micro services spread across 15,000 repos.
Like even in my own team we own about 25 different services that all have their own repos, and the AI needs to figure out all the connections between the data between them. We have agents that do all the fancy agent stuff - they can search the entire companies code base, they have one-pagers for all our standard practices, but tbh - for somethings they just need direction.
Like, this takes brain power - and of course coding isn't even half my job - my real job is talking to business people who know nothing about how logical systems work and translating that into software and telling them how their ideas *cant* work in some cases. Juniors need to learn all that shit too.
ReplyDeleteI agree with everything you're saying, i just don't trust the powers that be in the tech world to realise this.
They're just hearing "AI can make a standalone CRUD controller with 0 dependencies 100x faster than any junior engineer" and laying off juniors.
It'll bite them in the ass eventually but by that point I think we'll have a massive skill gap.
I think the issue is it isn't profitable at that price point.
ReplyDeleteBeen a dev for 25 years and I don’t pay some 300$ a month subscription… I use plenty of AI tools for assistance, research, and to answer questions when I read technical literature, but I haven’t spend that kinda money yet..
ReplyDeleteHowever, I was talking to a friend who has been warming me up to the notion. I’m no stranger to agents and tooling, but the price tag for max service seems appalling to me. Regardless, I’ll do some more reading on the offers as they are today… it’s been a minute since I last checked.
I didn’t know about claude code. If companies are saving on headcount maybe they will dish out for subscriptions. They should keep the basic features free and charge for integration and business use. i believe copilot is like that already. If companies can foot the bill for it, maybe it will work.
ReplyDeleteCircular financing what make this bubble most dangerous. All it takes is one leg to fail. Look on the bright side. Index based ETFs will survive and LETF versions will hit rock bottom with massive top side. Buying those when underlying hits 20% of last all time high might be a smart buy.
ReplyDeleteYou mean, we take out one billionaire we take out all the billionaires and private equity..... I am listening.
DeleteGood luck with that
DeleteNah you won’t take them out. They could lose 99% of their wealth and still live a comfortable life. When this thing comes crashing down working people will be the ones to suffer most.
DeleteHow wealth transfer happens.
DeleteJesus you’re spamming this across all AI related subs. The endless stream of AI IS A BÅBBLE OMGOMG is extreme tedious and repetitive and click bait.
ReplyDeleteShows you know very little about where it’s heading
DeleteWhere is it heading?
What happens to OpenAI if they don't achieve AGI in the next couple years?
They already redefined AGI on the backend. Publicly, they're still selling the god AI that will solve science and scarcity. Privately, it's primed to be just a beefed up LLM. So OpenAI will claim ChatGPT has achieved AGI regardless.
Maybe then the investors will realize they've been sold a lie.
If you haven’t used Cursor with GPT 5.2 Codex or Claude Opus 4.5, you shouldn’t be able to have an opinion on AI. This shit is scary good. Office work is trivial when compared to SWE. There will be a massive reduction in workforce in the next few years.
ReplyDeleteHave you used LM Studio with openai/gpt-oss-20b - because its pretty good and it just runs on your MacBook
DeleteYes but it’s not profitable And the amount they will need you to spend to make it profitable is more than most can spend. They are counting on getting people hooked and hoping the increase prices will be tolerated.
DeleteThe whole bet for AI in software is dependent on people generating enough software that hosting all of it on these cloud providers allows them to make back the money subsidizing the token generation used to create the software. The problem is, how much software generated by AI will actually be profitable and thus worth consuming cloud compute infrastructure?
DeleteBut, sir. +1 btw, its not profitable until they get rid of the workforce. Then its all red hot honey magma lovely goodness. I am told.
DeleteThey are building AI factories like we used to build mainframes. The advancements are going exponential.
DeleteInvesting in infrastructure is expensive. Model costs keep falling and intelligence keeps getting cheaper. I think the people investing billions know this lol
DeleteInfrastructure is generally beneficial for everyone though. Does AI benefit anyone outside of corporate moguls trying to lay off tens of thousands of people to increase profit margins?
DeleteArguably large corpos have the least benefits.
DeleteAnd you figure that how…?
DeleteLarge corpos have large projects, that are complex and LLMs don't handle that very well. It's also scary to de facto hand over your entire business to another corpo. Big boys won't be firing anyone any time soon.
DeleteThe most benefits from LLMs come in areas that don't need reliability and are on a budget - start ups, mom and pop shops, businesses without dedicated IT, etc. can now avoid hiring devs or artists/animators or at least come much further with AI for much cheaper and faster.
The better the product must be, the more you will need capable humans involved.
But even outside of business, if you are a casual user, it's just a better search engine. I would probably pay 5$-10$ a month for this alone. You do have to double check the sources, but it's better that having to fight your way through "10 best" and other low quality and irrelevant slop that Google spits out.
DeleteEnterprise AI tools don’t collect and train on your data, and LLMs absolutely can handle big code bases.
No offense but you’re speaking out of your ass. It sounds like you’re referring to AI tools from 2024. Every tech exec that I talk to has their engineers on Cursor/Claude Code.
Source: ex-FAANG and working for a F50 company right now.
DeleteI don't care if they collect or train, I would only care that another company is a hard dependency for me to do anything - remember, I don't have people who know how the fuck anything works anymore and am relying 100% on AI tools.
I also don't know what your point is with saying engineers are using cursor and claude, there is a big difference between having engineers use it and them being autonomous.
Need you to define "large" here big dawg
DeleteI think Google will win this one. They have all the data and the build out. But the cost to others is extremely extensive. To recoup like an Amazon you need to be the only offering.
DeleteNo pension for you, all year
ReplyDeleteInshallahj
ReplyDeleteYes
ReplyDeleteWe can certainly go with out it, thats how you know its bubbling. Im still not sure how I would actually integrate into my life. Seems like a nothing burger at least for myself
ReplyDeleteAI doesn’t exist, so yes
ReplyDeleteSaw some stuff saying that Open AI's financials are looking quite dire
ReplyDeleteSeems like part of the problem is that we're reaching the point where achieving even a marginal improvement in output requires exponential increase in hardware and power.
The destruction of tremendous amounts of paper wealth owned by our tech overlords is very clearly a nearly unmitigated win for the general public.
ReplyDeleteI hope every ai company goes bankrupt
ReplyDeleteAI is just the next platform of societal control. Please burst
ReplyDeleteNo, the biggest bubble is the US Government debt balance:spending.
ReplyDeleteIt’s not a bad bet. But they should have kept all those software engineers to push it farther. AI isn’t where it needs to be to be fully functioning.
ReplyDeleteThey are too short sighted and that’s why this bubble exists.
Yes
ReplyDeleteI think a large portion of this is businesses thinking they can cut their work force in half and save money and their afraid that if they don't jump on the AI tech bubble they'll be left behind and cease to exist, problem is that there is heavy resistance from both their customer base and the people they will have to layoff/fire so it's a lose lose situation for everyone involved especially in this post capitalist bullshit that we find ourselves in.
ReplyDeleteMy opinion is we're going to repeat patterns seen in Japan's lost decade in the 1990s. Money printing up until we're at 200% debt to GDP. The Fed Rate staying in single digits. Inflation at 5 to 10 percent each year until silver is at $250 an ounce and gold at 10k an ounce while the government lies and says its between 2 and 5 percent.
ReplyDeleteThe dow will hit 100k but a trip to taco bell will be $50 and a tank of gas will be $150 while rent hits $4,000 a month.
When the government bailed out Silicon Valley Bank I think it signaled that there are too many people in government that would opt for short term inflation pain via money printing / QE rather than a legit 1970s style recession in an attempt to fix the system. I don't think we'll ever see someone as principled as Paul Volcker who was running the Fed back in the 1970s who would actually have the balls to raise the fed rate up to 15% and sink the economy and have everyone running to dump their money in to Bonds.
I believe staying in large caps and out of cash will continue to be a winning strategy to ride the massive wave of cash that will continue to hit the market the next 10 years.
But lets see if we have a legit recession anytime soon. Cheap money keeps the music going for a few more quarters!
Honestly it feels like the dot com bubble. Too much investment in a short period of time for the technology to be able to return profit. The bubble will burst and there will be some winners and many bankruptcies. But the technology will remain and continue to progress, and over time companies will learn to capitalize on it more profitably.
ReplyDeleteHa, no problem here, my 401k got wiped out in the "Great Recession"
ReplyDeleteRetirement is a myth.
AI is gonna crash as soon as someone says they can't grow anymore due to lack of equipment such as memory or nVidia posts a loss.
ReplyDeleteAI is the biggest bubble in history... so far.
ReplyDeleteThe "privatize the gains, socialize the losses" will be so great that it will incite widespread rebellion.
The biggest bubble so far.
ReplyDeleteThe bubble WILL pop.
ReplyDeleteRussian Ai tasked with attacking the west, has taken over.
ReplyDeleteThe SP crash is going to be epic. But it may take yet literally decades to happen, nobody knows.
ReplyDeleteI hope it goes away in the creative field. I want to create my own art. I also don't want A.I. to steal my art.
ReplyDeleteYeah that's not happening but a price correction might
Deletetools remain, absurd valuations bite the dust. look at cisco.
DeleteYes
ReplyDeleteGod, I hope so. It’s stealing data from Artists and others to populate
ReplyDeleteAI is being viewed as the next industrial revolution or the next printing press, so everyone with capital wants to own it in advance because having an early lead has historically let to an effortless monopoly and huge profits... passively.
ReplyDeletethe unfortunate truth here is AI does not replace mechanical tasks. and all work is still largely mechanical in nature or else we wouldnt need a human for it, wed have digitized it or at least put it on paper and turned it into an automated letter and then offloaded the task to someone else and reduced the number of workers in the company already.
so we keep investing into more and more digitial enhancements, which is great, but ultimately incapable of replacing humans.
money changing hands at the top, or investors giving other companies large sums of money doesnt matter at all because its rich peoples money, its untaxed. its never recollected at any point in time. some other rich person just ends up with a couple more zeros.
Don't forget they're buying real equipment and building real buildings and expanding real infrastructure to satisfy promises about trying to make AI work.
DeleteWhat happens when those promises turn out unfulfilled?
we end up with massive amounts of e waste and land+water pollution 👍👍👍
DeleteAI is just another web 3.0 scam, but one that investors finally bought into. That doesn't make it any less BS than ape jpegs and etherium.
DeleteHonestly true artificial general intelligence will likely be like the industrial revolution, but I don’t think it’s possible LLMs can be AGI. They work like a more advanced word prediction like you see in many programs but it doesn’t understand what it’s saying, only what the next most likely token is based on the previous tokens.
Delete
DeleteIt's not like that already, LLM is capable of thinking about answer and not just token-after-token, like it was. It's better every year/month and advancing pretty fast.
Also, you can check how Ai solves mathematical questions by itself and how it gets better here every year. Astonishing.
Historically, a lot of money was thrown at such industrial innovations, and in a boom and bust cycle that “lead” helped a lucky few get a head start, while most lost their money because technology wasn’t able to deliver on the hype. This happened with canals, railroads, and the internet. The short term results tends to be useful but underperform, and AI will be no different.
Deletethe truth is, the current crop of “LLM AI” isn’t actually intelligent. Billionaires spent a whole lotta marketing money to co-opt the term “AI”
Deletecurrent AIs are effectively a dumb human sounding dictionary, it’s great when the actual content doesn’t really matter and only needs to sound human (IE TPS reports to your boss)
generative AI is great for a cheap laugh and that’s about it.
Not actually true.
DeleteI have used to to :
- write software for controlling DMX lights
- identify CNC controller boards from photos taken through ventilation holes
- debug musical hardware issues with linux
- advise on CNC feeds/speeds for cutting metal.
- write JScad code to generate STL code for making guitar bodies.
And dozens of other things of a similar nature. It is actually really good at things that can be checked right away.
a google search for each of this is pretty much just as quick - so effectively your “AI” just summarized it for you.
Deletethat isn’t actually intelligence and most likely you had to check it to see if it was actually true.
so basically cliff notes for search results that sound human LOL
That is total rubbish.
DeleteI am actually pretty wary of this tech, but you are speaking from a POV of ignorance.
Use google to identify the driver in the attached picture.
https://imgur.com/a/H2jtoH7
You won't be able to do it.
Elsewhere on this thread is another programmer with 30-40 years experience. I started programming in 1979.
I asked if it was possible to control a DMX light with python, and not only did it say yes, it also wrote the code - which worked first time, and took about 4 seconds... which included how to install all the dependencies and how to run it.
Let's see you do that with google. I have decades of commercial programming experience - which includes spending fucking DAYS with problems like those above, searching with google through discussions (which more often than not are 15 years out of date), trying to find something that might work.
If google was that effective these discussion boards wouldn't exist - they're what happens when google doesn't deliver, and they have to resort to asking other people - across continents so these convos take days as well.
..
Here's another one : Try doing a google search that will write the code to superimpose in inverted parabola onto a circle so it makes an egg shape.
You won't be able to do it.
..
The mother of all stupidity is people barging in with judgements about things they know nothing about.
That's you.
1.Quick google image search with that picture
Deletehttps://www.google.com/search?vsrid=CO2Rve7J-evmggEQAhgBIiRhOGUwZjM3NS01NTYzLTRiZGQtOWQ3My1mN2MxMjE4NmE5MTYyBiICYmkoBDjKo4Ct1aeSAw&vsint=CAIqDAoCCAcSAggKGAEgATojChYNAAAAPxUAAAA_HQAAgD8lAACAPzABEP8HGK8EJQAAgD8&udm=26&lns_mode=un&source=lns.web.gisbubb&vsdim=1023,559&gsessionid=e4er4qEJxDqm684M7DGnQjad9eUXqoLu1ifO7ODb9dcOwMSHFyr0pg&lsessionid=_Y1itbNtiA968hYkHkGDdoTkIgopL5RWpJULP3jrThR0G7fgTX3fXg&lns_surface=26&authuser=0&lns_vfs=e&qsubts=1769377638360&biw=1750&bih=875&hl=en
2.Quick google: is it possible to control a DMX light with python
https://www.google.com/search?q=is+it+possible+to+control+a+DMX+light+with+python
Top links literally shows https://pypi.org/project/PyDMXControl/
sure i guess if you're too lazy to read through the tutorial, the first page literally gives you copy/paste code and how to install dependencies.
3. Quick google: code to superimpose in inverted parabola onto a circle so it makes an egg shape numpy
https://www.google.com/search?q=code+to+superimpose+in+inverted+parabola+onto+a+circle+so+it+makes+an+egg+shape+numpy
Top results show stackoverflow link with copy/paste
https://stackoverflow.com/questions/50171718/superimpose-two-parabolas-on-each-other-and-limit-display-ranges
AI can be useful, but the way you describe it makes me fear you’re outsourcing your learning.
DeleteIf you're that interested in learning, maybe start with a question rather than a declaration.
DeleteYou never know. You might learn something yourself.
DeleteI hear you, I'm doing tons of serious things with it too. It's just passed a threshold in the last month or two where it really can handle complex tech stuff (write software, test, iterate, deploy, monitor...) almost entirely on it's own.
I've been coding for 40 years, professional dev in various guises for 30, and in the last few weeks I realized I probably won't write any more code.
But with AI's help, I'm building and shipping maybe 10x?
Anyone making the argument that genai isn't useful just isn't paying attention.
Yea - I'm getting a lot of that.
DeleteThere seems to be this eternal memetic froth of people going "ooh, AI can't do this, it can't do that"... without actually having tried it themselves.
I got hooked on Midjourney for about a year - anyone who says it can't do original art because everything is a pastiche of things that have been done before, is simply wrong - and probably hasn't actually tangled with it.... tried to feed it weird stuff to see what it's native imagination is like.
It is fucking incredible - and I have spent a lot of time in art galleries, all over the world.
That said, I'm not entirely opposed to going full Butlerian Jihad on the whole thing until it can be wrested from the hands of billionaires. The whole alignment issue is moot if it is being developed by humans and human-systems that are already profoundly mal-aligned to human survival.
Another problem with the critiques is that they are tending to split Human vs AI into a binary - when actually it's more a case of the boundary between the two coming down. Less a case of "was this created by an AI" than "how was AI used in the co-creation of this?"
Back to the Butlerian Jihad - I think there's an opportunity/market waiting out there that caters to anti-AI purists - and to be fair, I can understand that perspective. I like cat videos but they've got to be real cats.
If you use it for things that can easily be checked, why don't you just check in the first place?
DeleteBecause I don't know what to check.
DeleteThe DMX light controller for example - although I am a programmer, it would take me hours and hours of googling to try to find out how to write the code for that particular protocol. To be honest, I wouldn't really know where to start. I didn't even know if it was possible to do it without going to machine-code level or whatever.
All of the things in the list above involve hours if not days of work.
https://imgur.com/a/H2jtoH7
That is the photo of the CNC controller that I took through a ventilation hole... GPT was able to correctly identify it in fairly minute detail - and tell me how to set up limit-switches etc.
How would I begin to find out what to even check if I was doing it on my own? Google pictures of controller-boards until I found one that looked the same?
GPT did it in about 4 seconds.
You trust it way too much.
DeleteHow the fuck would you know what I trust?
DeleteDid you miss the bit where I said
"It is actually really good at things that can be checked right away."
..
Here's a nuanced observation : I am old enough to remember when calculators were not a thing, and time was when if I was confronted with something like "17 * 53" I'd work out how to do it. Now my mind just goes blank and I reach for a calculator.
I can feel a similar sort of "abandonment of diagnostic attention" creeping in with AI... things like "how to change the default settings in Chrome"... instead of working it out, I can feel my mind going kindof blank, and I reach for a GPT.
So I can kindof see it making people dumber - but it seems to be making people who are against it stupider as well - and this comment section, and your comment is a classic example of this.
You think you know what other people are thinking.
You don't.
Your perspective is odd but I hope it gives you comfort.
DeleteBoth the printing press and the industrial revolution took some time to kickstart and deliver actual results.
DeleteHere with AI the gamble is so big and soon on the first stages that someone is going to get hurt economically.
I believe LLM are going to be used but bit how they think or hope and not with those astonishing results promised.
All work is largely mechanical? Everybody that is able to work remotely or work from home during covid would be exceptions to that rule. The entire knowledge-work sector. Even if we have no huge advances in robotics to automate physical work, just automating knowledge-work could easily justify current valuations. But if it doesn't pan out like that then yes, we're in a huge bubble that's going to have huge consequences if/when it bursts.
Deleteyou misunderstand.
Deleteyou signing a piece of paper and putting it a folder in a binder for an audit is mechanical. every piece of walking, handshake, print job, phone call, is still "mechanical".
i guess you can replace mechanical = real workd task. analog is the oldschool world. digital = can be automated, scripted, or managed by an AI.
if youre doing VOIP you can script calls. if youre actually punching in on the keypad thats mechanical.
My point is that if a task has been automated, you no longer need to sign a piece of paper or put it into a folder for the task to be completed. Automating it removes all the physical elements. Knowledge-work is the processing of information. When humans do knowledge-work today they might walk around offices holding pieces of paper, shake hands and make phone calls. But that's just because humans are in the loop, not because knowledge-work requires mechanical steps.
DeleteIf your job requires physically manipulating objects (like a plumber fixing a pipe) then it'll need robots to automate. If you're emailing and using spreadsheets, your job could well be automated by AI. A lot of people work in offices, they don't have mechanical jobs, they process information.
thats the thing, you cant automate all the office work that currently exists. theres still a mandatory human component.
DeleteI'm not saying AI is capable of replacing all knowledge-workers today. But can you give me some examples of mandatory human components? What knowledge-work is impossible to automate?
Deletefly to japan, have a night out on the town to earn cultural respect, copy mannerisms, convince investors. because its a person you need to convince, not another ai.
DeleteThat's certainly one of the last use cases to be automated but it's still perfectly feasible and therefore inevitable. The most successful investors will have their AI agents negotiate with your AI agent and thousands of other AI agents to find the best investment opportunities. People will still negotiate the old way face to face for many years. But they'll slowly become obsolete and be replaced by more efficient systems.
DeleteBut if we pay taxes on that money???? How will we beat China???????? /S
DeleteAnother bailout incoming? Most likely
ReplyDeleteNo tears for any bucket over here if that shit pops.
ReplyDeleteI hope so.
ReplyDeleteYeah thats lowkey scary tbh. Early tech waves often overbuild infra first. Cloud and internet looked insane early too. Some firms will crash, but the tech itself usually sticks around
ReplyDeleteExcept the infrastructure is worthless after a few years. This isn’t railroad tracks or fiber cable. AI is a scam.
DeleteCloud is still looking insane. Azure still grows 40% YoY.
Delete