AI Added 'Basically Zero' to US Economic Growth Last Year, Goldman Sachs Says
gizmodo.com
gizmodo.com
I have been a paid AI practitioner since 1982, so I appreciate the benefits of AI - it is just that I hate the almost religious tech belief that real AI will happen from exponential cost increases for LLM training and inference for essentially linear gains.
I get that some lazy ass people have turned vibe coding and development into what I consider an activity sort-of like mindlessly scrolling social media.
Where are they?
Are we sure that's not a misunderstanding of the terminology? Artificial diamonds, such as cubic zirconia, are not diamonds, and nobody thinks they are. 'Artificial' means it's not the real thing. When will conscious, actual intelligence be called 'synthetic intelligence' instead of 'artificial'?
Incidentally, this comment was written by AI.
I saw someone on the news claiming this recently, but he ran an AI consultancy firm so I suspect he was trying to drum up business.
When computers have super-human level intelligence, we might be making similar distinctions. Intelligence IS intelligence, whether it's from a machine or an organism. LLMs might not get us there but something machine will eventually.
People who declare that AGI is coming.
P.S., hope your cat have nice day.
And nobody working in the space either as ML/AI practitioners, or as philosophers, or as cognitive scientists, even thinks we know what consciousness is, or what is required to create it. So there would be no way to tell if an AI is conscious because we haven’t yet managed to reliably tell if humans, or dogs, or chimpanzees or whales are conscious.
The claim that is often made is that more work on the current generation of AI tech will lead to AGI at a human or better level. I agree with Yann Lecun that this is unlikely.
I would draw a separate line around sapience, and particularly the capacity for suffering, maybe indeed attributing it to mammals and birds but not insects, but consciousness seems more widespread to me.
I've always doubted it, but then again I've also been skeptical about claims that humans have these capabilities.
Synthetic sounds more neutral, aside from bringing microplastics to my mind.
I guess the field of artificial life has the same issue.
As another comment pointed out, you don't necessarily need consciousness for intelligence. And you don't need either of those for goal oriented behavior.
My favorite example is the humble refrigerator. (The old one, without the microchips!) It has a goal (target temperature), it senses its environment (current temperature), and takes action based on that (turn cooling on or off).
A cuter example is the dandelion seed. It "wants" to fly. Obviously! So you can display goal directed behavior as the result of natural forces moving through you. (Arguably electricity and glucose also fall in that category, but... Yeah...)
LLMs, conscious or not, moved into that category this year, in a big way. (e.g. Opus and Codex routinely bypassing security restrictions in the pursuit of the goal.)
Does it really have goals, or does it merely appear to act as though it has them? Does it appear to act as though it has consciousness?
(I forget who said it: it won't really disrupt the global economic system, it will merely appear to do so ;)
Also, here I am! :)
I haven't met him, but a famous (pre-ChatGPT) counterexample is Blake Lemoine:
> In June 2022, LaMDA gained widespread attention when Google engineer Blake Lemoine made claims that the chatbot had become sentient. (https://en.wikipedia.org/wiki/LaMDA).
It's also not uncommon here to see someone respond to a comment questioning the consciousness or sentience of LLMs with the question along the lines of "how do you know anyone is conscious/sentient?" They're not being direct with their beliefs (I believe as a kind of motte and bailey tactic), but the implication is they think LLM are sentient and bristle when someone suggests otherwise.
One can bypass the whole sentience discussion and say that AI stands for Automated Inference.
If actual, conscious intelligence were to manifest synthetically, as in silicon-based rather than carbon-based, it is a losing battle to convince people because of the philosophical “problem of other minds.”
If there is a functional equivalence between meatspace intelligence and synthetic, it will surely have enough value to reinforce itself, philosophical debates aside.
On terminology, I would argue for non-biological intelligence. People can be awfully bioist (biological racist).
It depends a bit what you mean by conscious but assuming it's human like then it incorporates a lot of feelings, vision, sound, thoughts and the like, things that are not language really. But we do it with neurons and some chemicals and I imagine you could do something like that with artificial neural networks and some computer version of the chemistry, but not just language really.
I naturally have a hard time stopping when almost done with something, but with AI everything feels "close" to a big breakthrough.
Just one more turn... Until suddenly it's way later than I thought and hardly have time to interact with my family.
The reason I believe this is because I recently went through a really annoying battle with Claude trying to get it to stop being so strict with its sandbox. I wanted it to simply load some sanitized text from a source online, and it just would not do it. The sessions when I was sorting that out were so much easier to stop and moderate than the ones where everything just kept flowing effortlessly.
Robert Solow, Noble Prize winning economist, 1987.
1)
What he means to say is, say you needed to get something done. You could ask AI to write you a Python script which does the job. Next time around you could use the same Python script. But that's not how people are using AI, they basically think of a prompt as the only source of input, and the output of the prompt as the job they want get done.
So instead of reusing the Python script, they basically re-prompt the same problem again and again.
While this gives an initial productivity boost, you now arrive at a new plateau.
2)
Second problem is ideally you must be using the Python script written once and improve it over time. An ever improving Python script over time should do most of your day job.
That's not happening. Instead since re-prompting is common, people are now executing a list of prompts to get complex work done, and then making it a workflow.
So ideally there should be a never ending productivity increase but when you sell a prompt as a product, people use it as a black box to get things done.
A lot of this has to do with lack of automation/programming mindset to begin with.
> Robert Solow, Noble Prize winning economist, 1987.
Some skeptic was wrong in the past, therefore we should disbelieve every skeptic, forever.
That's the argument, right?
A similar historical thing is when factories went from steam engines to electricity. Steam factories had one big engine connected mechanically to many tools and conveniences in the factory. So they replace the one big steam engine with an electric motor. Really not much better. It took time for them to realize they wire the factory and have each device have its own electric motor. That was more efficient and more flexible. Technology that changes how you work takes a long time to adopt.
I just had a meeting yesterday when someone from the customer support team vibe-coded a solution in a few hours. The boss said, "Let's just give this as a gift; this product is not our focus and I want to show them how AI makes us work fast."
Junior developers will find it harder to be hired and trained. The case for lesser known artists and musicians is much worse. The scientific literature will be flooded by low quality AI slop with questionable veracity. Drafts of Good debut novels will be harder to find. When someone writes a love song, their romantic partner(s) will have to question if it was LLM generated. Nobody will be able to trust video footage of any kind and will have a much harder time telling what is the truth.
I don't think standard economic indicators are tuned to detect these externalities in the short to medium term.
I think most people will retreat into smaller spaces where they can rely on people to not deceive them. Everyone is moving to discord/group chats now for any sort of trustworthy information. This might be a good thing honestly. It was probably never good that we all got our information from the same place.
This. I think generative AI will mostly generate destruction. Not in the nuking cities sense, but in hollowing out institutions and social bonds, especially the complicated and large-scale kind that have enabled advanced civilization. In many ways, things will revert to a more primitive state: only really knowing people in your local vicinity (no making friends online, because it'll be mostly dead-internet bots out there), only really knowing the news you see yourself, more reliance on rumor and hearsay, removal of the ability for the little guy to challenge and disprove institutional propaganda (e.g. can't start a blog and put up some photos and have people believe your story about what happened), etc.
- "Thousands of CEOs just admitted AI had no impact on employment or productivity..." https://fortune.com/2026/02/17/ai-productivity-paradox-ceo-s...
- “Over 80% of companies report no productivity gains from AI…” https://www.tomshardware.com/tech-industry/artificial-intell...
But fundamentally, large shifts like this are like steering a super tanker, the effects take time to percolate through economies as large and diversified as the US. This is the Solow paradox / productivity paradox https://en.wikipedia.org/wiki/Productivity_paradox
> The term can refer to the more general disconnect between powerful computer technologies and weak productivity growththe same firms "predict sizable impacts" over the next three years
late 2025 was an inflection point for a lot of companies
There will be a period like we are in now where dramatic capability gain (like recent coding gains) take a while for people to adapt to, however, I think the change will be much faster. Even the speed of uptake in coding tools over the last 3 months has been faster than I predicted. I think we'll see other shifts like this in different sectors where it changes almost over a series of a few months.
That isn’t actually true though, right now everyone has a hard dependency on a cloud service. That is currently sold to them at deep discount by companies that are losing billions.
When the market eventually corrects it’ll be interesting to see how much AI ends up costing. At the very least it will be comparable to the broadband internet connection you mentioned. Possibly a whole lot more.
Isn't that a huge red flag? If customers are being given this product at a discount and it still isn't showing a positive ROI for them, what makes people think it will improve once we're charged full price?
Financially this feels similar to Uber's business plan in the 2010s; undercut the market with unsound pricing propped up by venture capital (PE was literally subsidising taxi fares; they admitted this and their intention to readjust, but no one seemed to care) then stop manipulating the market and allow fares to even out at (gasp) what it cost to get a cab before Uber.
The difference here is that the LLM market is human productivity; enormous subsidies are afforded to Anthropic, OpenAI etc. in the form of VC or compute credit, but eventually those debts will be called in, the free-to-use aspect will vanish because it's simply not profitable, and we'll be left with several premium products that only a few people will actually pay for, and even then that may not be enough to cover their costs. That's when the bubble will burst.
There’s the scenario where LLMs get more efficient in size, and to get 2026 SOTA performance you will be able to get it from consumer grade laptop.
Sure with a 1000B parameter you will get better performance but the average person will have it write some python script, not derive new physics equations.
So in a sense the demand for LLM intelligence with reach a plateau (arguably we are there today for avg person) so there will not be any subsidy required, because the avg person will not need the latest and greatest.
There’s not the same demand pattern for something like uber.
But isn't that bad for the AI companies, too? Because then people just run an ~2026 SOTA performance open source model on their laptop for free and not pay any subscription.
Regular folks will not pay Anthropic, but NSA, NASA or research labs might.
I’m not implying this will be a good time for AI companies. I am saying AI as a technology can provide value without it being controlled by only 3 companies.
$20 a month is a lot of money, I don't think the "convenience and flexibility" you get would actually be worth it, unless you've 1) got money to burn, 2) lack the skills to install software, 3) the open source community totally fails to develop a reasonable installer. The LLM service would probably be akin to a scam preying on ignorance, like those companies that will rent you a water softener for like $100/month.
That’s.. kinda the question.
And also that may be the case for Anthropic who have fewer free users, a large enterprise business, and less generous rate limits on their subscriptions. I don't know if OpenAI or Google have commented. I suspect OpenAI is in a worse position given their massive non-paying consumer base.
In the past 30 days I have burned $78.19 in API token costs with my $20/month Claude Pro subscription. In January I burnt over $300 in API token costs.
EDIT: also, the casual or gym-style members that pay every month but barely use the service are of course very valuable wrt margins
They're spending more than they're making. For the foreseeable future, saying "we could be profitable if we stopped training" if goofy, because they can't stop. If they do, no one will want to use their product because it will be overtaken by competitors within three months.
I get it that in 10 years all of this might peak and we're gonna be content using old models, but that'll be a very different landscape and Anthropic might not be a part of it anymore if they don't start making money before that.
I would personally be happy using gpt 5.3 codex for the foreseeable future, with just improvements in harnesses
IMO we're already at the point where even if these company collapse and the models end up being sold at the cost of inference (no new training), we would be massively ahead
Models are already super useful, but if you can make them more useful by burning cash people are willing to hand you, why not?
I'm pretty sure that in corpo-speak "inference" excludes the cost of datacenter construction, GPUs and other hardware, manual data cleaning, R&D, administration, etc - basically everything except the power bill for inference.
I have absolutely no problem with companies that run inference only - plenty of them offer open models as a service - they're usefull and their accounting can be believed... but they don't have near $ Trillion valuations and they don't misallocate capital on a vast scale as the frontier models do.
The point of the OP is that closed models don't pay for themselves and, on the scale of the US economy, they provide minuscule economic advantages compared to the enormous investments they consume.
So they are not profitable now & they have no idea of when they ever will be.
Worse, Gemini has guaranteed funding for continued training whenever the AI hype bubble pops.
Anthropic & OpenAI's only saving grace is that Google is generally terrible at product.
I was talking about Anthropic, but run rates don't need to go down, they just need to scale with revenue. For Anthropic specifically, this seems to already be the case.
OpenAI I don't know much about, but it would make sense if they were running at a terrible loss due to the ubiquity of free ChatGPT.
> Worse, Gemini has guaranteed funding for continued training whenever the AI hype bubble pops.
I don't see a scenario in which Anthropic has any problem financing their activity given their conversion rate of inputs to recurring revenue. Generally, bubbles popping means companies with bad balance sheets and bad economics die, but that just doesn't apply to Anthropic IMO.
OpenAI though, hard to say. They've lost all of the good will being the first mover gave them at this point, so they'll need to really lead product to make the economics work for them.
They're not losing billions on inference, they're losing billions in the arms race of training.
Effective use of these AI tools need high critical thinking skills which are in short supply.
But as the organization slowly learns and adapts I'm sure the capability gains will materialize.
"The Productivity Paradox" is what they called it when people were skeptical that computer would end up finding a place in the office. There are articles from the 90s complaining about how much people are spending on buying computers for no real impact on productivity https://dl.acm.org/doi/10.1145/163298.163309
Once confronted with reality we have a "productivity paradox"?
To my eyes, the problem is not the productivity gain arriving slowly, but the immediate draining of funding from virtually all other areas of innovation.
But they may get rich soon, which is all that really matters to them.
It's not that AI can't be useful, but that there's a learning curve, and early in the learning curve we should expect as many resources to be spent learning as resources are saved by using the thing. A macro level view of the economy as a whole sees this as "zero economic growth".
So if you want to think of it in economic terms, some software consulting firm that would otherwise have made six figures instead did not. The vast majority of the money we would have spent stayed in our pocket. Slight decreases like this in “velocity of money” no doubt add up to significant sums.
Is your company a software firm or considered something outside of pure software?
I needed and embedded document based database, a friend of mine with 30 years experience was vibe coding a database in Rust and I asked him if he can make it support Swift and be embedded in iOS and in few minutes he delivered that using Claude. Then I started vibe coding on it with Codex adding features I wanted and integrating it into my project. It worked as expected. I think it is close to reaching parity with MongDB, years of work vibe coded in a weekend.
There’s going to be fundamental changes in how we program computers and consequently the IT industry.
But on company level I see it as a risk: suddently you might have 50 new small apps created by people who might not even work at the company who are not constantly tested for security/privacy ... but more important who once done are not pushing the frontier of how a much better solution might be in that area cause nobody is putting time into them. So as time passes by this has the risk to become legacy software used to run your business. yes of course you can point an AI to all of them and prompt it to make them better but that means focus on that instead of your core business.
Maybe we will see solutions appearing to manage this kind of tech debt.
If this is happening on a widespread basis in the economy we should see evidence of it sometime this year and that's what investors are anticipating with SaaS stocks.
I am an economic dummie, but wouldn't the metric be revenue per employee?
It takes time for technology to show measurable impact in enormous economies. No reason why AI will be any different.
The iPhone killer UX + App store release can be directly traced to the growth in tech in the subsequent years its release.
We might have been possibly better of actually, with the Apple walled garden abominations and user device lockdowns not being dragged into the mainstream.
Not clear that this is a helpful interpretation, other than "we're in the primordial ooze stage and the thing that matters will be something none of the current players have", but that's hard to take to the bank :-)
Personally I think AI is unlikely to go the way of NFTs and it shows actual promise. What I'm much less convinced of is that it will prove valuable in a way that's even remotely within the same order of magnitude as the investments being pumped into it. The Internet didn't begin as a massive black hole sucking all the light out of the room for anything else before it really started showing commensurate ROI.
Most idiots like Columbus died in obscurity.
The thing I'm making with the APIs is very helpful to me, maybe it'll be helpful to others, who knows.
Right now the frontier AI companies are explicitly running a kind of chicken race - increasing the burn rates so much that it gets harder and harder. With the hopes that they (and not their competitor) will be the one left standing. Especially OpenAI and Antropic, but non-AI companies like Oracle have also joined. If they keep it going, the likely outcome is that one of them folds - and the other(s) reap the rewards.
Utility (per cost) will go up the tougher the competition. Money captured by single entity possibly down with increased competition.
I think there are two layers of uncertainty here. One is, as you say, if the value is worth the investment. The other and possibly bigger issue is who is going to capture the value and how.
Assuming AI turns out to be wildly valuable, I'm not at all convinced that at the end of this money spending race that the companies pouring many billions of dollars into commercial LLMs are going to end up notably ahead of open models that are running the race on the cheap by drafting behind the "frontier" models.
For now the frontier models can stay ahead by burning heaps of money but if/when progress slows toward a limit whatever lead they have is going to quickly evaporate.
At some point I suspect some ugly legal battles as some attempt to construct some sort of moat that doesn't automatically drain after a few months of slowed progress. Google's recent complaining about people distilling gemini could be an early signal of this.
I have no idea how any of that would shake out legally, but I have a hard time sympathizing with commercial LLM providers (who slurped up most existing human knowledge without permission) if/when they start to get upset about people ripping them off.
It's not good enough to just say oreo ceos say we need to more oreos.
There's a real grey area where these tools are useful in some capacity, and in that confusion we're spending billions. Too may people are saying too conflicting things and chaos is never good for clear long-term growth.
Either that 20 years is completelly inapplicable to AI, or we're in for a world of hurt. There's no in between given the kinds of bets that have been made.
They don’t have time to wait for all the companies to pick up use of AI tooling in their own pace.
So they lie and try to manufacture demand. Well demand is there but they have to manufacture FOMO so that demand materializes now and not in 20 or 10 years.
The AI use of GPUs didn’t stem from a glut of outdated, discarded units with nearly no market value. All of those old discarded GPUs were, and still are, worthless digital refuse.
The closest analog i can think of to what you’re referring to is cluster computing with old commodity PCs that got companies like Google and Hotmail off the ground… for a few years until they could afford big boy servers and now all of those, and most current PCs on the verge of obsolescence, are also worthless digital refuse.
The big difference is that Google et al chose those PC clusters because they were cheap, commodity pieces right off-the-bat, not because they were narrowly scoped specialty hardware pieces that collectively cost hundreds of billions of dollars.
Your supposition fails to account for our history with hardware in any reasonable way.
This isn’t a normal tech expenditure— the scale of this threatens the economy in a serious way if they get it wrong. That’s 401ks, IRAs, pension plans, houses foreclosed on, jobs lost, surgeries skipped… if we took a tiny fraction of this race-to-hypeland and put towards childhood food insecurity, we could be living in a fundamentally different looking society. The big takeaway from this whole ordeal has nothing to do with semiconductors — it is that rich guys playing with other people’s money singularly focused on becoming king of the hill are still terrible stewards of our financial system.
Divorcing research from "learning by doing" is a recipe for a bureaucratic ivory tower. If you only funnel money into pure research without the messy, expensive, and often "wasteful" reality of large-scale deployment, you end up with an economy of academic metrics rather than industrial power.
The most damning evidence against the "research-only" model is the birth of the Transformer architecture. It did not emerge from an ivory tower funded by bureaucratic grants or academic peer-review cycles; it was forged in the fires of industrial practice.
History shows that a fixation on immediate social utility or "rational" cost analysis can be a strategic trap. During the same era, Qing Dynasty bureaucrats employed your exact logic, arguing that the astronomical costs of industrialization and rail were a waste of resources better spent elsewhere. By prioritizing short-term stability over "expensive" technological leaps, they missed the industrial window entirely. Two decades later, they faced an industrialized Japan in 1894 and suffered a total collapse. The "waste" of one generation is frequently the essential infrastructure of the next.
To be fair, it isn't necessarily the same people doing both at once. Sometimes there are two groups under the same general banner, where one makes the big-claims, and another responds to perceived criticism of their lesser-claim.
An even bigger problem is that people listen to them even after they say rationally implausible things. When even Yann LeCunn is putting his arms up and saying "this approach won't work," it's pretty bad.
whether or not these companies can turn a profit - time will tell. but I am betting that our massively profitable companies (which are biggest spenders of course) perhaps know what they are doing and just maybe they should get the benefit of the doubt until they are proven wrong. but if I had to make a wager and on one side I have google, microsoft, amazon, meta... and on the other side I have bunch of AI bubble people with a bunch of time to predict a "crash" I'd put my money on the former...
quite the opposite is happening as evidenced from last earnings reports…
>Many observers disagree that any meaningful "productivity paradox" exists and others, while acknowledging the disconnect between IT capacity and spending, view it less as a paradox than a series of unwarranted assumptions about the impact of technology on productivity. In the latter view, this disconnect is emblematic of our need to understand and do a better job of deploying the technology that becomes available to us rather than an arcane paradox that by its nature is difficult to unravel.
Today you have to be blind to not see the change that is coming.
World has its own (massive) inertia, burocracy present in businesses accounting for a big part in it.
AI itself is moving fast but not at infinite speeds. We start to have good enough tooling but it's not yet available to everyone and it still hangs on too many hacks that will need to crystalize. People have a lot of mess to sort out in their projects to start taking full advantage of AI tooling - in general everybody has to do bottom up cleanup and documentation of all their projects, setup skills and whatnot and that's assuming their corp is ok with it, not blocking it and "using ai" doesn't mean that "you can copy paste code to/from copilot 365".
As people say - something changed around Dec/Jan. We're only now going to start seeing noticable changes and changes themselves will start speeding up as well. But it all takes time.
the change that is coming.
Why? It’s descriptive of the “past”. While you’re trying to predict the near/far “future” and project your assumptions. Two different things.
If you replace OpenClaw with any number of other hot LLM products/projects, I’ve been hearing that same exact sentiment for numerous 6-to-12-month periods. I’d argue we have no idea how long it’s doing to be, but it’s probably not very soon.
Or will AGI build the fusion energy
Things are actually slowing down. And society will still see AI adding little to next years report. The costs still outweigh the benefits.
Yes, Anthropic decided they wanted to IPO and got the hype machine in full swing.
Don’t get me wrong LLMs are here to stay but how we’re currently using them is likely going to change a lot. Stuff like this:
> in general everybody has to do bottom up cleanup and documentation of all their projects, setup skills and whatnot and that's assuming their corp is ok with it, not blocking it
Is not needed to get a lot out of AI, and is mostly snake oil. Integrating them with actionable feedback is, but that takes a lot of time and rethinking of some existing systems.
I don’t like the Internet analogy cause that’s like producing a new raw material, but AI is gonna be like Excel eventually (one of the most important pieces of software in the world).
Adult life doesn’t have to be boring drudgery, you know. I mean, it mostly is, but the rare moments of childlike joy and excitement are some of the best parts.
As far as the putting a damper on anything, nope it doesn’t. And it never will.
The people excited about AI are excited because of the impacts they see on their own jobs and daily lives. We don’t care what Goldman Sachs has to say about productivity.
Someone I deeply respect, Clifford Stollm wrote a book called “Silicon Snake Oil — Second thoughts on the information highway" in 1995. And while he was and is a brilliant person, Stoll was wrong.
Smart people are terrible at predicting the most consequential changes in our future – even when they're familiar with the technology. I wrote a bit about my thesis why here, https://1517.substack.com/p/inside-v-outside-context-problem...
Don't make his mistake. Don't look away from the change being wrought. The world has changed and our history now has a new, sharp dividing chapter "Before ChatGPT | After ChatGPT"
and that chapter will go down right next to "Before Trinity | After Trinity"; "Before PC | After PC"; "Before 'Internet' | After 'Internet'"†
† Yes, I know I'm referring to the Web. But we're still using the dark fiber from the .com boom.
"On top of that, there is currently no reliable way to accurately measure how AI use among businesses and consumers contributes to economic growth."
No doubt people are using it work ( https://www.gallup.com/workplace/701195/frequent-workplace-c... ) the question is how much productivity results and to whom does it accrue.
Partially this is AI capability (both today and in the past), partially this is people taking time to change their tools.
See his interview in Dwarkesh's podcast: https://www.youtube.com/watch?v=c0-0gGdDJyE&t=4983s
I'm genuinely not sure. We are all computer people in this forum, so it may have improved our lives. But for many people, information technology has lessened the time spent in a given week or year on activities they find meaningful.
https://www.washingtonpost.com/technology/2026/02/23/ai-econ...
For example when there are large scale layoffs attributed to AI (which in some cases may be a smokescreen to reduce headcount), the people that are affected reduce their spending to compensate.
Less spending means less revenue for other companies, resulting in more layoffs, something of a feedback loop?Therefore perceived productivity gains maybe a net negative in the macro economic sense.
I am no economist so I don't know if any of that makes sense, but it's something I regularly wonder about.
Now it's adding "Basically Zero" to US Economic Growth
https://gizmodo.com/wall-street-apparently-believes-the-futu... This one's from bloomberg
Does Gizmodo just repeat random stuff they hear like some kind of stochastic click generating parrot? Low GDP this quarter is due to low government spendint. Hardware is still extremely cheap and isn't getting better much faster. America has taken a major lead by buying out all the fabtime. If I ran Taiwan I would have nationalized TSMC, horded all the chips and allied with Japan and SK for a joint AI and defense pact.
The key innovation of the LLM beyond anything technical is that, across the board, government can now see the potential of ML and "AGI". The current American left (and Europeans leaders) is absolutely hopeless, while they make "not one drop for data signs" we are quickly moving towards a future where your job is creating data mostly by doing whatever you want. 10 years ago data was worth 1/9th it's current worth, in 10 years it will be at least 9x it's current worth today. 22+% year on year for the last 14 years 50+% during lockdown. It's strangely hard for people to imagine that work becomes easier over time even though it's always been true. There is a huge difference between letting people capture the value of their labor and preventing that value from being created in the first place.
We need to get past the hype first and let the cash grabbers crash.
After that, with a clear mind we can finally think about engineering this technology in a sane and useful way.
Are we saying that llm's have zero economic growth, or are we saying that the sum of winners and losers in llm usage are zero or less than zero?
I think there's many examples of llms resulting in winners, or maybe this signal is just very high in the tech space.
But maybe there's not enough reporting on the losers in llms at the moment? (E.g. did llm displace their jobs, they have llm use cases that failed, etc, etc)
These opinions masquerading as statistics is why governments create departments to publish trustworthy statistics.
On real projects and real teams you can’t do this. If you did what you did in your example you’d say log 8hrs of work, right? Your team lead will ask a simple question: “did you write this code or was it AI-assisted? and what exactly here took 8hrs?” so you could do this once and 2nd time you’d be changing the status on linkedin to looking for work
With all this recent Claw stuff, it's weird that as people who should be championing the opposite due to our field of study or industry, some of us are now pushing a method of automation that is akin to robo vaccums randomly tracking dogshit across the carpet.
In my working environment, people get dressed down for repeatedly communicating incorrect information. If they do it repeatedly in an automated fashion they will be publically shamed if they are senior enough.
I have no idea what benefit a human-in-loop for sending an automatically generated emails or agent generated sdks or buliding blocks has when there is no guarentee or even a probability of correctness attached to the result. The effort for vaildating and editing a generated email can be equally or greater than manually writing a regular email let alone one of certain complexity or significance.
And what do we do to create to try to guarentee a semblance of correctness? We add another layer of automated validation performed by, you guessed it, the same crew of wacky fuzzy operators that can inject correct sounding gibberish or business workflows at any moment.
It's almost like trying to build a house of cards faster than the speed with which it is collapsing. There seems to be a morbid fascination among even the best of us with how far things can be taken until this way forward leads to some indisputable catastrophe.
Is it possible that this sort of problem will be fixed? Hypothetically, what would happen in a scenario where one of these apps can do in 1 hr the work that would take a developer a month, reliably? Or is your premise that will NEVER happen?
And yeah, blah blah they burn money blah blah. Check Anthropic CEO interviews. He openly describe the balance problem : - cost of training a new model - newly built infra ratio of training vs inference - market adoption, that is despite extremely quick is not unlimited, since even market is not unlimited.
essentially it's a tricky balance, between you do not invest today you will loose tomorrow vs you invest too much and go bankrupt next year.
But now we have something else happening. It's hard to find an application for something that makes a lot of mistakes. That's not the same issue. The issue then was that no one had written the software yet. Everyone knew what software needed writing. The future was obvious. Here, not so much. We can't see how to make it not make mistakes.
We have to hope someone will come up with a solution to that. Otherwise their big bets on something non-productive won't pan out the same way that the computer did, and we're all going to suffer for it.
Buy buy buy buy.
We don't even have enough data centers.
2. Also a good portion of contribution of AI is where it's not taken into account by metrics like GDP. I have seen an explosion of FOSS projects especially in my own area and I'm sure it's pretty much are the case for many other areas.
3. Also there will be a sink-like effect with effects of AI on net wealth production. Like Earth's own oxygenation event which took billions for the produced oxygen to (after saturating all the sinks like Earth' iron reserves and turning them to Iron ore) ultimately find its way into the atmosphere.
4. Also I'm not sure what is exactly being counted on as contribution of AI to economy. Is the data-center build-outs and growth of chip companies,etc. included in this metric in AI's favor? ...
When companies can no longer afford to just keep running AI data centers at a loss, we will suddenly have a lot more data centers than we need, who will benefit from these? Who could have use for the hardware for other purposes?
I have no doubt that people will use this to axe grind about they think AI is dumb in general, but I feel like that misses the point that this is mostly about data center construction contributing to GDP.
Economists and businesses are calling BS and saying AI is cool, but basically adding zero measurable value with 95% of AI projects failing.
The truth is likely somewhere in the middle, but it seems unlikely this bubble can continue much longer.
And most jobs that can be automated already has been automated using traditional software.
Having a higher-paid, qualified employee supervising multiple AIs as the human only needs to spot for mistakes - maybe.
> If AI can't do 100% of a job then you can't remove the job.
I'm not sure if LLMs will change that or not