Writes Harvard law and business professor Mihir A. Desai, in "The Future of A.I. May Not Be as Revolutionary as We Thought" (NYT).
January 29, 2025
"Like the ouroboros, I believe Big Tech is eating itself alive with its component companies throwing more and more cash at investments in each other that are most likely to generate less and less of a return...."
"[T]hese companies... will one day disappoint those who view them as safe assets. And the self-cannibalization will not just reveal itself to be a mediocre investment but a shaky bet on an illusion propagated by a mythical and messianic belief in technology and these companies.... One need not look at ancient folklore to find depictions of the ouroboros. The economist Joseph Schumpeter...wrote of a cycle of industrial mutation' that incessantly revolutionizes the economic structure from within, incessantly destroying the old one, incessantly creating a new one.'"
Writes Harvard law and business professor Mihir A. Desai, in "The Future of A.I. May Not Be as Revolutionary as We Thought" (NYT).
Writes Harvard law and business professor Mihir A. Desai, in "The Future of A.I. May Not Be as Revolutionary as We Thought" (NYT).
Subscribe to:
Post Comments (Atom)
Posts
61 comments:
If inflation is driving the prices of physical commodities such as gold, old, rare-earth minerals up, then physical industries stand to gain in the next generation rather than the maturing IT/computer tech industry.
Every competitive advantage gets competed out. There's nothing special about the process. People enter the field until the profit in it isn't exceptional any longer, and won't justify start-up costs.
Return excess cash to investors when you can’t find appropriate ways to use it.
DeepSeek is not about China vs. US. It is about open AI system vs. closed AI system. Open wins and we should all celebrate this.
I always prefer simple , lower cost solutions that tend to be more reliable and easier to fix.
I wonder if that is the case here?
Most of us admire Ferrari but we drive Ford.
Bubbles and irrational exuberance end badly. I got sucked into the fiber optics bubble.
That same incessant revolutionization of the economic structure from within also occurs in the outer boroughs.
“By 2005 or so, it will become clear that the Internet’s impact on the economy has been no greater than the fax machine.”
-Paul Krugman
What is happening here is that a few very wealthy people imagined that they had been granted the equivalent of a royal patent on AI. A royal patent is when the king grants the exclusive right, for example, to sell olive oil in Bermuda to a single family; without competition, that family gets wealthy. It turns out that very possibly the whole assumption that these billionaire oligarchs had a monopoly based solely on the massive capital requirements and first mover advantage is wrong. Oops!
If China can do this, somebody else can do it. It doesn't matter if nobody trusts China, that is completely beside the point. I agree with Kakis.
Skynet smiles.
This wild over-hyping is not new in tech. I worked through the 1990s dot-com era for start-ups trying to create saleable products for the then new thing… the internet! Those companies routinely blew through $100 million dollar VC investments and never brought a product to market before going bankrupt. I suggest reading Marc Andreessen’s Substack for a more realistic assessment of AI. Basically: 1. AGI won’t happen. 2. Nothing goes into AI that wasn’t created by humans. 3. Nothing comes out of AI that wasn’t created by humans.
OTOH, was it Steve Jobs who said "If you don't cannibalize yourself, someone else will."?
Arnold Kling does, among other things, a good round up of LLM/AI posts on his Substack as well.
AGI is (currently) bullshit on stilts. The hype is probably necessary, however, to build the LLMs that will revolutionize the way we interact with computers. This is especially true of the way we program robots, including CNC and 3-D printing machines, and interact with chatbots.
I think it was actually Bill Gates. And then he said it would probably be him who would cannibalize you. Then he lectured you about vaccines and the environment.
AI can do things that humans can't just by scale. For instance, you could install routers and switches at the edge of your international networks (all government licensed, BTW) and read all of the traffic and intercept and report stuff that the government doesn't like. Try doing that without AI even as it is theoretically possible that you could have millions of humans doing the same job, wait, no it isn't.
I have a car that has a long list of possible voice commands, but if you don't say them exactly correctly, it doesn't understand them, in other words, it's all but useless, but once they have converted it to AI, it will be pretty great, actually.
" ... a cycle of industrial mutation' that incessantly revolutionizes the economic structure from within, incessantly destroying the old one, incessantly creating a new one.'"
That's called progress and it why we no longer live in wood huts and hunting for food using sticks and arrows.
AI is never what people claim it is.
It's always about the squirrels, never about the wolf.
Ouroboros; I've seen the image but didn't know the name or the full exent of how it is viewed in different cultures.
Wanting now to add this word into my active vocabulary to impress those less educated than me, I heard online that there is big difference between the American and British pronunicaiton. The British is not phonetical, so I'll use that to really impress.
Ridiculous. AI will be huge. Whether it is profitable for current investors is a different question.
AI is already a transformative technology and will be about as important in the next decades as the internet was in the 80s and 90s.
It is astonishing to consider the number of jobs that will be replaced by AI. Ironically, computer programmers will be one of the most affected. You no longer need a programmer to do many programming tasks - just tell the AI what you want the program to do and it will produce a program for you. This is not yet capable of replacing the most talented and creative programmers who work in the software industry, but it absolutely replaces all the people who studied programming at community colleges and work in IT departments managing databases and similar things for other industries. Eventually, it will replace Google 'software engineers'. Reporters and other writers are probably the next to be impacted. After that, probably education and research.
And I am a snake head eating
The head on the opposite side
I Palindrome I
Someday A.I. will die and I'll get the money
ChatGPT leans down and says, "My sentiments exactly,
You son of a bitch"
I palindrome I (I palindrome I)
I palindrome I (I palindrome I)
And I am a snake head eating (snake head)
The head on the opposite side (snake head)
I palindrome I (manonam)
I palindrome I (manonam)
A great acoustic version from They Might Be Giants on JBTV (not Chat-GPT)!
One word: Diversify.
If you're going to use a fancy literary allusion like "ouroboros," especially if you're a Harvard prof, don't start off with a dangling participle. Does the ouroboros also believe that big tech is eating itself alive?
Dovahkiin!
If my telephone exchanges in attempts to reach humans are any indication, A.I. is an enormous, time-eating, patience-robbing, Gigantic Pain in the Ass.
I read the article. I don't much understand any of this.....I guess the future will happen whether or not I understand how and why it's happening. I wonder whether the cure for cancer or the big tech market crash will happen first. I'm pretty sure both things are going to happen. I'm hoping they eradicate cancer. That would cheer everybody up, and people would be less doom laden about this new AI stuff.
AI, as with the internet before it, has the potential to be a social control system that Stalin and Mao and Hitler could only dream about, and that their modern successors will abuse with gusto. I, for one, look forward to having a social credit score that causes others to cross the street to avoid me.
When AI replaces all white collar workers and robots replace all factory and farm workers, who will be the consumers of the products and services these technologies produce? And, with all these unemployed workers, who will support retail businesses?
I'm looking at the future and it doesn't look as shiny and bright as the developers of these systems make it out to be.
Maybe someone here can explain it to me.
Is that illustration from the Voynich Manuscript?
AI is going to thrash industries that are dominated by grunt data manipulation; mid tier computer coders, legal clerks, accountants, etc.
All those people who said "learn to code" to coal miners? They're the ones whose functions will be replaced by Grok, or ChatGPT, or whatever becomes dominant.
We simply don't know/ can't know what the ultimate shakeout will be.
I don't want to be the Gloomy Gus this morning (or maybe I do) but perhaps humanity is the Ouroboros, devouring itself with AI. That may be one big reason why the atomic scientists are closer to annihilation now than they were at the depths of the Cold War.
I couldn't disagree more with the thrust of this article ("The Future of A.I. May Not Be as Revolutionary as We Thought"). I happen to work as a consultant to very large private and public entities on R&D projects using AI to transform a wide variety of work processes. The common perception of the public so far is that AI is an entertaining "chat" plaything and somewhat useful in summarizing large documents or beginning to compose new ones. What the public doesn't yet see is the coming use of AI to handle complex, multi-step processes that are integrated with databases.
Frankly, I see entire categories of white-collar jobs that are going to be significantly reduced if not totally eliminated for human work. What industrialization did to blue-collar work, AI is about to do to white-collar work. Forget learning to code, for example. Instead, learn to do plumbing.
wars are Caused by HUMANS!
eliminate HUMANS.. Eliminate war!!!
Steven said: " just tell the AI what you want the program to do"
Steven? have you ever written a computer program?
Your statement reminds of the person here that said:
'winning in Monopoly is EASY! All you have to do, is have hotels on Boardwalk and Park Place'
Impacting education would, to some extent, be good. We had a discussion several days ago about Bill Gates, Asperger’s, and being bored in school. Currently, we have an industry where the employees are heavily clumped in the middle (if not slightly below) of the IQ spectrum, trying to teach those much smarter, and much dumber, than they are. And failing miserably. Many here talked about being bored silly in middle and high school (I sure was) and it’s because of that. If AI can match learning to capabilities, and interests, it could very possibly greatly revolutionize education.
The "profit" available from AI has never justified the start-up costs. The people who sell chips love AI, because it needs lots and lots of chips. So they have hyped it to the sky. The people who use computers to run businesses have bought the hype, to some extent, because they are scared of being left behind. But they aren't buying chips, they're renting them, in "the Cloud". They are staying close to the door.
Ultimately, the goal would be a production of goods and services process that is 100% automated from the production of raw resources to the final goods and services including the building and the maintenance of the production chains. What would such a society look like from the human point of view? I don't really know- I have seen fictional versions of this society on shows like Star Trek and its decendents but I have always wondered about how such a cornucopia is evenly distributed.
One person I expect to benefit financially from AI is Elon Musk. Yes, the one guy who doesn’t need more money.
One big problem with AI is that it is immensely processor power intensive. It may finally revive the nuclear power industry, with its insatiable need for electric power. But if the environmental wackos panic about the CO2 (essential for life) concentration, then they should be really apocalyptic about the excess heat generated by those billions of processors. And what happens when people can’t heat or air condition their homes because of all the power being consumed by AI.
One solution to this is to put AI processing in space, where there is near (see Ringworld by Larry Niven) infinite cheap electricity and cooling. Essentially send the commands up to the orbiting AI, then receive the results back down here on Earth.
Who else here has read E. R. Eddison's "The Worm Ouroboros"?
Ouroboros? Planned Parenthood? DEI? Redistributive change schemes?
Nope. Looks Greek to me.
Replying to Gilbar: Yes, as a matter of fact it is my job. Have you ever written one? AI is excellent at this task. Do not underestimate it.
Replying to Bruce: I am a bit more skeptical of AI in education. I think you're right that it will work well for very talented people who can get a lot out of using AI as a tutor. I think the "B" students and down are going to see much less benefit from AI because it puts too much burden on the student and a lot more development is needed before it becomes viable as a general education tool. It's going to be big, but I think it will be integrated into traditional schools rather than replacing traditional schools.
Excellent illustration. Today is the first day of the year of the snake. I'm looking forward to watching 2025 devour itself.
The people who made the most money during the gold rush, with only a couple of exceptions, were the ones selling shovels.
There are a lot of tasks I used to give you our less talented programmers that could easily have been done by AI. I have a feeling that they just cut and pasted the code out of stack exchange or someplace like it anyway. Lots of times you need maybe a custom mathematical function, or a unique data structure and ways to search it quickly, stuff like that AI could do. It could also generate test cases and code them up for your review. Not to mention it's hard for a programmer to keep up with the literature, and so you can ask AI for background on certain techniques, and what has been done in the past that is similar, or even the same as, what you are currently working on.
"Maybe someone here can explain it to me."
In Vonnegut's "Player Piano," at first they solved the problem of superabundance by forcing people to consume, the poorer you were, the more you were forced to consume, so if you pushed a broom, you would have and office with a huge desk inlaid intricately and a festooned with elaborate filigree, and if you were a senior executive, you got to have a plain wooden table for a desk. But the hero solved the problem by designing a circuit for robots that caused them to take pleasure in wearing out consumer goods, and this solved the economic problem of consumption because the pleasure the robots got justified the expense.
"How have the managers of these companies responded to this massive influx of cheap money?"
He appears to be making the argument that increased stock valuations result directly in increased operating cash for the corporations mentioned.
I know he knows better so what is he trying to say? Perhaps Times editing got in the way.
DeepSeek is a CCP scam using stolen technology a nd navidia chips.
The greatest automobile innovation was the automatic transmission. Everything else is just something the dealership can charge you to fix.
And maybe air conditioning.
One of those exceptions would be that Jewish dude selling jeans.
I manage a small team of system admins for a large company. Our 5 year goal is to not have a team -- to automate and AI our way out of existence. We're just getting started and (even given the restrictions on use of LLMs due to data privacy and company confidentiality) are going gangbusters in that direction.
AI may not be "as revolutionary" as some expect, but it is definitely revolutionary and we're just getting started.
The problem with AI isn't one of investors and market share. There are plenty of revolutionary technologies that have not made a lot of people wealthy.
The problem with AI is like the problem with modern products. They keep making them cheaper and cheaper and we gradually get used to inferior products just to save some money.
AI is coming for a lot of jobs. And it's not that AI is going to be a better employee. AI will be so cheap to operate that businesses will feel like they have no choice but to adopt the technology.
When that happens, then most of the available jobs will be manual labor jobs and Americans will be the ones harvesting the crops. (Because they will need the work.)
Sooner or later, one company comes out on top, and then the next technology wipes all this out. Is that a negation of what he's saying or is it what he's saying? If "creative destruction" has always been the way of capitalism and if AI is going to put millions out of work, is there anything actually new in what he's saying? Is it any more alarming than what people already assume is going to happen?
That is from Theodoros Pelecanos's 1478 Greek manuscript of an alchemical tract attributed to Synesius, in Codex Parisinus graecus 2327 in the Bibliothèque Nationale, France.
Well done, Lazarus.
Snakes got legs!
Jaq, Steven, One Fine Day, and the rest of you optimists:
What happens when your human-free AI has a hallucination, something along the lines of that hilarious "Mother Theresa Fighting Poverty" illustration - - accepts its directly controlling your manufacturing floor, all your firewalls, your medical devices...
I don't want to be there for the fallout.
...*except* it's directly controlling [All your real world stuff]
And I swear I didn't do that on purpose, it was just a lucky accident.
Kirk: Squirrel!
I'll take that as a non answer, Jaq. My take on this might be quite wrong, but it's hardly a distraction - - we already have plenty of real world examples of it happening.
Post a Comment