is AI gonna take our jobs?

[ahalpert]

A bubble 35 years ago is not a trend. History yes. Predictor? Maybe. Trend no.

Your responses make me wonder if you even read his article. All that dude does it look at trends and try to explain them, as in the graph of technological innovation-driven productivity increases. The underlying point is that the bubble psychology of the first tech stocks boom looks similar to what's happening today. Stock market valuations reflect group psychology more than anything else.

Okay, that's your guess. Got it, you're staying out. Everybody should do what's right for them.

Again, my interest in the discussion is not primarily buy/sell. It's what's going to happen to our jobs. And therefore - are the breathless predictions of AI salesmen true?

But my personal investment path as of now is just to buy the s&p. I make half my annual buy by year's end and half by tax day (in accordance with solo 401k deadlines). And I don't worry about what's happening in the market.

Those questions don't concern me. Fundamentals are progress measurements for energy, agriculture, manufacturing, transportation, not picking winners and losers.

Productivity growth is fundamentals. And you talked about buying NVIDIA. And you act like Krugman is giving crappy stock picking advice. So...seems like you are concerned.

That shouldn't surprise anyone. Internet was a defense project, funding development of ARPANET, for sharing information, connecting universities, TCP/IP and email, commercialization in the '80s, web, browser, search, e-commerce and social media in the '90s, 2000's on broadband, mobile internet, Wi-Fi, and social media platforms. You said, late 19th century. Stay at home connectivity doesn't build anything except itself! Are you really comparing the tech contributions of social media and wifi to the nation building that took place during Vanderbilt, Carnegie, Rockefeller, J.P. Morgan? Electrification, Steel, Energy, the assembly line, transportation, aviation? When Obama said, "You didn't build that, somebody else built that," he was right and he was talking to you. The internet is nothing without the basic industries to power it.

I'm comparing it because of the psychology. We were told the internet would change everything. In some ways it did. But the economy has undergone much bigger and faster changes in the hundred years before we were born. Now again we're being told that AI will change everything. So what an economist would ask is what will that mean in terms of productivity growth and GDP.

Stock valuations and bubble development are separate issues from economic development.

 

[Tom Roper]

Your responses make me wonder if you even read his article. All that dude does it look at trends and try to explain them, as in the graph of technological innovation-driven productivity increases. The underlying point is that the bubble psychology of the first tech stocks boom looks similar to what's happening today.

I read what you excerpted. I assume you didn't misquote him. He compared the AI frenzy to the frenzy over the internet 25 years ago. So 25 years of "frenzy" begets bubble psychology for an answer? Brilliant, I'll have a Guiness. Every single business day on the stock exchange is some kind of frenzy. It's a live marketplace. Every single transaction is a reaction to something.

The point of trends is to provide insights that help us make predictions about the future. Looking backward 25 years to point out the obvious, that there will be another flood, another earthquake is just dumb.

 

[ahalpert]

I read what you excerpted. I assume you didn't misquote him. He compared the AI frenzy to the frenzy over the internet 25 years ago. So 25 years of "frenzy" begets bubble psychology for an answer? Brilliant, I'll have a Guiness. Every single business day on the stock exchange is some kind of frenzy. It's a live marketplace. Every single transaction is a reaction to something.

The point of trends is to provide insights that help us make predictions about the future. Looking backward 25 years to point out the obvious, that there will be another flood, another earthquake is just dumb.

This feels pretty clear and obvious to me. The point is that these two tech bubbles bear a lot in common, from the marketing messaging to the scale of the build-up to the fact that investors don't really understand what they're buying into. The parallel is much closer to the first tech bubble than to any other "frenzy" you care to point to, whether it's everyday trading or the real estate bubble of the '00s. And understanding the first wave of technological change and market dynamics may be useful for understanding what's happening now, along with understanding the differences. Especially when not everyone is admitting that this is a bubble. AI salesmen like Altman or Musk will tell you that AGI is on the horizon. Altman says we should all get UBI plus shares in tech companies and that no one will have to work. ("techno-feudalism")

Historical perspective is useful. Most people don't have any. Or the experience to analyze macroeconomic data.

Unrelatedly, here's a hit on Anthropic rejecting the use of AI in its job applications:

"AI company Anthropic’s ironic warning to job candidates: ‘Please do not use AI’"

“While we encourage people to use AI systems during their role to help them work faster and more effectively, please do not use AI assistants during the application process,” the policy reads. “We want to understand your personal interest in Anthropic without mediation through an AI system, and we also want to evaluate your non-AI-assisted communication skills.”

AI company Anthropic’s ironic warning to job candidates: 'Please do not use AI' | Fortune

The tech juggernaut wants to field communication skills without help from tech, and Anthropic isn’t the only employer pushing back.

fortune.com

Oh, last thing I was gonna add was in terms of personal investing thoughts I was thinking about hedging against American tech decline by investing in an ETF of Chinese tech stocks. Not decided yet though.

 

[ahalpert]

Google faked Gemini AI output in Super Bowl ad

Gemini didn’t write that incorrect Gouda stat at all.

www.theverge.com

"Google faked Gemini AI output in Super Bowl ad"

This doesn't surprise me one bit since I recently edited an ad for an AI service in which much of the "AI-generated" content within was not generated by AI or was not generated by their service but by a different AI service. These people are such liars.

 

[ahalpert]

Tom, another perspective on the bubble question. Click through for a few more posts in his thread:

https://twitter.com/x/status/1887861216933994715

 

[Tom Roper]

The person making this statement seems to believe that, in response to concerns about innovation, there's a rush to invest in AI, which could be driven by hype rather than real, sustainable growth. A "wild bubble" suggests that investors might be overvaluing AI companies, expecting huge returns without the underlying innovation to back it up. If it is or isn't should make no difference, don't put all your chips on one square, or at least don't leave them there, take some gains while they are ahead and move on, don't wait for today's unknowns to turn into tomorrow's problems. Be ahead of the burst is the safe play but the last one to sell before it happens is the biggest winner. Is it different with any other gamble?

 

[ahalpert]

The person making this statement seems to believe that, in response to concerns about innovation, there's a rush to invest in AI, which could be driven by hype rather than real, sustainable growth. A "wild bubble" suggests that investors might be overvaluing AI companies, expecting huge returns without the underlying innovation to back it up. If it is or isn't should make no difference, don't put all your chips on one square, or at least don't leave them there, take some gains while they are ahead and move on, don't wait for today's unknowns to turn into tomorrow's problems. Be ahead of the burst is the safe play but the last one to sell before it happens is the biggest winner. Is it different with any other gamble?

Well, again, I'm not interested in timing the bubble. I'm curious about the future of our economy and our industry. AI salesmen like Altman are saying that AGI (artificial general intelligence, in other words actual artificial intelligence) and even ASI (artificial super intelligence) are just around the corner. If that were true then tech stocks are, if anything, undervalued. And my job would be months away from obsolescence. But I don't believe that at all.

That guy's point is purely psychological. He sees the vitriolic reaction of the captains of industry against an op-ed by a bureaucrat who isn't even in office anymore as a sign that they're bluffing. If they were really holding a royal flush then it wouldn't matter what anyone else said or thought.

 

[ahalpert]

This is the crux of the issue: AI can't or won't generate images of left-handed people writing. Because the vast majority of images that it's trained on show right-handed writers.

Now, if the machine were actually intelligent then it would be easy for it to infer from what it has seen of right-handed writers and simply generate a mirror image. And it should be easy for an intelligent machine to understand what a left-handed writer is even if it's never seen one.

This is all you need to know about the current state of AI. It's not intelligent. It's a really fancy photocopier.

Indian Prime Minister Says AI Can't Generate Images of Left-Handed People Writing

Is that true?

petapixel.com

 

[Tom Roper]

This is all you need to know about the current state of AI. It's not intelligent.

The only gain from shutting the door on knowledge is a closed mind.

 

[NorBro]

oh no he di'int

 

[ahalpert]

The only gain from shutting the door on knowledge is a closed mind.

Hey man I use AI all the time. I'm just trying to predict the future by understanding the present

 

[ahalpert]

"First Legal Ruling on AI, Copyright, and Training Data Goes the Way of Creators"

First Legal Ruling on AI, Copyright, and Training Data Goes the Way of Creators

This case will be cited in future legal arguments.

petapixel.com

^ this is a limited ruling regarding a law brief AI search engine or something

"Adobe’s AI Video Generator is Nowhere Near Ready to Charge $30 a Month"

Adobe's AI Video Generator is Nowhere Near Ready to Charge $30 a Month

Adobe's newly-available Firefly text-to-video generator is nowhere near good enough good enough to warrant any price.

petapixel.com

^ dunno if this is true, I'm just reporting the headline

 

[Bassman2003]

Very good article. They tried it and the videos were horrible. Yet Adobe still wants to charge for it! This does not mean all AI is or will be this bad, but not the best optics when the largest player in the digital software imaging space has this type of faceplant.

 

[NorBro]

I didn't read the article, but one thing I've learned from filming/editing hundreds of case study videos in my career for very techy companies no one's ever heard of is that big companies, even with all of their money, may rely on many others (more than one may assume) for their overall success.

Of course at some point you're so big that you do 80-90%+ in house (took Apple almost half a century for the M-series) and you have everyone and mostly everything you need - but in this case, this isn't really Adobe's specialty.

I'm sure it will be at some point, but this AI video stuff (like the good kind) has only been around for a couple of years, still new to everyone.

 

[Bassman2003]

I agree. Too big and too proud. Normally a person or company would not charge for a good or service that was not ready for sale.

 

[ahalpert]

Very good article. They tried it and the videos were horrible. Yet Adobe still wants to charge for it! This does not mean all AI is or will be this bad, but not the best optics when the largest player in the digital software imaging space has this type of faceplant.

I still haven't read it but it's possible that if Adobe is being more ethical than other companies in choosing the content that in trains on then it may hamstring the project. Maybe they're trying to use only public domain stuff or stuff they have legal access to. So maybe the weakness of their model reflects doing the right thing. I dunno.

I will say that sora is very hit or miss for me and I think it works best when it's giving me something more similar to exactly what it trained on.

And I think that photoshop has the best generative photo editor available. I use generative fill all the time for photo editing now and it lets me do higher quality photoshopping than I know how to do and it lets me do it very fast. But I basically never fill in the prompt field. I leave it blank and let the machine generate what it thinks is most likely. In general I think that the less that I prompt and the more I let the program rely on its exact training data, the better the result. It's not actually capable of making anything.

 

[ahalpert]

This is big. Jason Kint is the head of a digital media trade association. I have been following him for his insights into the antitrust trials of Google, Amazon, etc. He's not the voice of God but he doesn't throw around hot takes casually. He just pinned a new tweet stating that Facebook has committed numerous IP violations in the creation of its AI and that the internal company documents unsealed in an ongoing lawsuit are "starting to look a lot like a crime scene." Click the tweet to see the rest of the thread and read the screenshotted documents to see for yourself.

https://twitter.com/x/status/1892977369826161045

 

[Bassman2003]

Whatever! Let them all burn. The obvious is going to be known and I think we will find out that many in the world do not respect copyright when it is in the way of something they want. All the money being spent... For what? I just do not see the human race level benefits outside of some medical research areas.

 

[NorBro]

It was always my [maybe misconceived] understanding that when you're a company that big you don't get to not commit crimes.

I'm not saying anyone committed crimes, I just always assumed because of how the world is, big companies are put in situations where they have to commit many crimes, sometimes frequently, to continue doing business and staying on top.

Are you surprised about any crimes, violations, illegality, in general - or that AI systems specifically may not have been built honestly?

 

[Bassman2003]

Are you surprised about any crimes, violations, illegality, in general - or that AI systems specifically may not have been built honestly?

Not at all. Unhappy, sad are better descriptors that these mega entities are driving our world now. I can only guess that they are planning something a lot larger than realistic fake videos and generative fill to match the amount of investment dollars. The big tech firms have committed to 70, 80, 100 billion dollars a year in AI spending. It just makes me think "what are we in for"?