Dot.Com Era/AI Era?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

HanSolo wrote: Sun Mar 31, 2024 4:30 pm
Logan Roy wrote: Sun Mar 31, 2024 3:08 pm
HanSolo wrote: Sun Mar 31, 2024 11:35 am
Logan Roy wrote: Sun Mar 31, 2024 8:15 am Both the human brain and an artificial neural network are 'black box' technologies. We know how they process information and the structures they use, but they're too complex to really break down much beyond that .. So all we've got is the output .. And if a computer neural net is generating similar or better results, it's probably doing something very similar.
Either that, or it's the ELIZA effect on steroids.

https://en.wikipedia.org/wiki/ELIZA_effect
Other way around, though.

It's that the way humans actually work is just cold computation. The experience of being human is something that emerges on top of a whole lot of unconscious brain processing – and is part of a very abstractified simulation of reality that humans share to simplify the task of navigation and having social connections and things.
That's an interesting belief system, if one wants to subscribe to it. It doesn't have much to do with investing, unless one is investing in science fictions films or something, as the above ideas are sometimes a basis for such stories.
It's responding to your suggestion. So it has as much to do with investing as the link to the ELIZA effect.

It's also the standard scientific view of how brains work. Daniel Dennett calls what you're referring to the 'Cartesian Theatre'.
gougou
Posts: 1342
Joined: Thu Sep 28, 2017 7:42 pm

Re: Dot.Com Era/AI Era?

Post by gougou »

Logan Roy wrote: Sun Mar 31, 2024 3:08 pm
HanSolo wrote: Sun Mar 31, 2024 11:35 am
Logan Roy wrote: Sun Mar 31, 2024 8:15 am
gougou wrote: Sat Mar 30, 2024 6:13 pm Computer can do certain things better than human. It can do calculations better. It can play chess better. In the future it might drive cars better than human.

With a lot of computational power and large language model it can play this “reasoning” and question-answer game better than human. You may call it human-like ability just like playing chess is a human-like ability. But it’s not anything more than that like some people are suggesting.
That's a common perspective. But I think it's probably wrong.

Both the human brain and an artificial neural network are 'black box' technologies. We know how they process information and the structures they use, but they're too complex to really break down much beyond that .. So all we've got is the output .. And if a computer neural net is generating similar or better results, it's probably doing something very similar.
Either that, or it's the ELIZA effect on steroids.

https://en.wikipedia.org/wiki/ELIZA_effect
Other way around, though.

It's that the way humans actually work is just cold computation. The experience of being human is something that emerges on top of a whole lot of unconscious brain processing – and is part of a very abstractified simulation of reality that humans share to simplify the task of navigation and having social connections and things.
Nobody understands how human brain works. What you said is pure speculation.

Just look at how difficult self-driving is. With so many sensors, so much power consumption the AI is still inferior to human drivers.
The sillier the market’s behavior, the greater the opportunity for the business like investor.
User avatar
nisiprius
Advisory Board
Posts: 52348
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Dot.Com Era/AI Era?

Post by nisiprius »

I just thought of another example of forgotten hype. In the late 1950s and early 1960s there was a great deal of hype around "teaching machines," based on the work of B. F. Skinner and colleagues. It was widely speculated, feared, hoped, that they would literally replace human teachers. Lots of technical argle-bargle about optimum reinforcement schedules, and the relative merits of linear versus branching devices, etc. etc.

One example. Time Magazine, November 7th, 1960:
New York's Collegiate School for boys tried teaching ma-
chines in math, found that 73 students completed in only two weeks an abstract-algebra course that usually requires two months. The Roanoke public schools used teaching machines on 34 eighth-graders—with no oral teaching and no homework—and in less than one semester, all 34 completed a year's work in algebra.
The great revolution hasn't happened. Programmed instruction of various sorts is with us, as part of a landscape of teaching and learning techniques, but just another part. It wasn't transformational the way it was hyped to be, and it didn't put teachers out of work.

Come to think of it, there was also an idea that AV techniques would revolutionize teaching. Why have a human teacher give the same lecture over and over again, when you could film or videotype it once and then just have students watch the recorded lecture forever after? Many people were hugely impressed by the training films produced by the U.S. armed forces--many by Disney--and thought they were seeing something transformational in nature.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

gougou wrote: Sun Mar 31, 2024 7:48 pm
Logan Roy wrote: Sun Mar 31, 2024 3:08 pm
HanSolo wrote: Sun Mar 31, 2024 11:35 am
Logan Roy wrote: Sun Mar 31, 2024 8:15 am
gougou wrote: Sat Mar 30, 2024 6:13 pm Computer can do certain things better than human. It can do calculations better. It can play chess better. In the future it might drive cars better than human.

With a lot of computational power and large language model it can play this “reasoning” and question-answer game better than human. You may call it human-like ability just like playing chess is a human-like ability. But it’s not anything more than that like some people are suggesting.
That's a common perspective. But I think it's probably wrong.

Both the human brain and an artificial neural network are 'black box' technologies. We know how they process information and the structures they use, but they're too complex to really break down much beyond that .. So all we've got is the output .. And if a computer neural net is generating similar or better results, it's probably doing something very similar.
Either that, or it's the ELIZA effect on steroids.

https://en.wikipedia.org/wiki/ELIZA_effect
Other way around, though.

It's that the way humans actually work is just cold computation. The experience of being human is something that emerges on top of a whole lot of unconscious brain processing – and is part of a very abstractified simulation of reality that humans share to simplify the task of navigation and having social connections and things.
Nobody understands how human brain works. What you said is pure speculation.

Just look at how difficult self-driving is. With so many sensors, so much power consumption the AI is still inferior to human drivers.
But there are serious academic subjects, like computational neuroscience, that probably get a fair bit right. They are based on how we know neurones work, and how we know computation works, and our ability to map and simulate the neural networks of simpler lifeforms .. When it comes to more complex behaviours, the fact they seem to emerge in systems like GPT probably gives us more insight into emergent intelligence in human brains too.

It's a sensible, logical, evidence-based way to look at things. Certainly far less speculative than any alternative, and more scientific than throwing our hands up in the air and sighing.

And I'm not sure AI's necessarily inferior at driving. Again, that seems like an assumption without evidence .. Waymo has 7+ million miles of driverless data now, and their self-driving cars are nearly 7x less likely than human drivers to be involved in a serious crash; 2.3x less likely to be involved in a crash. By all accounts, humans are quite bad at driving. But the standard for an autonomous vehicle is one that almost never has an accident.
User avatar
HanSolo
Posts: 2377
Joined: Thu Jul 19, 2012 3:18 am

Re: Dot.Com Era/AI Era?

Post by HanSolo »

Logan Roy wrote: Sun Mar 31, 2024 4:45 pm
HanSolo wrote: Sun Mar 31, 2024 4:30 pm
Logan Roy wrote: Sun Mar 31, 2024 3:08 pm
HanSolo wrote: Sun Mar 31, 2024 11:35 am
Logan Roy wrote: Sun Mar 31, 2024 8:15 am Both the human brain and an artificial neural network are 'black box' technologies. We know how they process information and the structures they use, but they're too complex to really break down much beyond that .. So all we've got is the output .. And if a computer neural net is generating similar or better results, it's probably doing something very similar.
Either that, or it's the ELIZA effect on steroids.

https://en.wikipedia.org/wiki/ELIZA_effect
Other way around, though.

It's that the way humans actually work is just cold computation. The experience of being human is something that emerges on top of a whole lot of unconscious brain processing – and is part of a very abstractified simulation of reality that humans share to simplify the task of navigation and having social connections and things.
That's an interesting belief system, if one wants to subscribe to it. It doesn't have much to do with investing, unless one is investing in science fictions films or something, as the above ideas are sometimes a basis for such stories.
It's responding to your suggestion. So it has as much to do with investing as the link to the ELIZA effect.
Well, someone had to bring an alternative view to the wild speculations about how things work (as if they were facts).

For the record, I wasn't the one who departed from the financial topic to discuss those wild speculations.
It's also the standard scientific view of how brains work.
There's nothing "standard" about it. It's an available model, speculation and/or belief system, that's all.
Strategic Macro Senior (top 1%, 2019 Bogleheads Contest)
loukycpa
Posts: 789
Joined: Wed Aug 05, 2020 9:52 am

Re: Dot.Com Era/AI Era?

Post by loukycpa »

Ignoring recent year (post 2020) highs, am I correct that CAPE 10 and P/E ratios have only reached these levels once? Which was in the late 90s (dot com era)?

I lived through this period. It was the beginning of accumulation stage for me. S&P 500 earnings were around 75 and dividends were around 30 in 1998. CAPE mid 30s and climbing. 10 year treasury rate around 5% or 6%.

15 years later at the end of 2013 S&P 500 earnings were around 130 and dividends 47 (inflation adjusted). So moderate growth happened, kind of on the low side of normal I think. Any productivity gains that came from the internet seem to have been realized by the public but not corporate shareholders. (That's actually how capitalism should work I think, when competition is alive and well.)

Source: https://www.multpl.com/shiller-pe
Source: https://www.multpl.com/s-p-500-earnings
Source: https://www.multpl.com/s-p-500-dividend
Source: https://www.multpl.com/10-year-treasury-rate

What played out over those same 15 years for corporate shareholders?

https://www.portfoliovisualizer.com/bac ... rdppnpFxfC

Despite that growth in S&P earnings and dividends, valuations in terms of earnings multiples came back down. As a result we had a 15 year period where despite a wild ride you basically got no benefit from having equities in your portfolio. These are the periods that make staying the course a very difficult thing to do.

No predictions being made that the next 15 years will be a repeat of 1998 through 2013. However, my caution would be that you need to have a plan and stick to it. And in that plan you always need to be ready to go through a period like this one. There was plenty of good stuff on the other side of this (over the next 10 years, 2014 to today), but you had to be able to stay the course to see it.

Incidentally, S&P 500 earnings are around 185 and dividends of 70 today (inflation adjusted). Both earnings and dividends have doubled in real terms over this time period (about 25 years). That is the reason we hold stocks in the first place and the reason to stay the course.

Mike Tyson: "Everyone has a plan until they get punched in the mouth."
Last edited by loukycpa on Tue Apr 02, 2024 2:34 pm, edited 2 times in total.
"The safe assumption for an investor is that over the next hundred years, the currency is going to zero." - Charlie Munger
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

HanSolo wrote: Mon Apr 01, 2024 6:32 am
Logan Roy wrote: Sun Mar 31, 2024 4:45 pm
HanSolo wrote: Sun Mar 31, 2024 4:30 pm
Logan Roy wrote: Sun Mar 31, 2024 3:08 pm
HanSolo wrote: Sun Mar 31, 2024 11:35 am

Either that, or it's the ELIZA effect on steroids.

https://en.wikipedia.org/wiki/ELIZA_effect
Other way around, though.

It's that the way humans actually work is just cold computation. The experience of being human is something that emerges on top of a whole lot of unconscious brain processing – and is part of a very abstractified simulation of reality that humans share to simplify the task of navigation and having social connections and things.
That's an interesting belief system, if one wants to subscribe to it. It doesn't have much to do with investing, unless one is investing in science fictions films or something, as the above ideas are sometimes a basis for such stories.
It's responding to your suggestion. So it has as much to do with investing as the link to the ELIZA effect.
Well, someone had to bring an alternative view to the wild speculations about how things work (as if they were facts).

For the record, I wasn't the one who departed from the financial topic to discuss those wild speculations.
It's also the standard scientific view of how brains work.
There's nothing "standard" about it. It's an available model, speculation and/or belief system, that's all.
Just on this .. The basics of how the brain works are widely accepted in science and medicine .. The brain is a network of neurones, connected by synapses, which transfer information between neurones in hierarchical structures.

This was the inspiration for artificial neural networks. And while the basic view of how the brain processes information isn't particularly speculative (we can test neurones, and see them firing impulses in networks), the fact that software based on this architecture is able to summarise text, write songs, hold conversations, use abstract reasoning, etc., when it's not been specifically programmed to do this, adds weight to this being a good approximation of how brains perform those tasks.
User avatar
HanSolo
Posts: 2377
Joined: Thu Jul 19, 2012 3:18 am

Re: Dot.Com Era/AI Era?

Post by HanSolo »

Logan Roy wrote: Mon Apr 01, 2024 3:29 pm Just on this .. The basics of how the brain works are widely accepted in science and medicine .. The brain is a network of neurones, connected by synapses, which transfer information between neurones in hierarchical structures.
My previous responses were about your previous assertions. Now, responding to the above, that can be true while not contradicting anything I said (including pointing out that some of your previous assertions were speculations and/or beliefs). If you have additional assertions on that topic, then let's just say the pursuit of sorting out what's fact, fiction, speculation or belief is reaching a point of diminishing returns (aka "beating dead horse"). Since this is the "Investing" category, perhaps something investing-related would be more interesting subject matter. Just a thought.
Strategic Macro Senior (top 1%, 2019 Bogleheads Contest)
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

HanSolo wrote: Mon Apr 01, 2024 6:57 pm
Logan Roy wrote: Mon Apr 01, 2024 3:29 pm Just on this .. The basics of how the brain works are widely accepted in science and medicine .. The brain is a network of neurones, connected by synapses, which transfer information between neurones in hierarchical structures.
My previous responses were about your previous assertions. Now, responding to the above, that can be true while not contradicting anything I said (including pointing out that some of your previous assertions were speculations and/or beliefs). If you have additional assertions on that topic, then let's just say the pursuit of sorting out what's fact, fiction, speculation or belief is reaching a point of diminishing returns (aka "beating dead horse"). Since this is the "Investing" category, perhaps something investing-related would be more interesting subject matter. Just a thought.
If you'd like to be more specific, I can address anything. I work in machine learning and studied computational neuroscience. I don't mean to derail this, but when someone tells me I'm talking out of my rear end, it's either that I've not explained something properly, or there's something really big that I don't know about.

What you're suggesting is an unscientific view seems to be this: "The experience of being human is something that emerges on top of a whole lot of unconscious brain processing" .. Afaic, that's a completely reasonable, academic viewpoint. I think it would be far more difficult to argue something significantly contrary to that.
User avatar
HanSolo
Posts: 2377
Joined: Thu Jul 19, 2012 3:18 am

Re: Dot.Com Era/AI Era?

Post by HanSolo »

Logan Roy wrote: Tue Apr 02, 2024 10:09 am If you'd like to be more specific, I can address anything.
My previous comments adequately addressed the points as needed. If you disagree with my responses, that's up to you.
What you're suggesting is an unscientific view seems to be this:
I responded to multiple assertions, some of which were speculative, or indicative of belief or other forms of narrative.
.. Afaic, that's a completely reasonable, academic viewpoint.
Yes. Sometimes it's reasonable to speculate something, or to create a narrative. But it's still a speculation/narrative.

This is getting repetitive. Not sure if you missed this:
HanSolo wrote: Mon Apr 01, 2024 6:57 pm If you have additional assertions on that topic, then let's just say the pursuit of sorting out what's fact, fiction, speculation or belief is reaching a point of diminishing returns (aka "beating dead horse"). Since this is the "Investing" category, perhaps something investing-related would be more interesting subject matter. Just a thought.
Strategic Macro Senior (top 1%, 2019 Bogleheads Contest)
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

HanSolo wrote: Tue Apr 02, 2024 10:20 am
Logan Roy wrote: Tue Apr 02, 2024 10:09 am If you'd like to be more specific, I can address anything.
My previous comments adequately addressed the points as needed. If you disagree with my responses, that's up to you.
What you're suggesting is an unscientific view seems to be this:
I responded to multiple assertions, some of which were speculative, or indicative of belief or other forms of narrative.
.. Afaic, that's a completely reasonable, academic viewpoint.
Yes. Sometimes it's reasonable to speculate something, or to create a narrative. But it's still a speculation/narrative.

This is getting repetitive. Not sure if you missed this:
HanSolo wrote: Mon Apr 01, 2024 6:57 pm If you have additional assertions on that topic, then let's just say the pursuit of sorting out what's fact, fiction, speculation or belief is reaching a point of diminishing returns (aka "beating dead horse"). Since this is the "Investing" category, perhaps something investing-related would be more interesting subject matter. Just a thought.
You typed all of that without providing the specifics I asked for. I'll assume you don't have it, and we can both move on.
Manny1066
Posts: 38
Joined: Fri Aug 04, 2023 3:22 pm

Re: Dot.Com Era/AI Era?

Post by Manny1066 »

My big worry is that inflation and Fed actions will put an end to the bull market, and many of these AI firms will get wrecked, or go under completely

Now I am not an economist, but from what I have read in the last 4-5 months has led me to conclude that inflation is NOT completely under control, and that there is a very good chance the Fed will not cut rates at all this year. If inflation picks up in late Summer, Powell might even hike.

Now what that means for the market is anyone's guess, but I already lived through a similar scenario in early 2000. And if we want another example, we need to go back to the early 70s.

In other words, a long-term, secular bear market may result. Securities will get punished, and the bond market will suffer in the short-term. Funds like VOO will get cut in half.

bt the other complication is the upcoming election. If the Fed does nothing, or even hikes, the administration will be furious, as it could send the economy into a recession by the 4th quarter.

Scary times ...
User avatar
HanSolo
Posts: 2377
Joined: Thu Jul 19, 2012 3:18 am

Re: Dot.Com Era/AI Era?

Post by HanSolo »

Logan Roy wrote: Tue Apr 02, 2024 10:50 am You typed all of that without providing the specifics I asked for. I'll assume you don't have it, and we can both move on.
You didn't ask for anything. You said, "if you'd like", and I have no motivation to revisit issues I already addressed adequately. I stand by my earlier responses. If you disagree, that's not a problem.

If nothing else, it's clear that the one assertion I made about AI was accurate: that AI has been known to, and will continue to, inspire endless pontification and prognostication.

The implication for investors? Tune out the noise. Stay the course.

I agree with moving on. Or getting back on-topic at least.
Strategic Macro Senior (top 1%, 2019 Bogleheads Contest)
User avatar
HanSolo
Posts: 2377
Joined: Thu Jul 19, 2012 3:18 am

Re: Dot.Com Era/AI Era?

Post by HanSolo »

Manny1066 wrote: Tue Apr 02, 2024 12:16 pm My big worry is that inflation and Fed actions will put an end to the bull market, and many of these AI firms will get wrecked, or go under completely
It's the companies that are flimsy, hyped-up or otherwise grossly overvalued that will suffer that fate as a result of the Fed merely doing its job. Strong, well-managed companies will do just fine, because they're managed in such as way as to be able to weather the economic ups and downs. This is how capitalism is supposed to work.

"Only when the tide goes out do you discover who's been swimming naked." -- Warren Buffett

In the meantime, there will be bull markets and bear markets, because that's what's normal. No need to worry.
Scary times ...
What would be scary is if capitalism didn't work.
Strategic Macro Senior (top 1%, 2019 Bogleheads Contest)
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

The thing with capitalism is it's ultimately profit destroying. The goal is always perfect competition. Hence Fisher's Uncommon Profits, etc. etc. Buffett and Munger talk/ed about this a lot. It's the core of that investing philosophy.

The extreme skewness of market returns is such that for us to get our inflation-beating return, we're reliant on a handful of businesses that don't conform to the rules of capitalism. Like Coca Cola – there's a premium on the price people are willing to pay that doesn't reduce to simple supply and demand. And we expect AI to have a deflationary effect. But in making things it operates in more efficient, it should also improve capitalism .. Which means every sectors becomes more like the airlines industry: such tight competition that no one really makes a profit, and investors don't really make anything.

I think the unexpected thing is that markets are right. Chipmakers are probably massively undervalued. Because while AI can bring every business up to the same level of peak efficiency, it creates an arms race. All businesses are destined to make no excess profit, but just to stay in the race, you need a lot of processing power. AI can be the ultimate disruptor. And it's everything that can be disrupted in this next era that's at risk.
Valuethinker
Posts: 49122
Joined: Fri May 11, 2007 11:07 am

Re: Dot.Com Era/AI Era?

Post by Valuethinker »

nisiprius wrote: Sun Mar 31, 2024 8:00 pm I just thought of another example of forgotten hype. In the late 1950s and early 1960s there was a great deal of hype around "teaching machines," based on the work of B. F. Skinner and colleagues. It was widely speculated, feared, hoped, that they would literally replace human teachers. Lots of technical argle-bargle about optimum reinforcement schedules, and the relative merits of linear versus branching devices, etc. etc.

One example. Time Magazine, November 7th, 1960:
New York's Collegiate School for boys tried teaching ma-
chines in math, found that 73 students completed in only two weeks an abstract-algebra course that usually requires two months. The Roanoke public schools used teaching machines on 34 eighth-graders—with no oral teaching and no homework—and in less than one semester, all 34 completed a year's work in algebra.
The great revolution hasn't happened. Programmed instruction of various sorts is with us, as part of a landscape of teaching and learning techniques, but just another part. It wasn't transformational the way it was hyped to be, and it didn't put teachers out of work.

Come to think of it, there was also an idea that AV techniques would revolutionize teaching. Why have a human teacher give the same lecture over and over again, when you could film or videotype it once and then just have students watch the recorded lecture forever after? Many people were hugely impressed by the training films produced by the U.S. armed forces--many by Disney--and thought they were seeing something transformational in nature.
My spouse is doing another undergraduate degree by distance learning. The Open University, which was once a world leader in this field (before the internet was in wide use). 1 in 4 MBAs in the UK is via the OU. Prime Minister Harold Wilson said its creation was his proudest achievement*.

I would have to say the quality of the recorded and written material is much better, on average, than my experience of in class instruction through several university degrees. So I am a huge fan of the OU.

But I do recognise there is something about being "kept up to the mark" by in class instruction. It may lodge in memory better. I will remember a lecture I had by Alan Bloom for the rest of my life, I imagine. And one by Norman Davies.

We waste a lot of professor time essentially preparing and delivering the same material in hundreds of identical lecture halls across the world. And there's that infamous Intro to Psych lectures at U of Toronto, with over 1200 students in Convocation Hall. What would be better is if students watched the lectures of video, in their own time, and attended tutorials with a lot more engaged learning. (maybe they do that now - a quick google doesn't answer that question)


* Wilson was a Labour Party Prime Minister -- not the party traditionally associated with being pro monarchy.** It is also true, however, that HM Queen Elizabeth II, dined with only 2 Prime Ministers at their residence, 10 Downing Street. The other was Sir Winston Churchill.

The Crown (Netflix) is of course total fiction. We don't know what HM QE II was like or what she thought. We never will - State Secrecy laws and the code of Omerta protect the Sovereign's weekly conversations with the Prime Minister into the forever.

Nonetheless there is a touch of Wilson in the meeting below, when after the tragedy of Aberfan, when a coal pit tip collapses down a slope and buries the village school, she confronts him (Olivia Colman just cannot do a role badly):

https://www.youtube.com/watch?v=DT-eH0iC87w

(in actual fact, it was at Aberfan when there was one of the few times HM QE II showed public emotion).

** stories do suggest the British Prime Minister (of her 14 or so) that HMQ really did *not* get on with was Margaret Thatcher, the Iron Lady. But that might be wish fulfilment in some quarters.
loukycpa
Posts: 789
Joined: Wed Aug 05, 2020 9:52 am

Re: Dot.Com Era/AI Era?

Post by loukycpa »

delete
Last edited by loukycpa on Tue Apr 02, 2024 2:33 pm, edited 1 time in total.
"The safe assumption for an investor is that over the next hundred years, the currency is going to zero." - Charlie Munger
loukycpa
Posts: 789
Joined: Wed Aug 05, 2020 9:52 am

Re: Dot.Com Era/AI Era?

Post by loukycpa »

delete
"The safe assumption for an investor is that over the next hundred years, the currency is going to zero." - Charlie Munger
Valuethinker
Posts: 49122
Joined: Fri May 11, 2007 11:07 am

Re: Dot.Com Era/AI Era?

Post by Valuethinker »

Logan Roy wrote: Tue Apr 02, 2024 2:00 pm The thing with capitalism is it's ultimately profit destroying. The goal is always perfect competition. Hence Fisher's Uncommon Profits, etc. etc. Buffett and Munger talk/ed about this a lot. It's the core of that investing philosophy.

The extreme skewness of market returns is such that for us to get our inflation-beating return, we're reliant on a handful of businesses that don't conform to the rules of capitalism. Like Coca Cola – there's a premium on the price people are willing to pay that doesn't reduce to simple supply and demand. And we expect AI to have a deflationary effect. But in making things it operates in more efficient, it should also improve capitalism .. Which means every sectors becomes more like the airlines industry: such tight competition that no one really makes a profit, and investors don't really make anything.
I think that is overly ... hopeful.

The reason competition doesn't crush returns for some companies is they are adroit at exploiting market advantages. Scale and network effects (Google and Facebook, Amazon; also Boeing and Airbus). Watch Google and FB stare down the Canadian government on paying for news content. Platform dominance (Apple, the Kindle, Microsoft). Portfolios of patents and processes (Kodak as was, Xerox). Domination of distribution chain (Anheuser-Busch, Coca Cola). Purchasing power (any big supermarket chain, Walmart). Locational advantages (WalMart, any mining company, airlines controlling hubs, Warren Buffett's railways). Not all of these tactics are legal, but the enthusiasm of regulators for confronting market dominance has varied over time, and has been quite weak in the last 30 years or so.

Any good business school strategy textbook - back to Michael Porter's Competitive Strategy and Competitive Advantage, will tell you how to do this - albeit with the problem that much of the secret is also illegal in some or all jurisdictions.
I think the unexpected thing is that markets are right. Chipmakers are probably massively undervalued. Because while AI can bring every business up to the same level of peak efficiency, it creates an arms race. All businesses are destined to make no excess profit, but just to stay in the race, you need a lot of processing power. AI can be the ultimate disruptor. And it's everything that can be disrupted in this next era that's at risk.
History tends to suggest over-hyping, particularly of anything that says "Artificial Intelligence" on the tin. We've been here before - at least 3 times previously? Boom and then bust.

Which doesn't mean that the technology will not have profound consequences. It's the *adoption* of a technology which creates the revolution -- the explosion of its benefits into the wider slipstream of economic activity and mass consumer use. The steam train. The telegraph. The radio. The airplane. The automobile. Each of these caused a revolution in the way people live and work. Not necessarily the companies making and deploying that innovation -- Dell Computer, anyone? GM?

On Chips, it's an incredibly capital intensive industry. Chip fabrication is really something only a handful of companies can do because the scale of investment required is so huge. There's a bigger ecosystem of "IP Only" companies of course, but that may, in turn, become their vulnerability.

In one sense the classic case is Advanced RISC Machines - ARM. ARM chips are the most common logical processors in the world by some large factor - in particular its dominance of mobile devices. But ARM is a licensing company - its actual revenues are not huge for something so profoundly important to the modern world we live in. ARM has changed the world, but it gets only modestly rich from that.

Coincident with that capital intensitivity, is cyclicality. Just like the oil industry. Or the mining industry. Although perhaps pharmaceuticals are a better analogy, given the IP protection extant.
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

Valuethinker wrote: Tue Apr 02, 2024 2:46 pm
Logan Roy wrote: Tue Apr 02, 2024 2:00 pm The thing with capitalism is it's ultimately profit destroying. The goal is always perfect competition. Hence Fisher's Uncommon Profits, etc. etc. Buffett and Munger talk/ed about this a lot. It's the core of that investing philosophy.

The extreme skewness of market returns is such that for us to get our inflation-beating return, we're reliant on a handful of businesses that don't conform to the rules of capitalism. Like Coca Cola – there's a premium on the price people are willing to pay that doesn't reduce to simple supply and demand. And we expect AI to have a deflationary effect. But in making things it operates in more efficient, it should also improve capitalism .. Which means every sectors becomes more like the airlines industry: such tight competition that no one really makes a profit, and investors don't really make anything.
I think that is overly ... hopeful.

The reason competition doesn't crush returns for some companies is they are adroit at exploiting market advantages. Scale and network effects (Google and Facebook, Amazon; also Boeing and Airbus). Watch Google and FB stare down the Canadian government on paying for news content. Platform dominance (Apple, the Kindle, Microsoft). Portfolios of patents and processes (Kodak as was, Xerox). Domination of distribution chain (Anheuser-Busch, Coca Cola). Purchasing power (any big supermarket chain, Walmart). Locational advantages (WalMart, any mining company, airlines controlling hubs, Warren Buffett's railways). Not all of these tactics are legal, but the enthusiasm of regulators for confronting market dominance has varied over time, and has been quite weak in the last 30 years or so.

Any good business school strategy textbook - back to Michael Porter's Competitive Strategy and Competitive Advantage, will tell you how to do this - albeit with the problem that much of the secret is also illegal in some or all jurisdictions.
I think the unexpected thing is that markets are right. Chipmakers are probably massively undervalued. Because while AI can bring every business up to the same level of peak efficiency, it creates an arms race. All businesses are destined to make no excess profit, but just to stay in the race, you need a lot of processing power. AI can be the ultimate disruptor. And it's everything that can be disrupted in this next era that's at risk.
History tends to suggest over-hyping, particularly of anything that says "Artificial Intelligence" on the tin. We've been here before - at least 3 times previously? Boom and then bust.

Which doesn't mean that the technology will not have profound consequences. It's the *adoption* of a technology which creates the revolution -- the explosion of its benefits into the wider slipstream of economic activity and mass consumer use. The steam train. The telegraph. The radio. The airplane. The automobile. Each of these caused a revolution in the way people live and work. Not necessarily the companies making and deploying that innovation -- Dell Computer, anyone? GM?

On Chips, it's an incredibly capital intensive industry. Chip fabrication is really something only a handful of companies can do because the scale of investment required is so huge. There's a bigger ecosystem of "IP Only" companies of course, but that may, in turn, become their vulnerability.

In one sense the classic case is Advanced RISC Machines - ARM. ARM chips are the most common logical processors in the world by some large factor - in particular its dominance of mobile devices. But ARM is a licensing company - its actual revenues are not huge for something so profoundly important to the modern world we live in. ARM has changed the world, but it gets only modestly rich from that.

Coincident with that capital intensitivity, is cyclicality. Just like the oil industry. Or the mining industry. Although perhaps pharmaceuticals are a better analogy, given the IP protection extant.
Exactly. And I think Zero to One has become the handbook on how to do that in the 'information economy'. Every tech platform I've adopted since reading that book has followed it to the letter .. But it's the failure to regulate today's Big Tech that has to be the biggest threat to capitalism. Since the 80s, every decade's at least been dominated by different tech firms. The hype never held because the landscape kept changing – but I worry that now you've got bloated, inefficiently run firms compounding unassailable leads with machine learning and data.

And yes, it's not the revolution (air travel) that generates the uncommon profits. I expect AI hype to overshoot, but what I don't think the average investor or analyst is really considering is what happens when the standard for video game AI characters is GPT5? It's already being done with mods for PC games. When you've got an 80 million-copy-selling console game with that as a standard, done really well, developers will never be able to go back. And that means a lot of cloud processing infrastructure. Or it gets very expensive, but I think these firms are going to make that investment on a 15-30 year timeframe, because otherwise they can't compete. Then an iOS update with GPT-like Siri. It's the sheer demand side of the equation. I think everyone expects boom and bust this time, and my fear is we're wrong (because during 2020-2022, I diversified into everything that looked solid and cheap, like TIPS and core infrastructure .. and I hope that prudence pays off .. otherwise I'll regret not just putting 70% in an index tracker).
User avatar
HanSolo
Posts: 2377
Joined: Thu Jul 19, 2012 3:18 am

Re: Dot.Com Era/AI Era?

Post by HanSolo »

Logan Roy wrote: Tue Apr 02, 2024 2:00 pm The thing with capitalism is it's ultimately profit destroying.
You're welcome to try some other system. They're out there.
AI can be the ultimate disruptor.
Just like the steam engine, electronics, air travel and outsourcing. There's always an ultimate disruptor.

The more things change, the more they stay the same.

This time isn't different. Tune out the noise. Stay the course.
Strategic Macro Senior (top 1%, 2019 Bogleheads Contest)
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

HanSolo wrote: Wed Apr 03, 2024 11:38 am
Logan Roy wrote: Tue Apr 02, 2024 2:00 pm The thing with capitalism is it's ultimately profit destroying.
You're welcome to try some other system. They're out there.
AI can be the ultimate disruptor.
Just like the steam engine, electronics, air travel and outsourcing. There's always an ultimate disruptor.

The more things change, the more they stay the same.

This time isn't different. Tune out the noise. Stay the course.
Profit destruction is why capitalism's worked. Feudalism – when a lord could buy all the produce, and sell it back to a starving village – was an uncompetitive supply/demand situation. And brands like Apple and Coca Cola do mean we still overpay for some things. And when they become monopolies (maybe more like Amazon, Google Ads) you don't necessarily have much choice.

I don't think capitalism will take us much further. Competition's been a way to outsource optimisation problems. But AI now offers a much more efficient way of achieving the same thing. And it can take over without having to rebuild the system. If AI analysts do a better job, then they wind up controlling the market anyway. At some point, presumably, someone realises: you don't actually need human traders. BlackRock can just allocate capital with AI-based systems, and wind up owning everything. China's planning the same thing – it's called digital authoritarianism.
User avatar
HanSolo
Posts: 2377
Joined: Thu Jul 19, 2012 3:18 am

Re: Dot.Com Era/AI Era?

Post by HanSolo »

Logan Roy wrote: Wed Apr 03, 2024 4:52 pm Profit destruction is why capitalism's worked.
As some have said, it's the worst economic system, except for all the others.
I don't think capitalism will take us much further.
They've been saying that for decades, if not centuries. That's a perpetual prognostication.
But AI now offers a much more efficient way of achieving the same thing. And it can take over without having to rebuild the system.
There's always something new that will "take over". That's another perpetual prognostication.

Which brings us back to:
HanSolo wrote: Wed Apr 03, 2024 11:38 am There's always an ultimate disruptor.

The more things change, the more they stay the same.

This time isn't different. Tune out the noise. Stay the course.
Strategic Macro Senior (top 1%, 2019 Bogleheads Contest)
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

HanSolo wrote: Thu Apr 04, 2024 1:28 am
Logan Roy wrote: Wed Apr 03, 2024 4:52 pm Profit destruction is why capitalism's worked.
As some have said, it's the worst economic system, except for all the others.
I don't think capitalism will take us much further.
They've been saying that for decades, if not centuries. That's a perpetual prognostication.
But AI now offers a much more efficient way of achieving the same thing. And it can take over without having to rebuild the system.
There's always something new that will "take over". That's another perpetual prognostication.

Which brings us back to:
HanSolo wrote: Wed Apr 03, 2024 11:38 am There's always an ultimate disruptor.

The more things change, the more they stay the same.

This time isn't different. Tune out the noise. Stay the course.
No one's saying sell all your stocks, or don't stay the course.

But the fact we're not still living under feudalism – which was the norm for about 600 years, until the 15th century – shows that things do change.
adave
Posts: 592
Joined: Thu Sep 13, 2007 7:17 pm
Location: Houston

Re: Dot.Com Era/AI Era?

Post by adave »

Mega cap tech stocks were tested in 2022 and bounced back well from rising interest rates. It seems that these companies are money printers and thus not as effected by rising rates. They seem to be able to survive / thrive in any economic environment.

Part of me believes the ultimate endpoint for any large network is several very large players become dominant and eat the rest.

It wouldn't shock me if the the mega cap Tech companies "win" the AI cycle. There is a certain winner take all dynamic in the Tech ecosystem imo.
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

adave wrote: Thu Apr 04, 2024 1:26 pm Mega cap tech stocks were tested in 2022 and bounced back well from rising interest rates. It seems that these companies are money printers and thus not as effected by rising rates. They seem to be able to survive / thrive in any economic environment.

Part of me believes the ultimate endpoint for any large network is several very large players become dominant and eat the rest.

It wouldn't shock me if the the mega cap Tech companies "win" the AI cycle. There is a certain winner take all dynamic in the Tech ecosystem imo.
I agree. And I think it's because governments haven't worked out how to regulate big tech, so they're all effectively allowed to operate as monopolies (in a practical rather than legal sense of the word), while growing at a pace at which anti-competitive lawsuits don't make a dent. So they're already too large to be effectively regulated .. China has arguably done a better job, but now who wants to invest in China?

So what's to stop them becoming more powerful than governments? They'll control everything; they'll be completely global and opaque; the consumer will keep opting into their ecosystems (there are large parts of the modern world you can't now access without a smartphone). And it's not necessarily a bad thing. Governments are responsible for almost all remaining conflict and famine in the world. But the individual will be back to having zero power. Every decision will be made for you.
User avatar
HanSolo
Posts: 2377
Joined: Thu Jul 19, 2012 3:18 am

Re: Dot.Com Era/AI Era?

Post by HanSolo »

Logan Roy wrote: Thu Apr 04, 2024 1:20 pm
HanSolo wrote: Thu Apr 04, 2024 1:28 am
HanSolo wrote: Wed Apr 03, 2024 11:38 am There's always an ultimate disruptor.

The more things change, the more they stay the same.

This time isn't different. Tune out the noise. Stay the course.
No one's saying sell all your stocks, or don't stay the course.
I was just taking the opportunity to note that my prognostication about AI continues to be the most reliable one so far.
But the fact we're not still living under feudalism – which was the norm for about 600 years, until the 15th century – shows that things do change.
I acknowledged that some things change. My comment was accurate within the context of the topic of this thread.
Logan Roy wrote: Thu Apr 04, 2024 1:53 pm So what's to stop them becoming more powerful than governments? They'll control everything; they'll be completely global and opaque; the consumer will keep opting into their ecosystems (there are large parts of the modern world you can't now access without a smartphone).
Maybe, maybe not. One way the above could be disrupted is if some savvy young people decide that "the new cool" is to not carry those devices around.

I don't have a prediction for this, so I won't bet one way or the other.
Strategic Macro Senior (top 1%, 2019 Bogleheads Contest)
unwitting_gulag
Posts: 695
Joined: Mon Dec 05, 2016 3:37 pm

Re: Dot.Com Era/AI Era?

Post by unwitting_gulag »

Logan Roy wrote: Thu Apr 04, 2024 1:53 pm
adave wrote: Thu Apr 04, 2024 1:26 pm Mega cap tech stocks were tested in 2022 and bounced back well from rising interest rates. It seems that these companies are money printers and thus not as effected by rising rates. They seem to be able to survive / thrive in any economic environment.

Part of me believes the ultimate endpoint for any large network is several very large players become dominant and eat the rest.

It wouldn't shock me if the the mega cap Tech companies "win" the AI cycle. There is a certain winner take all dynamic in the Tech ecosystem imo.
I agree. And I think it's because governments haven't worked out how to regulate big tech, so they're all effectively allowed to operate as monopolies (in a practical rather than legal sense of the word), while growing at a pace at which anti-competitive lawsuits don't make a dent. So they're already too large to be effectively regulated .. China has arguably done a better job, but now who wants to invest in China?

So what's to stop them becoming more powerful than governments? They'll control everything; they'll be completely global and opaque; the consumer will keep opting into their ecosystems (there are large parts of the modern world you can't now access without a smartphone). And it's not necessarily a bad thing. Governments are responsible for almost all remaining conflict and famine in the world. But the individual will be back to having zero power. Every decision will be made for you.
One wonders how different (or not) it was in prior cycles of mass-acceptance and heady growth of new technologies. Steam engines. Railroads. Telegraphy. Later, internal combustion engines, automobiles, telephones and radio. Still later, chemicals and computers. And so on. Each time, there's a technology revolution, with a spate of entrants into the market, followed by consolidation.

The question for us as investors, is whether small-caps and mid-caps will ever again (in the reasonably near future) have their day. If AI goes the consolidation route, and consolidation favors the mega-caps, will we for the foreseeable future have a magnificent-something-or-other (never mind its exact composition or number), or will the market broaden? The actionable thing here, is whether to overweight the S&P 500, capturing the largest of the large, or whether we revert to the old "factor" thinking, that smaller --> higher CAGR (eventually).
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

unwitting_gulag wrote: Thu Apr 04, 2024 4:45 pm
Logan Roy wrote: Thu Apr 04, 2024 1:53 pm
adave wrote: Thu Apr 04, 2024 1:26 pm Mega cap tech stocks were tested in 2022 and bounced back well from rising interest rates. It seems that these companies are money printers and thus not as effected by rising rates. They seem to be able to survive / thrive in any economic environment.

Part of me believes the ultimate endpoint for any large network is several very large players become dominant and eat the rest.

It wouldn't shock me if the the mega cap Tech companies "win" the AI cycle. There is a certain winner take all dynamic in the Tech ecosystem imo.
I agree. And I think it's because governments haven't worked out how to regulate big tech, so they're all effectively allowed to operate as monopolies (in a practical rather than legal sense of the word), while growing at a pace at which anti-competitive lawsuits don't make a dent. So they're already too large to be effectively regulated .. China has arguably done a better job, but now who wants to invest in China?

So what's to stop them becoming more powerful than governments? They'll control everything; they'll be completely global and opaque; the consumer will keep opting into their ecosystems (there are large parts of the modern world you can't now access without a smartphone). And it's not necessarily a bad thing. Governments are responsible for almost all remaining conflict and famine in the world. But the individual will be back to having zero power. Every decision will be made for you.
One wonders how different (or not) it was in prior cycles of mass-acceptance and heady growth of new technologies. Steam engines. Railroads. Telegraphy. Later, internal combustion engines, automobiles, telephones and radio. Still later, chemicals and computers. And so on. Each time, there's a technology revolution, with a spate of entrants into the market, followed by consolidation.

The question for us as investors, is whether small-caps and mid-caps will ever again (in the reasonably near future) have their day. If AI goes the consolidation route, and consolidation favors the mega-caps, will we for the foreseeable future have a magnificent-something-or-other (never mind its exact composition or number), or will the market broaden? The actionable thing here, is whether to overweight the S&P 500, capturing the largest of the large, or whether we revert to the old "factor" thinking, that smaller --> higher CAGR (eventually).
I think just from a technology perspective, it may depend whether that consolidation proves to be top down (everyone relying on AWS and Microsoft servers for their computing power, and smaller companies feeding large), or bottom up (something more blockchain, decentralised computing, open source – where small, innovative companies wield the same power as the large).

I always assumed the latter, because the ML community is very open source. If OpenAI's doing something this year, a teenager on GitHub will be doing it 6 months later. But now the moat has become the sheer processing and energy costs of training really large networks. That could still be done in a distributed fashion. And one trend right now is that GPT's so heavily censored, free alternatives are popping up that don't have those restrictions.

On the small-caps though. I wouldn't be tempted anyway. AQR did a piece on why there probably never was a small-cap premium. I think the likelihood of relative small-cap or value outperformance would be down to valuations in mega-cap tech being in bubble territory. And Ray Dalio's recent piece on whether the market's currently in a bubble allays quite a few of those fears.

https://www.aqr.com/Insights/Perspectiv ... 20finding.
gougou
Posts: 1342
Joined: Thu Sep 28, 2017 7:42 pm

Re: Dot.Com Era/AI Era?

Post by gougou »

adave wrote: Thu Apr 04, 2024 1:26 pm Mega cap tech stocks were tested in 2022 and bounced back well from rising interest rates. It seems that these companies are money printers and thus not as effected by rising rates. They seem to be able to survive / thrive in any economic environment.

Part of me believes the ultimate endpoint for any large network is several very large players become dominant and eat the rest.

It wouldn't shock me if the the mega cap Tech companies "win" the AI cycle. There is a certain winner take all dynamic in the Tech ecosystem imo.
Is there even going to be an "AI cycle"? I'm not seeing any evidence that the AI stuff is making money. The most promising AI tech is probably self-driving and that bubble is bursting and there's no timeline when the tech will be mature enough to be widely and profitably adopted. We really don't know if this AI thing will be bigger than the Internet or if it's just another fad that will be forgotten in a few years.

The dot.com bubble started in the 90s and internet companies began to dominate about 20 years later. But it's not the big tech companies from the 90s that dominated today's market. It's too early to say who's going to dominate AI. I'd be VERY surprised if any of the tech companies today will dominate AI in the future, even if AI is as big and promising as what people think of it today.
The sillier the market’s behavior, the greater the opportunity for the business like investor.
adave
Posts: 592
Joined: Thu Sep 13, 2007 7:17 pm
Location: Houston

Re: Dot.Com Era/AI Era?

Post by adave »

I think it is a good question, but the Tech giants of today have huge amounts of cash and it seems they are investing multi-billions into AI. I doubt they are doing this without the expectations of an enormous pay off down the road.

One possibility is the benefits of AI will be limited in terms of consumer facing applications but be a huge benefit to the internal functioning of these companies, leading to massive productivity gains and thus profits. Using AI to design better chips, target advertising, optimize cloud computing etc.

Of course, our index funds should see some of this benefit as well hopefully!
KlangFool
Posts: 31648
Joined: Sat Oct 11, 2008 12:35 pm

Re: Dot.Com Era/AI Era?

Post by KlangFool »

Folks,

The other problem about leading in this area are the huge upfront cost. And, it will be replaced with cheaper and faster system in a short time. So, whoever go first has a limited time to make their money. Whoever come next will be able to do the same with a significant cost advantage. And, all this has to do with capital expenditure. If someone can do the same with 50% less cost in one year, what happened to all those older equipment? What happened to the first movers that spent all those money? They will be wasted.

KlangFool
30% VWENX | 16% VFWAX/VTIAX | 14.5% VTSAX | 19.5% VBTLX | 10% VSIAX/VTMSX/VSMAX | 10% VSIGX| 30% Wellington 50% 3-funds 20% Mini-Larry
madbrain
Posts: 7177
Joined: Thu Jun 09, 2011 5:06 pm
Location: San Jose, California

Re: Dot.Com Era/AI Era?

Post by madbrain »

Logan Roy wrote: Fri Mar 29, 2024 5:50 pm Self-driving's slowly being worked in – as it should be. You want to get the issues ironed out at the small-scale. It's being used commercially in the UK, in Milton Keynes, in Singapore. People saw the potential a long time ago – it just takes the technology a while to catch up. What we might see as a 'failure' is probably more an effect of the media hyping things, then getting bored .. Autonomous vehicles haven't gone away - they'll come along as the pace they're meant to. There's no denying the commercial value.
Of course there is commercial value. But I'm not sure that you can just go from small scale to large scale easily. Apparently, some self-driving cars just cannot operate without a cell signal.

https://futurism.com/the-byte/self-driv ... m-wireless

This is a major design issue, that makes these cars useless for traveling in any area without a signal, or during a network outage or congestion. There may be strong cell signals in the small areas these cars are being tested in, but that will not be true when they need to venture outside of city limits. There are plenty of areas without any cell signal, some for dozens of miles even in the bay area, if you need to go to the beach.

If these cars really need a strong signal, they would need a satellite antenna. But even that would not work in tunnels or in areas with tall trees. SiriusXM often breaks up on CA SR-17 because of these issues.

Many roads are unmarked, also. There just aren't any lines in most of the streets in my neighborhood, including the one my home is on, and multiple adjacent streets. This poses significant challenges for self-driving cars. I just checked Google earth, and only about 1% of the streets in the foothills have markings.

I really want self-driving cars to work, as I have vision issues that affect night driving. But when you combine low/no cell signals, unmarked roads, and low visibility at night, all of which apply, I'm not at all confident that we will see a self-driving car that will be useful to me in my lifetime.
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

madbrain wrote: Thu Apr 04, 2024 7:53 pm
Logan Roy wrote: Fri Mar 29, 2024 5:50 pm Self-driving's slowly being worked in – as it should be. You want to get the issues ironed out at the small-scale. It's being used commercially in the UK, in Milton Keynes, in Singapore. People saw the potential a long time ago – it just takes the technology a while to catch up. What we might see as a 'failure' is probably more an effect of the media hyping things, then getting bored .. Autonomous vehicles haven't gone away - they'll come along as the pace they're meant to. There's no denying the commercial value.
Of course there is commercial value. But I'm not sure that you can just go from small scale to large scale easily. Apparently, some self-driving cars just cannot operate without a cell signal.

https://futurism.com/the-byte/self-driv ... m-wireless

This is a major design issue, that makes these cars useless for traveling in any area without a signal, or during a network outage or congestion. There may be strong cell signals in the small areas these cars are being tested in, but that will not be true when they need to venture outside of city limits. There are plenty of areas without any cell signal, some for dozens of miles even in the bay area, if you need to go to the beach.

If these cars really need a strong signal, they would need a satellite antenna. But even that would not work in tunnels or in areas with tall trees. SiriusXM often breaks up on CA SR-17 because of these issues.

Many roads are unmarked, also. There just aren't any lines in most of the streets in my neighborhood, including the one my home is on, and multiple adjacent streets. This poses significant challenges for self-driving cars. I just checked Google earth, and only about 1% of the streets in the foothills have markings.

I really want self-driving cars to work, as I have vision issues that affect night driving. But when you combine low/no cell signals, unmarked roads, and low visibility at night, all of which apply, I'm not at all confident that we will see a self-driving car that will be useful to me in my lifetime.
Well I think you'd have to say we're already there. Just that it's going to be localised for a while. So apparently "In the United States, commuters can hail AV rides using their mobile devices in Phoenix, Arizona and in San Francisco, California. In China, AVs have been licensed to ferry commercial passengers in Chongqing and Wuhan since August 2022.", and Milton Keynes, UK, has various AV services and trials.

It may even be that for a region to be licensed for AVs, the infrastructure has to be in place for (say) 5G and a certain standard of road markings. And then at some point you'll get AVs using motorways, and that'll connect cities, and then there'll be a tipping point, where only certain areas aren't licensed for AVs. But I would expect there to be an infrastructure aspect

I think the most interesting aspect is a traffic hive mind – a centralised route planner could ensure every road's used at optimum efficiency; in principle, they could drive on motorways practically bumper to bumper – like a big, modular train. And of course they could get you around cities, and park themselves outside cities. It could transform how we get around cities like London and NY.
madbrain
Posts: 7177
Joined: Thu Jun 09, 2011 5:06 pm
Location: San Jose, California

Re: Dot.Com Era/AI Era?

Post by madbrain »

Logan Roy wrote: Fri Apr 05, 2024 12:48 pm Well I think you'd have to say we're already there.
Actually we're clearly not there at all. Everything you wrote after that sentence is about the future, not the present.

I look forward to the day an AI can drive better than a human, in the locations I need it, not some arbitrary subset of locations subject to the whims of for-profit network operators. The technology today certainly isn't up to that task and the design is not even fit for purpose.
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

madbrain wrote: Fri Apr 05, 2024 5:19 pm
Logan Roy wrote: Fri Apr 05, 2024 12:48 pm Well I think you'd have to say we're already there.
Actually we're clearly not there at all. Everything you wrote after that sentence is about the future, not the present.

I look forward to the day an AI can drive better than a human, in the locations I need it, not some arbitrary subset of locations subject to the whims of for-profit network operators. The technology today certainly isn't up to that task and the design is not even fit for purpose.
No, that whole paragraph is about the here and now. That's quite clear. Where I live, we've got AV taxis, home deliveries and several AV trials ongoing. So you can't say the technology isn't *here*.
madbrain
Posts: 7177
Joined: Thu Jun 09, 2011 5:06 pm
Location: San Jose, California

Re: Dot.Com Era/AI Era?

Post by madbrain »

Logan Roy wrote: Fri Apr 05, 2024 6:12 pm No, that whole paragraph is about the here and now. That's quite clear. Where I live, we've got AV taxis, home deliveries and several AV trials ongoing. So you can't say the technology isn't *here*.
It certainly isn't *here* where I live. And I believe that is for technological reasons.
A self-driving car is useless if it can only go from point C to point D, IMO.
Logan Roy
Posts: 1891
Joined: Sun May 29, 2022 10:15 am

Re: Dot.Com Era/AI Era?

Post by Logan Roy »

madbrain wrote: Fri Apr 05, 2024 6:24 pm
Logan Roy wrote: Fri Apr 05, 2024 6:12 pm No, that whole paragraph is about the here and now. That's quite clear. Where I live, we've got AV taxis, home deliveries and several AV trials ongoing. So you can't say the technology isn't *here*.
It certainly isn't *here* where I live. And I believe that is for technological reasons.
A self-driving car is useless if it can only go from point C to point D, IMO.
This is what I was getting at .. (e.g.) In the early days, you could only get a mobile phone signal in certain built up areas – and gradually the infrastructure rolled out to cover most areas .. With AVs, it may be quite similar.

And it's not useless .. If AVs only operate as a shuttle service within metropolitan areas, it could still be transformative for easing congestion, taking strain off public transport, home deliveries, tourism, parking, etc.
madbrain
Posts: 7177
Joined: Thu Jun 09, 2011 5:06 pm
Location: San Jose, California

Re: Dot.Com Era/AI Era?

Post by madbrain »

Logan Roy wrote: Fri Apr 05, 2024 7:01 pm And it's not useless .. If AVs only operate as a shuttle service within metropolitan areas, it could still be transformative for easing congestion, taking strain off public transport, home deliveries, tourism, parking, etc.
All of the above can be performed with manned vehicles. They don't require autonomous vehicles. AVs may reduce costs if they work as well or better than human drivers. Cost is a very important issue for people with disabilities who cannot drive, and IMO, serving them would be the most transformative impact, if AVs can lower costs sufficiently to do that. But that will be all for naught if they are limited to certain areas, either for technical or commercial reasons. It certainly is useless to me if not available in my area, which it isn't currently.

I still cannot get a decent cell signal inside my home from any operator. I have been on a waiting list for fiber for 13 years. There is cable. But I'm switching to fixed wireless on my roof next week due to Comcast contract shenanigans.
billaster
Posts: 3022
Joined: Wed Apr 27, 2022 2:21 pm

Re: Dot.Com Era/AI Era?

Post by billaster »

Logan Roy wrote: Fri Apr 05, 2024 7:01 pm And it's not useless .. If AVs only operate as a shuttle service within metropolitan areas, it could still be transformative for easing congestion, taking strain off public transport, home deliveries, tourism, parking, etc.
That's what they claimed about ride-hailing apps like Uber and Lyft -- transformative!, disruptive!, innovative!. Reality was just the opposite. They increased congestion.
madbrain
Posts: 7177
Joined: Thu Jun 09, 2011 5:06 pm
Location: San Jose, California

Re: Dot.Com Era/AI Era?

Post by madbrain »

billaster wrote: Fri Apr 05, 2024 7:39 pm That's what they claimed about ride-hailing apps like Uber and Lyft -- transformative!, disruptive!, innovative!. Reality was just the opposite. They increased congestion.
+1 . Thus has been documented. Thus is partly due to too many drivers chasing the same passengers in high density areas. With AVs, one could hope for a more rational geographical distribution of vehicles. Assuming the algorithms are written properly ...
User avatar
HanSolo
Posts: 2377
Joined: Thu Jul 19, 2012 3:18 am

Re: Dot.Com Era/AI Era?

Post by HanSolo »

madbrain wrote: Fri Apr 05, 2024 8:04 pm Assuming the algorithms are written properly ...
One might check to see whether they're the same "brogrammers" who wrote the code for Boeing's MCAS (or, alternatively, are drinking equivalent amounts of beer while coding).
Strategic Macro Senior (top 1%, 2019 Bogleheads Contest)
case_of_ennui
Posts: 302
Joined: Mon Aug 17, 2020 5:07 pm
Location: Montana

Re: Dot.Com Era/AI Era?

Post by case_of_ennui »

HanSolo wrote: Sat Apr 06, 2024 6:48 am
madbrain wrote: Fri Apr 05, 2024 8:04 pm Assuming the algorithms are written properly ...
One might check to see whether they're the same "brogrammers" who wrote the code for Boeing's MCAS (or, alternatively, are drinking equivalent amounts of beer while coding).
I thought the problem with MCAS was it only used a single angle of attack sensor? With a lack of redundancy a faulty nose high reading from the sensor caused MCAS to trim the plane nose down. That's a design flaw/excessive cost cutting not any problem with the code.
User avatar
HanSolo
Posts: 2377
Joined: Thu Jul 19, 2012 3:18 am

Re: Dot.Com Era/AI Era?

Post by HanSolo »

case_of_ennui wrote: Sat Apr 06, 2024 7:16 am
HanSolo wrote: Sat Apr 06, 2024 6:48 am
madbrain wrote: Fri Apr 05, 2024 8:04 pm Assuming the algorithms are written properly ...
One might check to see whether they're the same "brogrammers" who wrote the code for Boeing's MCAS (or, alternatively, are drinking equivalent amounts of beer while coding).
I thought the problem with MCAS was it only used a single angle of attack sensor? With a lack of redundancy a faulty nose high reading from the sensor caused MCAS to trim the plane nose down. That's a design flaw/excessive cost cutting not any problem with the code.
There are reasons why we have a word like "brogrammer". Those reasons might apply not only to coders but also other people involved. As for MCAS, whether it was the code or something else, poor engineering is still poor engineering. And I can't prove it was caused by beer, but the fact remains that someone screwed up, and we don't know whether or not similar kinds of people are working on AV (in whatever job functions). Hence, the "assuming" in the post I quoted remains just that, an assumption.
Strategic Macro Senior (top 1%, 2019 Bogleheads Contest)
User avatar
firebirdparts
Posts: 4447
Joined: Thu Jun 13, 2019 4:21 pm
Location: Southern Appalachia

Re: Dot.Com Era/AI Era?

Post by firebirdparts »

billaster wrote: Fri Apr 05, 2024 7:39 pm
Logan Roy wrote: Fri Apr 05, 2024 7:01 pm And it's not useless .. If AVs only operate as a shuttle service within metropolitan areas, it could still be transformative for easing congestion, taking strain off public transport, home deliveries, tourism, parking, etc.
That's what they claimed about ride-hailing apps like Uber and Lyft -- transformative!, disruptive!, innovative!. Reality was just the opposite. They increased congestion.
Well, now, wait a minute. In the US, taxi service is very ripe for disruption. Not because of congestion. If you're a taxi consumer, you want more congestion by taxis where you happen to be standing. but because of the price. It was at extortion level.

There's no reason to think of the programming doing anything other than informing an army of independent contractors that don't much care what they're being paid. That's enough. All it needs to do. you don't need intelligence, artificial or the regular kind, to come up with competing against a bunch of price-fixing racketeers. Programming-wise, it's good enough. Seems like it anyway.
This time is the same
gougou
Posts: 1342
Joined: Thu Sep 28, 2017 7:42 pm

Re: Dot.Com Era/AI Era?

Post by gougou »

adave wrote: Thu Apr 04, 2024 7:22 pm I think it is a good question, but the Tech giants of today have huge amounts of cash and it seems they are investing multi-billions into AI. I doubt they are doing this without the expectations of an enormous pay off down the road.

One possibility is the benefits of AI will be limited in terms of consumer facing applications but be a huge benefit to the internal functioning of these companies, leading to massive productivity gains and thus profits. Using AI to design better chips, target advertising, optimize cloud computing etc.

Of course, our index funds should see some of this benefit as well hopefully!
The tech giants of today are incumbents not disruptors. They are spending huge capex on AI to maintain their market positions. If Google doesn’t spend enough on the AI arms race Microsoft might develop something to steal Google’s market.

But Google seems to be a loser in this race regardless. Google can put 10 ads on a search page but it’s hard to put any ad in an AI chatbot. It looks like the tech giants are spending 100s of billions $$$ to develop something that will make the search ads market smaller.

I think the competition is beneficial to the society. They’ll probably make it easier and more accessible for people to search for information. I don’t know about the stocks because those are monopolistic companies suddenly facing competition and possible disruption and huge capex requirements.
The sillier the market’s behavior, the greater the opportunity for the business like investor.
User avatar
nisiprius
Advisory Board
Posts: 52348
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Dot.Com Era/AI Era?

Post by nisiprius »

Logan Roy wrote: Fri Apr 05, 2024 12:48 pm...It may even be that for a region to be licensed for AVs, the infrastructure has to be in place for (say) 5G and a certain standard of road markings...
This.

And road "markings" will probably involve non-visual guidance, buried wires that will be "visible" through a light dusting of snow. And virtual "markings" in the form of GPS maps with resolution down to individual lane closures, updated hourly.

It's what happened with cars in the first place. Cars were hobbyist toys until the Good Roads Movement, 1870-1920, sparked first by bicyclists and then by motorists, successfully lobbied for public investment in road infrastructure. Even by the 1920s, the official map for the Lincoln Highway said that, coast to coast, "you can make the trip in 20 to 30 days driving approximately 10 hours a day." Then there were public policy decisions, e.g. Robert Moses managing to get an extraordinary proportion of all New York city and state expenditures allocated to roads, roads, parks, bridges, roads, roads, bridges, roads, and roads. And then, later, the Interstate Highway System.

It wasn't because the cars got better, it's because the roads got better.

And that's what will be needed for self-driving cars.
Last edited by nisiprius on Sat Apr 06, 2024 1:24 pm, edited 2 times in total.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
madbrain
Posts: 7177
Joined: Thu Jun 09, 2011 5:06 pm
Location: San Jose, California

Re: Dot.Com Era/AI Era?

Post by madbrain »

HanSolo wrote: Sat Apr 06, 2024 6:48 am One might check to see whether they're the same "brogrammers" who wrote the code for Boeing's MCAS (or, alternatively, are drinking equivalent amounts of beer while coding).
I just hope they aren't the same ones that came up with "surge pricing" for ride sharing. At least the number of AVs can be controlled/limited to actual demand by the system as opposed to the mad rush of drivers that happens today at certain times.
madbrain
Posts: 7177
Joined: Thu Jun 09, 2011 5:06 pm
Location: San Jose, California

Re: Dot.Com Era/AI Era?

Post by madbrain »

gougou wrote: Sat Apr 06, 2024 1:10 pm Google can put 10 ads on a search page but it’s hard to put any ad in an AI chatbot.
It's not hard to come up with ways to do that. I don't want to give them any ideas, so I'm not going to write what they are. But I think Google employs sufficiently smart people to figure it out.
madbrain
Posts: 7177
Joined: Thu Jun 09, 2011 5:06 pm
Location: San Jose, California

Re: Dot.Com Era/AI Era?

Post by madbrain »

nisiprius wrote: Sat Apr 06, 2024 1:20 pm It wasn't because the cars got better, it's because the roads got better.

And that's what will be needed for self-driving cars.
Given the state of some roads in California, that would point to a very difficult path for self-driving cars ahead. There are policy issues that are hard to resolve. The funding just hasnt kept up with inflation.
Then, there are local streets. I fought with my city's transportation department for years to get the street resurfaced. Finally succeeded last year. It hadn't been redone for at least the 13 years since I moved in. The city didn't have any records of the last time it was done prior to that. And of course, no street markings. Same as all streets in the neighborhood, except one.
Post Reply