Be careful of AI investing hype

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Topic Author
nisiprius
Advisory Board
Posts: 52216
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

Be careful of AI investing hype

Post by nisiprius »

I've just started William J. Bernstein's The Delusions Of Crowds: Why People Go Mad in Groups, which looks as if it is going to draw parallels between investing bubbles and end-of-the-world manias. It seems particularly apropos in the light of the AI-world talk of "the Singularity," when AI will become smarter than humans, resulting in The End Of The World As We Know It. I think AI combines a financial craze and an end-of-the-world fears in one neat package.

That's not to say new developments in AI won't be useful and make money. I believe IBM's Watson is finding limited uses and makes money for IBM. But it wasn't the Singularity, any more than an IBM computer beating a checkers champion in 1962.

Today's Washington Post reports what I should have guessed.

The tech industry was deflating. Then came ChatGPT.
Last year, Silicon Valley was drowning in layoffs and dour predictions. Artificial intelligence made the gloom go away.

SAN FRANCISCO — A year ago, the mood in Silicon Valley was dour. Big Tech stocks were falling, the cryptocurrency bubble had popped, and a wave of layoffs was beginning to sweep through the industry.

Then the artificial intelligence boom hit.

Since then, venture capitalists have been throwing money at AI start-ups, investing over $11 billion in May alone, according to data firm PitchBook, an increase of 86 percent over the same month last year. Companies from Moderna to Heinz have mentioned AI initiatives on recent earnings calls. And last week, AI chipmaker Nvidia became one of only a handful of companies in the world to hit $1 trillion in value.

In San Francisco, it’s suddenly impossible to escape the AI hysteria. In bars and restaurants, people are conversing about using ChatGPT and whether AI will take their jobs — or take over the world....
So Silicon Valley desperately wants large language models to be The New Big Thing.

The point is that as investors we should resist the temptation to do anything in particular. I think it's a bubble. Like all bubbles, yes, some people will get rich by making high-stakes bets, and it won't be me. But the title of Spencer Jakab's The Revolution That Wasn't: GameStop, Reddit, and the Fleecing of Small Investors speaks for itself. I was also impressed by a Rational Reminder podcast transcript, Lighting your Money on Fire with Thematic ETFs

At the start of 2021, not being a gamer, I hardly knew what nVIDIA was. Yet it has added about a percentage point to my Total Stock returns.

Meanwhile, I think it speaks volumes that both ARK Autonomous Tech. & Robotics ETF, ARKQ, and the iShares Robotics and Artificial Intelligence Multisector ETF, IRBO--blue lines--have lost money for anyone who invested in them since 2021--despite holding nVIDIA. They didn't hold enough of it--or soon enough--or enough of anything else--to make up for previous losses. There's no big upward spike at the end. A boring old S&P 500 index fund (red line), with a quarter of a teaspoon of AI, has outperformed both.

Image

(Note: I used a price chart in order to include recent daily values, but the picture is the same if you include dividends.)
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.
adave
Posts: 567
Joined: Thu Sep 13, 2007 7:17 pm
Location: Houston

Re: Be careful of AI investing hype

Post by adave »

NVDA is an interesting case.

It’s more like an AI infrastructure company.

Turns out GPUs are ideal for the type of compute required in AI and NVDA has a dominant position in this market.

I see no harm in holding a small portion just to go along with on the ride.
valleyrock
Posts: 1129
Joined: Sun Aug 12, 2018 7:12 am

Re: Be careful of AI investing hype

Post by valleyrock »

Heinz is using AI? Are they going to apply "AI" to photosynthesis? I don't think so.

One problem with all this hype is the failure to understand biology. Tech in general is linear. One line of code leads to the next, to put it simply. But when you throw in the nonlinearity of biology, things just don't work according to those linear perspectives. Oh, and us folks, we're animals, biological species, which throws a monkey wrench into the whole shebang.

When AI can produce ketchup in a replicator from its constituent atoms, now then we'll have something. Until then, AI is basically another form of vaporware, software that's promised, but never shows up. Maybe we need another term instead of vaporware. How about "vaporAI"? Goodness me, I have the vapors!
Nowizard
Posts: 4842
Joined: Tue Oct 23, 2007 5:33 pm

Re: Be careful of AI investing hype

Post by Nowizard »

Is it be careful because these stocks are inflated and unlikely to continue providing outsized returns, or is it that they are excellent companies that may revert to a lower level of acceptable return over time? The decision could lead to market timing involving selling shares owned and repurchasing if they drop significantly but continue to produce positive returns.

Tim
User avatar
burritoLover
Posts: 4097
Joined: Sun Jul 05, 2020 12:13 pm

Re: Be careful of AI investing hype

Post by burritoLover »

I find it odd that the people who buy into hype (FOMO) appear to be the same people who typically panic sell. If you are that nervous when times are bad, you'd think you'd stay conservative in a boring portfolio all the time.
User avatar
rob
Posts: 5247
Joined: Mon Feb 19, 2007 5:49 pm
Location: Here

Re: Be careful of AI investing hype

Post by rob »

There is no "AI"... it's a nonsense term for a set of algorithms (that are less deterministic than the past). Those algorithms more easily allow functions that appear like "magic" in some ways but are simply the application of maths and computing power. Part of the issue is the black box nature to algorithms.

I agree it's complete hype right now - hopefully it will quickly go back to what it should be and just improved outcomes from the application of algorithms.
| Rob | Its a dangerous business going out your front door. - J.R.R.Tolkien
User avatar
SimpleGift
Posts: 4477
Joined: Tue Feb 08, 2011 2:45 pm

Re: Be careful of AI investing hype

Post by SimpleGift »

nisiprius wrote: Mon Jun 05, 2023 8:05 am The point is that as investors we should resist the temptation to do anything in particular.
Couldn’t agree more. The hype around AI today is reminiscent of the Gartner hype cycle for new technologies:
The advent of ChatGPT in early 2023 is likely the technology innovation trigger for this hype cycle. Broad market index investors should eventually benefit from AI's widespread adoption and its "plateau of productivity."
Valuethinker
Posts: 49035
Joined: Fri May 11, 2007 11:07 am

Re: Be careful of AI investing hype

Post by Valuethinker »

valleyrock wrote: Mon Jun 05, 2023 8:30 am Heinz is using AI? Are they going to apply "AI" to photosynthesis? I don't think so.

One problem with all this hype is the failure to understand biology. Tech in general is linear. One line of code leads to the next, to put it simply. But when you throw in the nonlinearity of biology, things just don't work according to those linear perspectives. Oh, and us folks, we're animals, biological species, which throws a monkey wrench into the whole shebang.

When AI can produce ketchup in a replicator from its constituent atoms, now then we'll have something. Until then, AI is basically another form of vaporware, software that's promised, but never shows up. Maybe we need another term instead of vaporware. How about "vaporAI"? Goodness me, I have the vapors!
Consider how good natural language translation in Google has become-- a problem once deemed unsolvable. This is big stuff.

LLMs are capable of being non-linear. That's the whole point, they are emergent thinkers - they can do things their designers have not appreciated.

I don't have a good sense of how much is hype & how much is capability. I take notice of the fact that many of the initiators of this work (Geoffrey Hinton in particular) have publicly called for a halt until we figure out what the implications are and what the best controls are. They think it's serious enough to want to pause it.

So far what is publicly available seems a curiousity. But I don't imagine for one minute that's what is going on in the labs - that we are seeing anything like the state of the art.
strummer6969
Posts: 978
Joined: Tue Dec 15, 2020 10:59 am

Re: Be careful of AI investing hype

Post by strummer6969 »

adave wrote: Mon Jun 05, 2023 8:30 am NVDA is an interesting case.

It’s more like an AI infrastructure company.

Turns out GPUs are ideal for the type of compute required in AI and NVDA has a dominant position in this market.

I see no harm in holding a small portion just to go along with on the ride.
If you hold an index fund, you already have a decent portion in it. NVIDIA is 2.69% of the S&P.
JonFund
Posts: 197
Joined: Mon Feb 01, 2021 2:51 pm

Re: Be careful of AI investing hype

Post by JonFund »

Historically, new innovations in technology has spurred a strong rally in related stocks, followed by a sizeable pullback, and then over time, gradual growth. Look at the internet boom in 1998-2000; all a company had to do was add ".com" after their company name and the stock soared. Then, those stocks crashed but the winners came back slowly and consistently over time and did well. Ditto for 3D printing a few years ago. A few stocks soared, then crashed, but now the technology is growing at a healthy rate. Artificial Intelligence is no different. Over the last few weeks if your company could be even remotely connected to AI technology, the stock did well. This too shall correct, and then follow a healthy growth pattern. All the more reason to stay diversified over the long term and avoid the short-term "noise".
toddthebod
Posts: 5737
Joined: Wed May 18, 2022 12:42 pm

Re: Be careful of AI investing hype

Post by toddthebod »

As an accumulating investor who rebalances, I love stock market bubbles!
valleyrock
Posts: 1129
Joined: Sun Aug 12, 2018 7:12 am

Re: Be careful of AI investing hype

Post by valleyrock »

Valuethinker wrote: Mon Jun 05, 2023 9:44 am
valleyrock wrote: Mon Jun 05, 2023 8:30 am Heinz is using AI? Are they going to apply "AI" to photosynthesis? I don't think so.

One problem with all this hype is the failure to understand biology. Tech in general is linear. One line of code leads to the next, to put it simply. But when you throw in the nonlinearity of biology, things just don't work according to those linear perspectives. Oh, and us folks, we're animals, biological species, which throws a monkey wrench into the whole shebang.

When AI can produce ketchup in a replicator from its constituent atoms, now then we'll have something. Until then, AI is basically another form of vaporware, software that's promised, but never shows up. Maybe we need another term instead of vaporware. How about "vaporAI"? Goodness me, I have the vapors!
Consider how good natural language translation in Google has become-- a problem once deemed unsolvable. This is big stuff.

LLMs are capable of being non-linear. That's the whole point, they are emergent thinkers - they can do things their designers have not appreciated.

I don't have a good sense of how much is hype & how much is capability. I take notice of the fact that many of the initiators of this work (Geoffrey Hinton in particular) have publicly called for a halt until we figure out what the implications are and what the best controls are. They think it's serious enough to want to pause it.

So far what is publicly available seems a curiousity. But I don't imagine for one minute that's what is going on in the labs - that we are seeing anything like the state of the art.
Here's an interesting perspective on AI and the hype: https://www.theverge.com/23604075/ai-ch ... irror-test
adave
Posts: 567
Joined: Thu Sep 13, 2007 7:17 pm
Location: Houston

Re: Be careful of AI investing hype

Post by adave »

Whether or not these things are truly "intelligent" or ever become so almost doesn't matter if they end up massively increasing productivity across multiple industries. There is potential of massive economic impacts of these Bots even of they simply function as co-pilots and super-charge human productivity. In this case, the stocks will continue to do well imo.
User avatar
bertilak
Posts: 10725
Joined: Tue Aug 02, 2011 5:23 pm
Location: East of the Pecos, West of the Mississippi

Re: Be careful of AI investing hype

Post by bertilak »

JonFund wrote: Mon Jun 05, 2023 10:01 am Historically, new innovations in technology has spurred a strong rally in related stocks, followed by a sizeable pullback, and then over time, gradual growth. Look at the internet boom in 1998-2000; all a company had to do was add ".com" after their company name and the stock soared. Then, those stocks crashed but the winners came back slowly and consistently over time and did well. Ditto for 3D printing a few years ago. A few stocks soared, then crashed, but now the technology is growing at a healthy rate. Artificial Intelligence is no different. Over the last few weeks if your company could be even remotely connected to AI technology, the stock did well. This too shall correct, and then follow a healthy growth pattern. All the more reason to stay diversified over the long term and avoid the short-term "noise".
Agreed, with an additional observation:

The winners, those who defined the healthy growth pattern, were not easily identifiable in the early phases.

That is another point in favor broad diversification. I define broad as cap-weighted because that focuses on where the money is going (capitalization) and how well it does when it gets there. Money invested means action as opposed to talk.

I agree that there is room (and need) for fundamental analysis and even speculation, but that's all above my pay grade. Those whose pay grade does justify such action can fight it out amongst themselves and I will go with the results, as defined by the capitalization.
May neither drought nor rain nor blizzard disturb the joy juice in your gizzard. -- Squire Omar Barker (aka S.O.B.), the Cowboy Poet
ThankYouJack
Posts: 5704
Joined: Wed Oct 08, 2014 7:27 pm

Re: Be careful of AI investing hype

Post by ThankYouJack »

Isn't it "anti-Boglehead" to predict bubbles? If people currently feel confident about AI popping, why not invest accordingly?
User avatar
bertilak
Posts: 10725
Joined: Tue Aug 02, 2011 5:23 pm
Location: East of the Pecos, West of the Mississippi

Re: Be careful of AI investing hype

Post by bertilak »

ThankYouJack wrote: Mon Jun 05, 2023 10:49 am Isn't it "anti-Boglehead" to predict bubbles? If people currently feel confident about AI popping, why not invest accordingly?
As I implied in my post just above, I use capitalization as a confidence measure.
May neither drought nor rain nor blizzard disturb the joy juice in your gizzard. -- Squire Omar Barker (aka S.O.B.), the Cowboy Poet
Teague
Posts: 2524
Joined: Wed Nov 04, 2015 5:15 pm

Re: Be careful of AI investing hype

Post by Teague »

valleyrock wrote: Mon Jun 05, 2023 8:30 am Heinz is using AI? Are they going to apply "AI" to photosynthesis? I don't think so.

One problem with all this hype is the failure to understand biology. Tech in general is linear. One line of code leads to the next, to put it simply. But when you throw in the nonlinearity of biology, things just don't work according to those linear perspectives. Oh, and us folks, we're animals, biological species, which throws a monkey wrench into the whole shebang.

When AI can produce ketchup in a replicator from its constituent atoms, now then we'll have something. Until then, AI is basically another form of vaporware, software that's promised, but never shows up. Maybe we need another term instead of vaporware. How about "vaporAI"? Goodness me, I have the vapors!
I imagine Heinz wants to be an early adopter so they can be an industry leader and aren't forever stuck playing ketchup.
Semper Augustus
freyj6
Posts: 554
Joined: Thu Jun 19, 2014 2:09 am

Re: Be careful of AI investing hype

Post by freyj6 »

burritoLover wrote: Mon Jun 05, 2023 9:24 am I find it odd that the people who buy into hype (FOMO) appear to be the same people who typically panic sell. If you are that nervous when times are bad, you'd think you'd stay conservative in a boring portfolio all the time.
Even more interestingly, it seems to be many of the same people who bought into the last few hypes. ARKK, coins, NFTs, now AI.
strummer6969
Posts: 978
Joined: Tue Dec 15, 2020 10:59 am

Re: Be careful of AI investing hype

Post by strummer6969 »

burritoLover wrote: Mon Jun 05, 2023 9:24 am I find it odd that the people who buy into hype (FOMO) appear to be the same people who typically panic sell. If you are that nervous when times are bad, you'd think you'd stay conservative in a boring portfolio all the time.
I remember several threads last year about rotating from growth to value because the talking heads were saying value will do better in a rising rate environment. People were dumping highly profitable mega cap tech stocks just because they have some arbitrary 'growth' label attached to them. I had a feeling this would happen, and I should've taken it as a cue to do the opposite :twisted:
User avatar
HomerJ
Posts: 21282
Joined: Fri Jun 06, 2008 12:50 pm

Re: Be careful of AI investing hype

Post by HomerJ »

Teague wrote: Mon Jun 05, 2023 10:55 am
valleyrock wrote: Mon Jun 05, 2023 8:30 am Heinz is using AI? Are they going to apply "AI" to photosynthesis? I don't think so.

One problem with all this hype is the failure to understand biology. Tech in general is linear. One line of code leads to the next, to put it simply. But when you throw in the nonlinearity of biology, things just don't work according to those linear perspectives. Oh, and us folks, we're animals, biological species, which throws a monkey wrench into the whole shebang.

When AI can produce ketchup in a replicator from its constituent atoms, now then we'll have something. Until then, AI is basically another form of vaporware, software that's promised, but never shows up. Maybe we need another term instead of vaporware. How about "vaporAI"? Goodness me, I have the vapors!
I imagine Heinz wants to be an early adopter so they can be an industry leader and aren't forever stuck playing ketchup.
Heh, I wonder if a chatbot could have managed that sentence... :sharebeer
"The best tools available to us are shovels, not scalpels. Don't get carried away." - vanBogle59
Logan Roy
Posts: 1838
Joined: Sun May 29, 2022 10:15 am

Re: Be careful of AI investing hype

Post by Logan Roy »

rob wrote: Mon Jun 05, 2023 9:27 am There is no "AI"... it's a nonsense term for a set of algorithms (that are less deterministic than the past). Those algorithms more easily allow functions that appear like "magic" in some ways but are simply the application of maths and computing power. Part of the issue is the black box nature to algorithms.

I agree it's complete hype right now - hopefully it will quickly go back to what it should be and just improved outcomes from the application of algorithms.
It's not clear that human intelligence is anything different. I work in machine learning, and while human intelligence feels a certain way, there's no magic there either. ML is really just a brute force calculus solver that can be applied in ways that makes things appear intelligent.

What *might* be magic, though – in both cases – is emergent behaviour.. Emergence is when systems of far more sophistication or complexity form from much simpler systems. So ant and termite behaviour can be very sophisticated across a colony, where lots of simple systems are interacting together. Langton's ant is a good computer equivalent.. I sound like I'm getting ahead of myself with GPT, but having worked with it so much, I'm veering towards thinking there's something resembling real intelligence in there, that probably emerges somewhere up the processing hierarchy. You only get that sense working with it on complex problems.
valleyrock
Posts: 1129
Joined: Sun Aug 12, 2018 7:12 am

Re: Be careful of AI investing hype

Post by valleyrock »

adave wrote: Mon Jun 05, 2023 10:40 am Whether or not these things are truly "intelligent" or ever become so almost doesn't matter if they end up massively increasing productivity across multiple industries. There is potential of massive economic impacts of these Bots even of they simply function as co-pilots and super-charge human productivity. In this case, the stocks will continue to do well imo.
This is a valid point. People may very well lose their jobs, like the attorney who submitted a filing written by ChatGPT with citations which the attorney did not very, and which were ficticious! The judge wasn't happy.

Some writers are apparently already losing their jobs because "AI" produces text that is considered "good enough," far less expensive to generate than paying a writer's salary.

And there are areas where it's known that huge numbers of employees are superfluous/redundant. "AI," or LLMs, may well become an "excuse" to get rid of people who, for example, shuffle paper at insurance companies and other places. When I drive by an insurance company's office, I'm amazed at the huge buildings, full of paper shufflers. (Well, not necessarily shuffling paper... I mean the digital equivalent.)

So, from these perspectives, and perhaps waxing a bit cynical as I'm wont to do, maybe one investment strategy is companies with huge numbers of white collar employees. The upcoming layoffs due to "AI" will lead to huge savings because at least half the cost of running an operation is salaries, so profits are bound to increase dramatically. There's probably already and ETF for companies that will profit from AI, not just the AI companies themselves. Some companies will profit by getting smarter, others will profit from simply replacing people with LLMs. It may very well be a whacky time.
valleyrock
Posts: 1129
Joined: Sun Aug 12, 2018 7:12 am

Re: Be careful of AI investing hype

Post by valleyrock »

Teague wrote: Mon Jun 05, 2023 10:55 am
valleyrock wrote: Mon Jun 05, 2023 8:30 am Heinz is using AI? Are they going to apply "AI" to photosynthesis? I don't think so.

One problem with all this hype is the failure to understand biology. Tech in general is linear. One line of code leads to the next, to put it simply. But when you throw in the nonlinearity of biology, things just don't work according to those linear perspectives. Oh, and us folks, we're animals, biological species, which throws a monkey wrench into the whole shebang.

When AI can produce ketchup in a replicator from its constituent atoms, now then we'll have something. Until then, AI is basically another form of vaporware, software that's promised, but never shows up. Maybe we need another term instead of vaporware. How about "vaporAI"? Goodness me, I have the vapors!
I imagine Heinz wants to be an early adopter so they can be an industry leader and aren't forever stuck playing ketchup.
Haha.

Actually, ketchup exhibits interesting rheological properties. It has to be subjected to a shear stress to get it to flow, as we know from having to shake or pound a ketchup bottle. Maybe Heinz can use AI to develop ketchup that flows at a lower shear stress, thus helping prevent over ketchuping.
Logan Roy
Posts: 1838
Joined: Sun May 29, 2022 10:15 am

Re: Be careful of AI investing hype

Post by Logan Roy »

The art of investing has always been working out who benefits and how.

The 20th century saw the birth of the mainframe; microprocessor; home computer; internet; smart phone, etc. Things that have increased human productivity and changed the way every business and individual works.. And yet, from about 1960 to 2010, Tech was the worst performing sector in the stock market.

The two things I'm happy to add to at the moment are digital infrastructure (private infrastructure: fibre optics, data storage, etc.) and the FTSE All World, because I think the only obvious beneficiary of AI is human productivity.. More interesting problems might be how deflationary it is – esp. if it's going to replace a great many jobs (I don't buy that it will create nearly as many). It may necessitate Universal Basic Income. How does that change things? AI is an open source community, so I don't think these businesses will have moats (which drive abnormal profits) around AI. I think you've got to look at which businesses can implement AI to boost margins. AI may become the only necessary capital allocator in markets – so do we still have risk premiums, when growth and inflation are stuck at 1%?
User avatar
Topic Author
nisiprius
Advisory Board
Posts: 52216
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: Be careful of AI investing hype

Post by nisiprius »

The question isn't whether the technology known as AI is "intelligent," nor whether it is useful. The question is whether we are in the grip of "AI hysteria," to use the Post's phrase, and a potential investing mania surrounding it.
adave wrote: Mon Jun 05, 2023 10:40 am Whether or not these things are truly "intelligent" or ever become so almost doesn't matter if they end up massively increasing productivity across multiple industries. There is potential of massive economic impacts of these Bots even of they simply function as co-pilots and super-charge human productivity. In this case, the stocks will continue to do well imo.
But how long will that take? Huge improvements in productivity were expected in the 1970s and 1980s from the widespread application of computer in business. For thirty years there was a "productivity paradox" because it took that long for the expected gains to actually show up.

The Internet began in 1983, and it was easy to see it as important and productivity-enhancing. When commercial activity became permitted in 1993, there was hype, following by the spectacular tech bust in 2000. We buy our pet food over the Internet today (from chewy.com), but we had to wait twenty years after the collapse of pets.com for the promise of "pet food over the internet" to be realized.

You say "the stocks" will "continue" to do well, but "AI" is going to have its share of Osborne Computer Companies and Webvans and pets.com along the way. To say nothing of near-scams (companies making AI noises without really being seriously in the business... like the Long Island Ice Tea company rebranding itself is the Long Blockchain company to take advantage of crypto mania but mysteriously never getting around to actually buying crypto mining rigs).
Last edited by nisiprius on Mon Jun 05, 2023 11:30 am, 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.
User avatar
K72
Posts: 440
Joined: Wed Dec 05, 2018 7:04 pm

Re: Be careful of AI investing hype

Post by K72 »

So far I'm not impressed with ChatGPT. Both DW and I have dabbled, though not for investing (her for a book she is writing and me for Python programming). In every case, the result looked impressive on the surface but was utter nonsense. Won't catch me anytime soon trying ChatGPT for investing.
Last edited by K72 on Mon Jun 05, 2023 11:28 am, edited 1 time in total.
All we want are the facts...
User avatar
burritoLover
Posts: 4097
Joined: Sun Jul 05, 2020 12:13 pm

Re: Be careful of AI investing hype

Post by burritoLover »

freyj6 wrote: Mon Jun 05, 2023 11:00 am
burritoLover wrote: Mon Jun 05, 2023 9:24 am I find it odd that the people who buy into hype (FOMO) appear to be the same people who typically panic sell. If you are that nervous when times are bad, you'd think you'd stay conservative in a boring portfolio all the time.
Even more interestingly, it seems to be many of the same people who bought into the last few hypes. ARKK, coins, NFTs, now AI.
Yeah, I can understand getting burned once and then taking a step back but to continue to pound your head against the wall looking for gold is just insane.
User avatar
rob
Posts: 5247
Joined: Mon Feb 19, 2007 5:49 pm
Location: Here

Re: Be careful of AI investing hype

Post by rob »

Logan Roy wrote: Mon Jun 05, 2023 11:09 am
rob wrote: Mon Jun 05, 2023 9:27 am There is no "AI"... it's a nonsense term for a set of algorithms (that are less deterministic than the past). Those algorithms more easily allow functions that appear like "magic" in some ways but are simply the application of maths and computing power. Part of the issue is the black box nature to algorithms.

I agree it's complete hype right now - hopefully it will quickly go back to what it should be and just improved outcomes from the application of algorithms.
It's not clear that human intelligence is anything different. I work in machine learning, and while human intelligence feels a certain way, there's no magic there either. ML is really just a brute force calculus solver that can be applied in ways that makes things appear intelligent.

What *might* be magic, though – in both cases – is emergent behavior.. Emergence is when systems of far more sophistication or complexity form from much simpler systems. So ant and termite behavior can be very sophisticated across a colony, where lots of simple systems are interacting together. Langton's ant is a good computer equivalent.. I sound like I'm getting ahead of myself with GPT, but having worked with it so much, I'm veering towards thinking there's something resembling real intelligence in there, that probably emerges somewhere up the processing hierarchy. You only get that sense working with it on complex problems.
Yeah, I also live in that world. I just don't see the emergence you do and I guess I see human intelligence as far more than GPT or similar, regardless of mechanically how it might be done (and I don't think anyone knows that).

Complex outcomes sure, perplexing how it got there sure but not the "I" part of "AI" IMO. I mean "magic" in the sense of a bow & arrow killing an animal over distance compared to a someone who only knows about a club or someone who can navigate in the open ocean in the age of no time-keeper, sextant or compass. It's simply a lack of that knowledge to the observer, not intelligence by the "AI" insight creator.

Edit: Added the underline - certainly didn't mean to imply a topic that would be ban... I mean the creation of the insight. Re-reading people might misunderstand my words.
Last edited by rob on Mon Jun 05, 2023 1:11 pm, edited 1 time in total.
| Rob | Its a dangerous business going out your front door. - J.R.R.Tolkien
rebellovw
Posts: 1748
Joined: Tue Aug 16, 2016 4:30 pm

Re: Be careful of AI investing hype

Post by rebellovw »

The pets.com sockpuppet will now have a brain!

Anyhow great stuff nsiprius thanks for it - really nice read. No changes on my side.
InvestingGeek
Posts: 351
Joined: Tue Aug 28, 2018 6:31 am

Re: Be careful of AI investing hype

Post by InvestingGeek »

valleyrock wrote: Mon Jun 05, 2023 8:30 am Heinz is using AI? Are they going to apply "AI" to photosynthesis? I don't think so.
This is exactly what people used to say about watching TV shows and movies on their phones, or anyone using cloud laptops like Chromebooks before they became ubiquitous. This is why innovators make boatloads of money and naysayers don't.

I don't just see tomatoes and photosynthesis when I look at Heinz. There are multiple areas where ML and LLM tech can be used by a ketchup company right from farming, food processing and manufacturing to inventory management, delivery logistics and customer service. The possibilities literally are endless.

As a Boglehead investor that doesn't know anything about the tech, it is perfectly reasonable to stick with the three-fund portfolio. But let's be honest to acknowledge that instead of ignorantly knocking the tech.
User avatar
Topic Author
nisiprius
Advisory Board
Posts: 52216
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: Be careful of AI investing hype

Post by nisiprius »

Business plan, 2023-style

1) Design a toothbrush that uses generative AI.
2) ????
3) PROFIT!!!
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.
toddthebod
Posts: 5737
Joined: Wed May 18, 2022 12:42 pm

Re: Be careful of AI investing hype

Post by toddthebod »

nisiprius wrote: Mon Jun 05, 2023 11:38 am Business plan, 2023-style

1) Design a toothbrush that uses generative AI.
2) ????
3) PROFIT!!!
https://www.usa.philips.com/a-w/about/n ... stige.html
The Philips Sonicare app is powered by artificial intelligence and works in perfect harmony with the Philips Sonicare 9900 Prestige.
KlangFool
Posts: 31526
Joined: Sat Oct 11, 2008 12:35 pm

Re: Be careful of AI investing hype

Post by KlangFool »

K72 wrote: Mon Jun 05, 2023 11:27 am So far I'm not impressed with ChatGPT. Both DW and I have dabbled, though not for investing (her for a book she is writing and me for Python programming). In every case, the result looked impressive on the surface but was utter nonsense. Won't catch me anytime soon trying ChatGPT for investing.
But, it takes intelligence to notice the utter nonsense. Perhaps, we should use ChatGPT as a new kind of "Turing Test". Something to separate folks that can notice utter nonsense versus those that can't.

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
jumbo shrimp
Posts: 104
Joined: Wed Apr 12, 2023 3:43 pm

Re: Be careful of AI investing hype

Post by jumbo shrimp »

I agree with OP. I don't know how big or small the AI bubble is, but I am not into the hype. The announcement of AI has made more waves than actual results. That's my observation. However, I do believe over time that whatever is considered AI will be used by every company as a normal day-to-day operation. It will become another productive common place tool. This is similar to cloud computing. And if companies don't use it, they are left behind. It's just I doubt the hype matches the actual roadmap.

On another note, I do like the results coming from AI.
Logan Roy
Posts: 1838
Joined: Sun May 29, 2022 10:15 am

Re: Be careful of AI investing hype

Post by Logan Roy »

rob wrote: Mon Jun 05, 2023 11:31 am
Logan Roy wrote: Mon Jun 05, 2023 11:09 am
rob wrote: Mon Jun 05, 2023 9:27 am There is no "AI"... it's a nonsense term for a set of algorithms (that are less deterministic than the past). Those algorithms more easily allow functions that appear like "magic" in some ways but are simply the application of maths and computing power. Part of the issue is the black box nature to algorithms.

I agree it's complete hype right now - hopefully it will quickly go back to what it should be and just improved outcomes from the application of algorithms.
It's not clear that human intelligence is anything different. I work in machine learning, and while human intelligence feels a certain way, there's no magic there either. ML is really just a brute force calculus solver that can be applied in ways that makes things appear intelligent.

What *might* be magic, though – in both cases – is emergent behavior.. Emergence is when systems of far more sophistication or complexity form from much simpler systems. So ant and termite behavior can be very sophisticated across a colony, where lots of simple systems are interacting together. Langton's ant is a good computer equivalent.. I sound like I'm getting ahead of myself with GPT, but having worked with it so much, I'm veering towards thinking there's something resembling real intelligence in there, that probably emerges somewhere up the processing hierarchy. You only get that sense working with it on complex problems.
Yeah, I also live in that world. I just don't see the emergence you do and I guess I see human intelligence as far more than GPT or similar, regardless of mechanically how it might be done (and I don't think anyone knows that).

Complex outcomes sure, perplexing how it got there sure but not the "I" part of "AI" IMO. I mean "magic" in the sense of a bow & arrow killing an animal over distance compared to a someone who only knows about a club or someone who can navigate in the open ocean in the age of no time-keeper, sextant or compass. It's simply a lack of that knowledge to the observer, not intelligence by the creator.
Well the other angle I come at it from is things I studied around college, like Thomas Metzinger and computational neuroscience.. And I think what the brain does boils down to the same thing as a computer neural network. And things like Stable Diffusion look a lot like human imagination, to me. I don't think there is a 'magic' or a ghost in the machine – I think we're just computational networks that evolved to transport and propagate genetic information.

But the part of human intelligence that's experienced is a heavily abstractified version of what the brain's doing – like the ripples on the top of a pond, or the display on a car sat nav – where we have a Self that just makes decisions and experiences colours, sounds and feelings, because that kind of information can be used by the system to analyse and modify future behaviours. But underneath the 'Cartesian Theatre' of consciousness, I think you've got something a lot like GPT. Controlled by no one. Experienced by no one. Just synapses firing into a void and some highly abstractified narrative to go along with it, as our own high level operating system. May be OT.

It's GPT's ability to solve programming problems that convinces me there's an intelligence that isn't simply learnt and replicated in there.
strummer6969
Posts: 978
Joined: Tue Dec 15, 2020 10:59 am

Re: Be careful of AI investing hype

Post by strummer6969 »

Logan Roy wrote: Mon Jun 05, 2023 11:52 am
rob wrote: Mon Jun 05, 2023 11:31 am
Logan Roy wrote: Mon Jun 05, 2023 11:09 am
rob wrote: Mon Jun 05, 2023 9:27 am There is no "AI"... it's a nonsense term for a set of algorithms (that are less deterministic than the past). Those algorithms more easily allow functions that appear like "magic" in some ways but are simply the application of maths and computing power. Part of the issue is the black box nature to algorithms.

I agree it's complete hype right now - hopefully it will quickly go back to what it should be and just improved outcomes from the application of algorithms.
It's not clear that human intelligence is anything different. I work in machine learning, and while human intelligence feels a certain way, there's no magic there either. ML is really just a brute force calculus solver that can be applied in ways that makes things appear intelligent.

What *might* be magic, though – in both cases – is emergent behavior.. Emergence is when systems of far more sophistication or complexity form from much simpler systems. So ant and termite behavior can be very sophisticated across a colony, where lots of simple systems are interacting together. Langton's ant is a good computer equivalent.. I sound like I'm getting ahead of myself with GPT, but having worked with it so much, I'm veering towards thinking there's something resembling real intelligence in there, that probably emerges somewhere up the processing hierarchy. You only get that sense working with it on complex problems.
Yeah, I also live in that world. I just don't see the emergence you do and I guess I see human intelligence as far more than GPT or similar, regardless of mechanically how it might be done (and I don't think anyone knows that).

Complex outcomes sure, perplexing how it got there sure but not the "I" part of "AI" IMO. I mean "magic" in the sense of a bow & arrow killing an animal over distance compared to a someone who only knows about a club or someone who can navigate in the open ocean in the age of no time-keeper, sextant or compass. It's simply a lack of that knowledge to the observer, not intelligence by the creator.
Well the other angle I come at it from is things I studied around college, like Thomas Metzinger and computational neuroscience.. And I think what the brain does boils down to the same thing as a computer neural network. And things like Stable Diffusion look a lot like human imagination, to me. I don't think there is a 'magic' or a ghost in the machine – I think we're just computational networks that evolved to transport and propagate genetic information.

But the part of human intelligence that's experienced is a heavily abstractified version of what the brain's doing – like the ripples on the top of a pond, or the display on a car sat nav – where we have a Self that just makes decisions and experiences colours, sounds and feelings, because that kind of information can be used by the system to analyse and modify future behaviours. But underneath the 'Cartesian Theatre' of consciousness, I think you've got something a lot like GPT. Controlled by no one. Experienced by no one. Just synapses firing into a void and some highly abstractified narrative to go along with it, as our own high level operating system. May be OT.

It's GPT's ability to solve programming problems that convinces me there's an intelligence that isn't simply learnt and replicated in there.
The human brain is the product of millions of years of evolution. Through the machinery of our brains, we analyze data that we obtain through our sensory perceptions. The brain can store up to 2.5 quadrillion bytes of data. I can't even imagine how much data the brain processes in a lifetime. AI is nowhere close to replacing humans en masse. I think that is in the realm of science fiction.
User avatar
rob
Posts: 5247
Joined: Mon Feb 19, 2007 5:49 pm
Location: Here

Re: Be careful of AI investing hype

Post by rob »

Logan Roy wrote: Mon Jun 05, 2023 11:52 amIt's GPT's ability to solve programming problems that convinces me there's an intelligence that isn't simply learnt and replicated in there.
I've played with this a fair bit (under gpt3 at least) and I'm just unimpressed by the quality and design of code it can generate. Sometimes it does well but I suspect that it was just able to find an exact solution on (pick your fav code web site here). It's okay at reviewing simple code and suggesting red flags.

To the brain: I don't know but agree there is no ghost in the machine, just disagree on how far GPT and alike are from bare basic "I". I just don't see it at all.

To the social part: I think the stop and need gov regulation crowd is running too far ahead and look like Henny Penny to me - although it's way past time talking about the ethical usage of these types of things and related data ownership & privacy (rather than the whole terminator thing some are on).

To the investment side: It's impossible to figure out who will implement this well. There is no walled garden here, so there is no company to invest in directly and even the handful of tangential companies, who knows which will do well or badly.

To the graph of adoption: I think that is probably right, although it's impossible to know where on that graph it is today.
| Rob | Its a dangerous business going out your front door. - J.R.R.Tolkien
theorist
Posts: 1329
Joined: Sat Sep 28, 2019 11:39 am

Re: Be careful of AI investing hype

Post by theorist »

I have some experience with this, and my two cents worth:

— Unlike many of the much-heralded current future trends (web3, quantum computing,…), machine learning and its close cousin — AI — have made startling practical progress in recent years. There are very real use cases, they are exponentiating, and it will eventually benefit many workers (and reshift some of the landscape for careers).

— However, the investing in this area is very bubblicious. NVIDIA is now priced at a price to sales ratio that reminds of stories like Cisco around the tech bubble. It may well be a big winner, but it is hard to see how that isn’t already more than priced in.

Some of smaller players being bid up are much less certain to be big players, and are already similarly overpriced!

So despite my view on the actual progress in ML and AI, I am not investing with any special focus on those companies.
adave
Posts: 567
Joined: Thu Sep 13, 2007 7:17 pm
Location: Houston

Re: Be careful of AI investing hype

Post by adave »

I think it is a good point that human intelligence and learning comes from all our senses. Just look at a baby interacting with his or her environment. I don't think a Bot will ever be able to do this. Thus, whatever "intelligence" a Bot is able to achieve, it will be fundamentally different than human intellect... perhaps (hopefully) it will be complimentary.
secondopinion
Posts: 6011
Joined: Wed Dec 02, 2020 12:18 pm

Re: Be careful of AI investing hype

Post by secondopinion »

KlangFool wrote: Mon Jun 05, 2023 11:47 am
K72 wrote: Mon Jun 05, 2023 11:27 am So far I'm not impressed with ChatGPT. Both DW and I have dabbled, though not for investing (her for a book she is writing and me for Python programming). In every case, the result looked impressive on the surface but was utter nonsense. Won't catch me anytime soon trying ChatGPT for investing.
But, it takes intelligence to notice the utter nonsense. Perhaps, we should use ChatGPT as a new kind of "Turing Test". Something to separate folks that can notice utter nonsense versus those that can't.

KlangFool
+1. So far, it has to be used in an Arthur-Merlin type protocol (that is, we cannot trust it; so we might to submit tests to feel confident that the supposed information is valid).

I think too many blindly believe it…
Passive investing: not about making big bucks but making profits. Active investing: not about beating the market but meeting goals. Speculation: not about timing the market but taking profitable risks.
KlangFool
Posts: 31526
Joined: Sat Oct 11, 2008 12:35 pm

Re: Be careful of AI investing hype

Post by KlangFool »

secondopinion wrote: Mon Jun 05, 2023 1:04 pm
KlangFool wrote: Mon Jun 05, 2023 11:47 am
K72 wrote: Mon Jun 05, 2023 11:27 am So far I'm not impressed with ChatGPT. Both DW and I have dabbled, though not for investing (her for a book she is writing and me for Python programming). In every case, the result looked impressive on the surface but was utter nonsense. Won't catch me anytime soon trying ChatGPT for investing.
But, it takes intelligence to notice the utter nonsense. Perhaps, we should use ChatGPT as a new kind of "Turing Test". Something to separate folks that can notice utter nonsense versus those that can't.

KlangFool
+1. So far, it has to be used in an Arthur-Merlin type protocol (that is, we cannot trust it; so we might to submit tests to feel confident that the supposed information is valid).

I think too many blindly believe it…
But, this is normal and to be expected.

Common sense is highly uncommon.

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
cacophony
Posts: 1363
Joined: Tue Oct 16, 2007 9:12 pm

Re: Be careful of AI investing hype

Post by cacophony »

rob wrote: Mon Jun 05, 2023 12:57 pm
Logan Roy wrote: Mon Jun 05, 2023 11:52 amIt's GPT's ability to solve programming problems that convinces me there's an intelligence that isn't simply learnt and replicated in there.
I've played with this a fair bit (under gpt3 at least) and I'm just unimpressed by the quality and design of code it can generate. Sometimes it does well but I suspect that it was just able to find an exact solution on (pick your fav code web site here). It's okay at reviewing simple code and suggesting red flags.

...
If you're curious to try, this site is a free way to play with GPT-4 for development related inquiries : https://www.phind.com/
You need to enable the "Use Best Model" toggle, which is limited to 25 interactions every 4 hours.
User avatar
dogagility
Posts: 3237
Joined: Fri Feb 24, 2017 5:41 am

Re: Be careful of AI investing hype

Post by dogagility »

For the commercial world, AI/ChatGPT is the new buzzword analogous to use of the Sustainable/Sustainability buzzword.
Make sure you check out my list of certifications. The list is short, and there aren't any. - Eric 0. from SMA
User avatar
rob
Posts: 5247
Joined: Mon Feb 19, 2007 5:49 pm
Location: Here

Re: Be careful of AI investing hype

Post by rob »

cacophony wrote: Mon Jun 05, 2023 1:15 pm If you're curious to try, this site is a free way to play with GPT-4 for development related inquiries : https://www.phind.com/
You need to enable the "Use Best Model" toggle, which is limited to 25 interactions every 4 hours.
Thx... I'll give it a go.

As above, I think there are some great use-cases for (LLM & similar) algorithm based solutions (eg. I love the generated content in beta photoshop which is frustratingly close to amazingly useful) and I think they can be used - carefully - as an aid to certain tasks. Just think the rhetoric & expectations need to be dialed WAY down.... and they need to remove the "AI" label from the hype.

These things will not replace the vast number of professional jobs - although some for sure like any tech adoption; The people that can use it in the right spots will be superior to those that cannot adopt it in some way, just like search engines have replace needing to read all the manuals or code based solution websites have partially replaced the search engines (for certain roles).
| Rob | Its a dangerous business going out your front door. - J.R.R.Tolkien
rockstar
Posts: 6326
Joined: Mon Feb 03, 2020 5:51 pm

Re: Be careful of AI investing hype

Post by rockstar »

adave wrote: Mon Jun 05, 2023 8:30 am NVDA is an interesting case.

It’s more like an AI infrastructure company.

Turns out GPUs are ideal for the type of compute required in AI and NVDA has a dominant position in this market.

I see no harm in holding a small portion just to go along with on the ride.
NVDA has produced the cores that ML libraries use for years. This is nothing new. What is new is that models keep getting better and processing power keeps getting faster.
Logan Roy
Posts: 1838
Joined: Sun May 29, 2022 10:15 am

Re: Be careful of AI investing hype

Post by Logan Roy »

rob wrote: Mon Jun 05, 2023 12:57 pm
Logan Roy wrote: Mon Jun 05, 2023 11:52 amIt's GPT's ability to solve programming problems that convinces me there's an intelligence that isn't simply learnt and replicated in there.
I've played with this a fair bit (under gpt3 at least) and I'm just unimpressed by the quality and design of code it can generate. Sometimes it does well but I suspect that it was just able to find an exact solution on (pick your fav code web site here). It's okay at reviewing simple code and suggesting red flags.

To the brain: I don't know but agree there is no ghost in the machine, just disagree on how far GPT and alike are from bare basic "I". I just don't see it at all.

To the social part: I think the stop and need gov regulation crowd is running too far ahead and look like Henny Penny to me - although it's way past time talking about the ethical usage of these types of things and related data ownership & privacy (rather than the whole terminator thing some are on).

To the investment side: It's impossible to figure out who will implement this well. There is no walled garden here, so there is no company to invest in directly and even the handful of tangential companies, who knows which will do well or badly.

To the graph of adoption: I think that is probably right, although it's impossible to know where on that graph it is today.
In many cases, it feels like it's pasting code from Stack Overflow.. But then I'll give it specific problems, where it has to find innovative solutions or work out why something's not working as it should – and with QT, you get a lot of this. And it's found ways of working around known QT issues that the discussion boards don't seem to know about.. And as you get into prompting, and working through solutions with it, I don't think it could generate some of the solutions it does without an ability to conceptualise and apply principles from different areas.

It's nearly impossible to prove, either way. However, compared to discussions on Stack Overflow and QT forums, I'd say ChatGPT feels like it does a better job of understanding the problem than the people who populate those spaces do. I remember one of the first ML things I wrote was a simple reinforcement learning loop you could type into and condition it to learn things, like how to recognise and solve number sequences, using the in and out of an LSTM. And it felt so much like training an animal. There's something very organic about interacting with it directly.
Elysium
Posts: 4120
Joined: Mon Apr 02, 2007 6:22 pm

Re: Be careful of AI investing hype

Post by Elysium »

Is it really necessary to be careful of any hype, forget AI, let's call it excitement about a new technology, product, innovation, anything. Should we curb our excitement and if all of us did would there be any real inventions? The answer is clear, we need excitement from some people who build new things or else we would all still live with wheel barrows and horse driven carts.

That's one side of it, but then understandably there is something to be said about making financial mistakes from something that may be a hype or a bubble. I am just not sure AI investing is a hype at this point, it's an idea, which is real, that needs investment and encouragement. I don't think there is anything wrong with VCs funding these innovations. AI will make it's way into most applications in the next 10 years, it's just a natural extension of everything that came before it. We need to solve the problem of self driving, it's long overdue, we cannot advance with humans driving the cars always. Humans are very inefficient at driving cars, we make a lot of mistakes, if at all anything AI can help us in many ways to improve it, it doesn't need to fully take over. That's just one example, like that AI needs to find it's way into every application which is it's natural next level of growth.

What about investing? is there hype, yes, there will be some. But it's not at any bubble level. I remember looking at building applications using HTML in the mid 90s after HTTP and WWW became available, and it stuck me then this is exactly how we will build applications in future. The dot com bubble was a super hype, but that came much later, and even then the excitement wasn't unwanted, all of that eventually materialized, just took 20 years. The problem was about expectations of materializing everything in the next 2-3 years and having every single company dealing with dot com given a high value.

Similarly, AI wil be realized, but not every company dealing with it should be given a high value, when that happens we would know it's a bubble, and we can protect financially by being conscious about it. That being said, I will leave this post with a question, how do you protect from a bubble in a cap weighted indexing system when the index will take the most hyped up company to the top.
WhiteMaxima
Posts: 3338
Joined: Thu May 19, 2016 5:04 pm

Re: Be careful of AI investing hype

Post by WhiteMaxima »

If you invested in broard market like SPY500, you are not out of game or overloaded with AI.
Logan Roy
Posts: 1838
Joined: Sun May 29, 2022 10:15 am

Re: Be careful of AI investing hype

Post by Logan Roy »

strummer6969 wrote: Mon Jun 05, 2023 12:37 pm
Logan Roy wrote: Mon Jun 05, 2023 11:52 am
rob wrote: Mon Jun 05, 2023 11:31 am
Logan Roy wrote: Mon Jun 05, 2023 11:09 am
rob wrote: Mon Jun 05, 2023 9:27 am There is no "AI"... it's a nonsense term for a set of algorithms (that are less deterministic than the past). Those algorithms more easily allow functions that appear like "magic" in some ways but are simply the application of maths and computing power. Part of the issue is the black box nature to algorithms.

I agree it's complete hype right now - hopefully it will quickly go back to what it should be and just improved outcomes from the application of algorithms.
It's not clear that human intelligence is anything different. I work in machine learning, and while human intelligence feels a certain way, there's no magic there either. ML is really just a brute force calculus solver that can be applied in ways that makes things appear intelligent.

What *might* be magic, though – in both cases – is emergent behavior.. Emergence is when systems of far more sophistication or complexity form from much simpler systems. So ant and termite behavior can be very sophisticated across a colony, where lots of simple systems are interacting together. Langton's ant is a good computer equivalent.. I sound like I'm getting ahead of myself with GPT, but having worked with it so much, I'm veering towards thinking there's something resembling real intelligence in there, that probably emerges somewhere up the processing hierarchy. You only get that sense working with it on complex problems.
Yeah, I also live in that world. I just don't see the emergence you do and I guess I see human intelligence as far more than GPT or similar, regardless of mechanically how it might be done (and I don't think anyone knows that).

Complex outcomes sure, perplexing how it got there sure but not the "I" part of "AI" IMO. I mean "magic" in the sense of a bow & arrow killing an animal over distance compared to a someone who only knows about a club or someone who can navigate in the open ocean in the age of no time-keeper, sextant or compass. It's simply a lack of that knowledge to the observer, not intelligence by the creator.
Well the other angle I come at it from is things I studied around college, like Thomas Metzinger and computational neuroscience.. And I think what the brain does boils down to the same thing as a computer neural network. And things like Stable Diffusion look a lot like human imagination, to me. I don't think there is a 'magic' or a ghost in the machine – I think we're just computational networks that evolved to transport and propagate genetic information.

But the part of human intelligence that's experienced is a heavily abstractified version of what the brain's doing – like the ripples on the top of a pond, or the display on a car sat nav – where we have a Self that just makes decisions and experiences colours, sounds and feelings, because that kind of information can be used by the system to analyse and modify future behaviours. But underneath the 'Cartesian Theatre' of consciousness, I think you've got something a lot like GPT. Controlled by no one. Experienced by no one. Just synapses firing into a void and some highly abstractified narrative to go along with it, as our own high level operating system. May be OT.

It's GPT's ability to solve programming problems that convinces me there's an intelligence that isn't simply learnt and replicated in there.
The human brain is the product of millions of years of evolution. Through the machinery of our brains, we analyze data that we obtain through our sensory perceptions. The brain can store up to 2.5 quadrillion bytes of data. I can't even imagine how much data the brain processes in a lifetime. AI is nowhere close to replacing humans en masse. I think that is in the realm of science fiction.
The brain's also very slow – astonishingly slow compared to a modern computer. The human brain's storage is estimated to be 10-100 terrabytes – each synapse stores a byte, so it's how many synapses? It may be as much as 300 TB. This would only be 3x Nimbus solid state drives that any home computer can use.

There are things computers obviously do a lot better, and likely things an organic computer does a lot better. But the speed of evolution in the computer realm is incomparable.
valleyrock
Posts: 1129
Joined: Sun Aug 12, 2018 7:12 am

Re: Be careful of AI investing hype

Post by valleyrock »

rockstar wrote: Mon Jun 05, 2023 1:33 pm
adave wrote: Mon Jun 05, 2023 8:30 am NVDA is an interesting case.

It’s more like an AI infrastructure company.

Turns out GPUs are ideal for the type of compute required in AI and NVDA has a dominant position in this market.

I see no harm in holding a small portion just to go along with on the ride.
NVDA has produced the cores that ML libraries use for years. This is nothing new. What is new is that models keep getting better and processing power keeps getting faster.
Yes, but what are the models? I remember the days of neural networks. Actually they are still being hyped. Create a bunch of linkages with no fundamental basis, use "training," again undefined and with no fundamental basis, let it rip and publish a bunch of research about it.

Full explanation/revealing of the algorithms/models will really be necessary to understand this LLM, and, to bring this back on-topic, to get any idea as to the financial and investing ramifications of them. Unfortunately, we'll see a lot of applications, lots of research, hand-wringing, and jawboning without people agitating for honest explanations/a full reveal of how this data is being processed in these LLMs. While some people might think it's rocket science, I'll bet that at its core, this stuff is so simple that those behind it have no trouble complicating the hell out of it (to adapt a phrase).
Locked