Conversation

Notices

1. Embed this notice
   Paul Cantrell (inthehands@hachyderm.io)'s status on Saturday, 13-Jul-2024 13:58:04 JST Paul Cantrell

   Huh. Goldman Sachs makes two surprising claim: one obvious but surprising to hear coming from •them•, and one just surprising.

   The obvious-but-wow-they-said-it part: current AI is expensive, unreliable, intrinsically limited:

   “To justify costs, [AI] must be able to solve complex problems, which it isn’t designed to do…not a matter of just some tweaks…the tech is nowhere near where it needs to be in order to be useful for even basic tasks”

   1/

   https://www.404media.co/goldman-sachs-ai-is-overhyped-wildly-expensive-and-unreliable/

   • Haelwenn /элвэн/ :triskell: likes this.
   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Saturday, 13-Jul-2024 13:59:38 JST Paul Cantrell

     in reply to

     This part gave me a grim chuckle (this is 404 Media, inner quote is from Goldman):

     “The paper ultimately questions whether generative AI will ever become the transformative technology that SV and large portions of the stock market are currently betting on, but says investors may continue to get rich anyway. ‘Despite these concerns and constraints, we still see room for the AI theme to run, either because AI starts to deliver on its promise, or because bubbles take a long time to burst.‘”

     2/

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Saturday, 13-Jul-2024 14:00:33 JST Paul Cantrell

     in reply to

     The just-surprising part: Goldman Sachs argues that AI so is hyped right now that optimism about it casts a shadow over •the entire stock market• for the next decade:

     Even if AI does manage to boost productivity, “we find that stocks often anticipate higher productivity growth before it materializes…Using our new long-term return forecasting framework, we find that a very favorable AI scenario may be required for the S&P 500 to deliver above-average returns in the coming decade”

     3/

   • Embed this notice
     Stu (tehstu@hachyderm.io)'s status on Saturday, 13-Jul-2024 14:01:57 JST Stu

     in reply to

     @inthehands that inner quote is quite depressing.

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Saturday, 13-Jul-2024 14:02:01 JST Paul Cantrell

     in reply to

     IOW, they think the entire stock market is already priced as if the alleged AI miracle is guaranteed to materialize. And if it doesn’t, well…then everything, everything is overpriced. “Outside of the most bullish AI scenario…we forecast that S&P 500 returns would be below their post-1950 average.“

     Even I, an AI cynic, am skeptical that AI hype has managed to overprice the •entire stock market• that severely. But then again, maybe I’m overestimating investors.

     4/

     Haelwenn /элвэн/ :triskell: likes this.
     Matthew Lyon repeated this.
   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Saturday, 13-Jul-2024 14:05:15 JST Paul Cantrell

     in reply to

     I suspect many people will say (1) “fuck Goldman Sachs” and (2) “who cares about the stock market, it’s not people.” To that I say (1) sure and (2) like it or not, this is a quantitative measure of forces that are shaping all of our human realities.

     Gullible people are building castles that are maybe (very probably, imo) castles made of sand — and they’re putting us inside them, like it or not.

     When business punches itself in the face, it’s not the people in power who bleed.

     5/

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Saturday, 13-Jul-2024 14:10:06 JST Paul Cantrell

     in reply to

     • Michael Dekker

     And what about Goldman Sachs themselves? If they think AI might be overhyped, are they putting their money where their mouth is? @dekk provides the answer with this old gem: https://hachyderm.io/@dekk/112777129969989267

     I suppose if “we’re in a bubble” were more actionable information, then bubbles would be less likely. Maybe one day, we’ll figure that out.

     In the meantime, I can only long for a less investor-driven world.

     /end

     Haelwenn /элвэн/ :triskell: likes this.
   • Embed this notice
     Osma A (osma@mas.to)'s status on Saturday, 13-Jul-2024 16:04:46 JST Osma A

     in reply to

     @inthehands
     1. For the S&P 500 to be overpriced not *everything* indexed to it has to be, just the overweight, over-AI'd tech stocks. A strong enough correction (one is due) will average the entire index down. Investment strategy: underweigh tech portfolio.

     2. Post-1950 annual returns may be unattainable because of a second phenomenon Goldman Sachs completely ignores: climate crisis.

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Saturday, 13-Jul-2024 16:04:46 JST Paul Cantrell

     in reply to

     • Osma A

     @osma
     Yeah. I •really• don’t think markets have properly priced in the climate crisis yet.

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Saturday, 13-Jul-2024 23:46:42 JST Paul Cantrell

     in reply to

     • Leonard Ritter

     @lritter
     No, no, there’s no changing story here: AGI has consistently been 10-15 years away since the 1950s.

   • Embed this notice
     Leonard Ritter (lritter@mastodon.gamedev.place)'s status on Saturday, 13-Jul-2024 23:46:43 JST Leonard Ritter

     in reply to

     @inthehands a few months ago, AGI was supposed to be around the corner, now we get this headline https://qz.com/openai-five-level-system-human-intelligence-ai-1851588122 - they are managing expectations. soon it's "achieving stage 2 will take longer than assumed". it's bear time.

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Monday, 15-Jul-2024 00:00:27 JST Paul Cantrell

     in reply to

     • Falcon Darkstar
     • Michael Dekker

     @falcon @dekk
     Wish I had the tools and the time to figure something like that out. I’d love to be able to just take one of the index funds and then filter out specific companies.

   • Embed this notice
     Falcon Darkstar (falcon@mastodon.falconk.rocks)'s status on Monday, 15-Jul-2024 00:00:28 JST Falcon Darkstar

     in reply to

     • Michael Dekker

     @inthehands @dekk I have built a portfolio that specifically excludes AI hype stocks for this exact reason, even to the point of avoiding index funds.

     Though to be fair, I put my IRA in over a hundred individual stocks instead of index funds some time ago initially because the index funds would put me in defense contractors, Tesla, surveillance crap, tobacco, and oil, and zero commission trading makes this feasible.

   • Embed this notice
     Paul Cantrell (inthehands@hachyderm.io)'s status on Wednesday, 17-Jul-2024 02:07:42 JST Paul Cantrell

     in reply to

     • Bill Hooker
     • mybarkingdogs

     @sennoma @mybarkingdogs
     Yup, ML has in recent years become very, very good at certain kinds of pattern recognition, and if excitement were focused on those applications, we’d be in a much better place now. But “sometimes-useful ‘find the thing’ tool” can’t generate the sort of froth-mouthed investor hype that “oh noes maybe its skynet” can, so…here we are.

   • Embed this notice
     Bill Hooker (sennoma@chaos.social)'s status on Wednesday, 17-Jul-2024 02:07:43 JST Bill Hooker

     in reply to

     • mybarkingdogs

     @mybarkingdogs @inthehands I'm training a machine learning model to Spot The Thing at work, and it's remarkably good. Like, "not done training and already better than most humans" good. If we can get it hooked into our workflow properly, a very conservative estimate is $1M/yr in savings. More importantly (to me; my employer might disagree), it will free up time for scientists to think instead of doing robot work.

     So yeah, fuck "AI" but with SOME caveats. 🙂

   • Embed this notice
     mybarkingdogs (mybarkingdogs@freeradical.zone)'s status on Wednesday, 17-Jul-2024 02:07:44 JST mybarkingdogs

     in reply to

     @inthehands Yeah, I'm in the "fuck goldman sachs but stopped clock" view here.

     And I'm not even wholly an AI cynic: I think that some LLMs, if care is taken with issues known to affect training data and power, *can* be fun to play with, can be artistic tools for making surrealism/remix art and music *if* their output is correctly labeled and obvious influences are credited, *and* some could handle rote tasks/busywork/"spot The Thing" tasks

     But "chatbots everywhere" is just bullshit hype