Hacker News

Top stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Fujitsu launches made-in-Japan next-generation CPU FUJITSU-MONAKA(global.fujitsu ↗)
    107comments
  2. Rate limits on GitLab.com are changing(about.gitlab.com ↗)
    37comments
  3. Artificial intelligence now beats some of the best human forecasters(economist.com ↗)
    22comments
  4. CrowdSec Source Code Leak(crowdsec.net ↗)
    1comments
  5. LLM Classification Is Feature Engineering(minimallysufficient.com ↗)
    8comments
  6. Whoisinspace.com/(whoisinspace.com ↗)
    5comments
  7. Launch HN: Skillsync (YC W26) – AI chat sessions made portable across agents(skillsync.com ↗)
    discuss
  8. One Year of Sponsored Servo Development(servo.org ↗)
    123comments
  9. Vinix – A modern operating system written in V(vinix-os.org ↗)
    7comments
  10. hister(github.com/asciimoo ↗)
    discuss
  11. Show HN: Die With Me – Claude and Codex rate limits as AIM away messages(diewithme.co ↗)
    2comments
  12. CCC invites all model citizens to 40C3(ccc.de ↗)
    64comments
  13. Nvidia announces native GPU programming in Rust(nvidia.com ↗)
    352comments
  14. Show HN: Share your AI Setup, Learn from others(mysetup.ai ↗)
    30comments
  15. My temporary PHP fix from 2014 has nearly 20M installs. Today I'm deprecating it(jakeasmith.com ↗)
    70comments
  16. The Relation Between Mathematics and Physics by Paul Dirac (1939)(cam.ac.uk ↗)
    36comments
  17. Keys Not Included: recovering the signing keys for US driver's license barcodes(ryan.science ↗)
    125comments
  18. Mastering Layout Engines in Graphviz: Dot vs. Neato vs. Twopi vs. Circo(visual-paradigm.com ↗)
    1comments
  19. GLM Built Its Own Inference Infrastructure(z.ai ↗)
    201comments
  20. Ask HN: How to recover Google auth after phone stolen?
    11comments
  21. Show HN: I built a new version of my fun spatial 3D online meeting app(flat.social ↗)
    48comments
  22. Grand MS-DOS Gaming General MIDI Showdown(johnnovak.net ↗)
    discuss
  23. Xiaomi Mimo 2.6 live post-training dashboard(xiaomi.com ↗)
    147comments
  24. Better Vector Search for Long Documents: Chunking Inside Manticore Search(manticoresearch.com ↗)
    10comments
  25. Lucasart's Afterlife(togameforlife.wordpress.com ↗)
    40comments
  26. Online Z3 Guide(microsoft.github.io ↗)
    14comments
  27. Cloudflare/Security-Audit-Skill(github.com/cloudflare ↗)
    33comments
  28. Show HN: An e-ink frame that hears birds and draws them as 1800s illustrations(github.com/arnegiacomo ↗)
    250comments
  29. Comparison of Malloc() Algorithms(egbert.net ↗)
    32comments
  30. Why Germany Is Building an Ark for U.S. Climate Data(yale.edu ↗)
    1comments

Artificial intelligence now beats some of the best human forecasters

37 pointsby 1h agoeconomist.com
22 comments
43m agoHN ↗

That is too bad for The Economist. Exor N.V and Agnelli might replace some pundits at The Economist.

33m agoHN ↗

The Economist has actually published other human forecasts many times, e.g. Metaculus or Good Judgment forecasts. They do year-end forecasts too.

Whether they draw on AI or other humans seems immaterial to the quality of their reporting.

42m agoHN ↗

Product idea: a LLM trained separately from mainline LLMs that anticipate market trends by analyzing how mainline LLMs will invest. As retail investors will probably use mainline AI for decisions going forward , one could get an edge.

"The AI-driven Market Hypothesis"

Please let me know where I should pick up my Nobel prize.

40m agoHN ↗

Then the next person needs an LLM trained to predict the LLM trained to predict the mainline LLM.

It's derivatives all the way down

3m agoHN ↗

"No one could have anticipated the market crash of 2028."

39m agoHN ↗

Would you not then also copy the investments? Or are you trying to inverse the trades by an unpredictable time factor reasoning that thanks to AI the underlying stock is over- or underpriced?

34m agoHN ↗

I know this is tongue-in-cheek, but I think your idea could actually work, but not in financial markets. (The "keynesian beauty contest" of trying to predict what others think been played out to death there.)

You could train a model to anticipating scientific trends. Or policy trends. Others will definitely use mainline LLMs to make decisions there, so they may be more predictable now!

39m agoHN ↗

Given the training data isn’t that more a win for the wisdom of the crowd?

28m agoHN ↗

Aren't forecasters already using 'artificial intelligence' for decades in the form of non-llm machine learning models?

14m agoHN ↗

You don’t even have to limit it to machine learning, the definition of forecasting is isomorphic to the definition of modeling, which, with the dilution of the term AI, is also isomorphic to the definition of AI.

More simply:

  - forecasting = modeling = AI

Edit: I’d even throw statistics into that extended equality, meaning that Bayes, Bernoulli and even the fellow named John Gaunt have a strong case for having invented AI.

6m agoHN ↗

forecasting = modeling = AI

I wouldn't go that far. Humans can forecast by modeling with their wetware, nothing "A" about it.

8m agoHN ↗

I think also a lot of it is intuition.

25m agoHN ↗

The test is when reflexivity kicks in and the prediction itself changes market behavior. LLMs usually melt there

8m agoHN ↗

they fixed it . need better paywall bypasses

14m agoHN ↗

So I can guess the AI companies can stop with their plans to infest AI with ads and they'll instead fully fund themselves by using their AI to gamble on stocks and the prediction market right? Surely the chatbots will just print money!

13m agoHN ↗

At the risk of sounding extremely naieve i have a question for the Wall St / quant / HFT folks lurking here ... but how hard would it actually be to brute force the math/algos behind Medallion Fund (or something in that general class) or even some of the average quant funds

I know it’s not just the math but execution, infrastructure, risk management, data, colocation (if ur an HFT) etc ... but LLMs seem like a pretty powerful apparatus for running experiments that .. a few years ago would have required fairly deep multidisplinary skills across coding .. stats .. and math ..

So assuming you have decent intuition for ideas .. how difficult would it actually be to reverseengineer / rediscover some of the underlying stuff?

5m agoHN ↗

This has to be the least surprising development to date given ml is a universal function estimator