frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Extending single-minus amplitudes to gravitons

https://openai.com/index/extending-single-minus-amplitudes-to-gravitons/
10•telotortium•2h ago

Comments

behnamoh•1h ago
I'm tired of these posts; LLMs are good for happy-path demos, that's it. And even then, their success rate depends on the prompter already knowing the answer!

Literally any out-of-distribution project in which I used LLMs lead to catastrophic failure. The models can't "see" stuff outside their training data.

semiquaver•56m ago
I legitimately can’t tell if you’re being serious. It kind of seems like you might be trying to parody LLM detractors that will never admit to their usefulness. If you’re serious, why choose to say so in this post, which includes hard evidence that you’re wrong?
behnamoh•51m ago
> which includes hard evidence that you’re wrong?

You should already know what to ask to extract the answer OpenAI claims gpt-5.2-pro gave them.

Then you should be lucky to get an answer that makes sense.

Then you should already know how to verify the model's response.

Only after all these steps should you cherry-pick the one-in-a-million successful response to feature on your website.

And finally, you should prove that the answer didn't already exist in the training data. It's highly likely that the problem was solved before and the model picked that up. I have yet to see a genuinely novel discovery these models can produce.

* I'm an LLM researcher, but that doesn't mean I should close my eyes to the unjustified hype around language models.

MajimasEyepatch•39m ago
According to the post, this result was first derived for gluons in a previous paper. That paper was provided to the model as context, and then the model was asked to derive an analogous result for gravitons, which presumably has not been done before. The authors claim it would have taken "considerable time" for human experts to derive the graviton result.

I don't see any reason to believe that this exact problem was solved before in the training data, but it's definitely an incremental result based on a very similar problem that the model had seen before.

JProthero•18m ago
>It's highly likely that the problem was solved before and the model picked that up.

If you can demonstrate that, I would put it to Strominger and his colleagues, and I imagine they would be obligated to cite your contribution in the peer-reviewed publication.

behnamoh•15m ago
> If you can demonstrate that, I would put it to Strominger and his colleagues, and I imagine they would be obligated to cite your contribution in the peer-reviewed publication.

There's one little problem: OpenAI isn't actually open and doesn't reveal which dataset they used for training.

JProthero•1m ago
This shouldn't prevent anyone from finding and reporting a similar pre-existing result in the literature.
JProthero•27m ago
Do you understand the purported result, and the verification? I don't, but I'm confident that Andrew Strominger wouldn't have agreed to put his name on this if he didn't think it was correct and interesting.

The human authors have positions at the Institute for Advanced Study (Einstein's old institution), Vanderbilt, Harvard (Strominger) and Cambridge in the UK.

If you have to gauge this by the reputation of the experts involved in it as I do, that seems like a good list to me.

MacBook Neo

https://www.apple.com/newsroom/2026/03/say-hello-to-macbook-neo/
1565•dm•11h ago•1883 comments

Googleworkspace/CLI

https://github.com/googleworkspace/cli
58•gonzalovargas•1h ago•20 comments

Building a new Flash

https://bill.newgrounds.com/news/post/1607118
320•TechPlasma•5h ago•82 comments

Dario Amodei calls OpenAI’s messaging around military deal ‘straight up lies’

https://techcrunch.com/2026/03/04/anthropic-ceo-dario-amodei-calls-openais-messaging-around-milit...
200•SilverElfin•2h ago•78 comments

Something is afoot in the land of Qwen

https://simonwillison.net/2026/Mar/4/qwen/
525•simonw•10h ago•249 comments

Picking Up a Zillion Pieces of Litter

https://www.sixstepstobetterhealth.com/litter.html
33•colinbartlett•3d ago•18 comments

Humans 40k yrs ago developed a system of conventional signs

https://www.pnas.org/doi/10.1073/pnas.2520385123
64•bikenaga•9h ago•29 comments

Chaos and Dystopian news for the dead internet survivors

https://www.fubardaily.com
3•anonnona8878•19m ago•0 comments

Moss is a pixel canvas where every brush is a tiny program

https://www.moss.town/
183•smusamashah•15h ago•22 comments

What's Driving Rising Business Costs?

https://libertystreeteconomics.newyorkfed.org/2026/03/whats-driving-rising-business-costs/
8•jnord•1h ago•2 comments

NanoGPT Slowrun: Language Modeling with Limited Data, Infinite Compute

https://qlabs.sh/slowrun
124•sdpmas•8h ago•25 comments

The View from RSS

https://www.carolinecrampton.com/the-view-from-rss/
75•Curiositry•5h ago•19 comments

BMW Group to deploy humanoid robots in production in Germany for the first time

https://www.press.bmwgroup.com/global/article/detail/T0455864EN/bmw-group-to-deploy-humanoid-robo...
85•JeanKage•4h ago•68 comments

NRC Issues First Commercial Reactor Construction Approval in 10 Years [pdf]

https://www.nrc.gov/sites/default/files/cdn/doc-collection-news/2026/26-028.pdf
45•Anon84•4h ago•20 comments

“It turns out” (2010)

https://jsomers.net/blog/it-turns-out
243•Munksgaard•11h ago•76 comments

An interactive map of Flock Cams

https://deflock.org/map#map=5/37.125286/-96.284180
508•anjel•7h ago•191 comments

Qwen3.5 Fine-Tuning Guide – Unsloth Documentation

https://unsloth.ai/docs/models/qwen3.5/fine-tune
280•bilsbie•14h ago•70 comments

Daemon (2006)

https://en.wikipedia.org/wiki/Daemon_(novel)
18•solomonb•9h ago•3 comments

Data Has Weight but Only on SSDs

https://cubiclenate.com/2026/03/04/data-has-weight-but-only-on-ssds-blathering/
71•LorenDB•7h ago•48 comments

Was Windows 1.0's lack of overlapping windows a legal or a technical matter?

https://retrocomputing.stackexchange.com/questions/32511/was-windows-1-0s-lack-of-overlapping-win...
50•SeenNotHeard•5h ago•33 comments

A bit of fluid mechanics from scratch not from scratch

https://tsvibt.blogspot.com/2026/02/a-bit-of-fluid-mechanics-from-scratch.html
31•surprisetalk•2d ago•11 comments

Raspberry Pi Pico as AM Radio Transmitter

https://www.pesfandiar.com/blog/2026/02/28/pico-am-radio-transmitter
73•pesfandiar•4d ago•28 comments

Roboflow (YC S20) Is Hiring a Security Engineer for AI Infra

https://roboflow.com/careers
1•yeldarb•8h ago

Glaze by Raycast

https://www.glazeapp.com/
194•romac•12h ago•119 comments

Making Firefox's right-click not suck with about:config

https://joshua.hu/firefox-making-right-click-not-suck
270•mmsc•7h ago•178 comments

Libre Solar – Open Hardware for Renewable Energy

https://libre.solar
213•evolve2k•3d ago•62 comments

Flip Distance of Convex Triangulations and Tree Rotation Is NP-Complete

https://arxiv.org/abs/2602.22874
17•nill0•4d ago•0 comments

Show HN: Vertex.js – A 1kloc SPA Framework

https://lukeb42.github.io/vertex-manual.html
37•LukeB42•3d ago•23 comments

Show HN: A shell-native cd-compatible directory jumper using power-law frecency

https://github.com/jghub/sd-switchdir
15•jghub•15h ago•1 comments

MyFirst Kids Watch Hacked. Access to Camera and Microphone

https://www.kth.se/en/om/nyheter/centrala-nyheter/kth-studenten-hackade-klocka-for-barn-1.1461249
113•jidoka•13h ago•31 comments