Hacker Newsnew | past | comments | ask | show | jobs | submit | Mr_P's commentslogin

"Recursive self improvement" is not new. Compilers have enabled recursive self improvement since the 1960s. Semi-supervised leaning is not new either.

When people talk about recursive self improvement, they mean autonomous self improvement.

No humans in the loop.


What semi-supervised learning tells us is that any system with no humans in the loop will at some point reach a limit, and require more supervised data.

Deep learning had no humans in the loop, but there's always a bigger outer loop with a human somewhere.


And, the explicit goal of these companies is to break that bigger outer loop. That's the reason for the investment, and the reason for the valuation.

You can place bets that these companies will be failures if you want.


thats cause none of these systems are Intelligent yet?

This is a robotics paper by Nvidia & UW researchers.

It is not an Alphabet/Deepmind paper, and it also has nothing to do with the "NIST" OCR dataset.

Was there really not a more appropriate name than "AlphaNIST"?



Woah, deepmind solved the national institute of standards and technology???


Situation Before: there are 14 competing standards.

Situation Soon: there are 15 trillion standards tailored to every possible situation.


I know you are joking, but that is absolutely a possible trajectory. Reliable and cheap LLMs could lead to software that re-implements everything, instead of using libraries.


Indeed, it can lead to the end of software as we know it. Instead everything may be written on the basis of a single user.


I thought of the National Institute of Standards and Technology... acronym/naming overload, like using "STL"!


That is indeed the reference


NIST makes sense, but co-opting Alpha is a bit odd.

Alpha historically had meant "Google DeepMind" not RL


Well that settles it! They'll just have to hire us. Then the universe will be back in order XD!


Congrats to the team on such a successful exit!

I'm curious how much the recent AI 3D modeling trend (Astra post-trained to use Blender) would have impacted this.

World Labs' entire tech stack becomes potentially obsolete when anyone can prompt GPT/Claude to create a simulation-ready 3D model from a photograph.


My guess is their tech can be used for spatial reasoning training environments, which gives a value beyond just output of a 3D scene.


If you click through to the paper, it has approximately nothing to do with what this HN post title suggests.


Yes, it appears the submitter rewrote the title (strike 1) without even reading TFA (strike 2). Not great.

Actual title: "New MIT Algorithm Meets Every Hard Constraint in Simulated Tests"


Indeed. Which is a shame because it's still pretty cool work.


If fairness this article is poorly written and doesn’t explain what they actually did at all.


Android and Chrome need on-device AI capabilities. Google can't lock down those weights like it can with server-side ML.

So it's easier to just release those models as open source and make it official, since someone would inevitably hack the weights out anyway.


Could say the same for camera processing in the Pixel Camera app or any other binary someone wants to re-use that comes included in a software distribution (seemingly for 'free'). They can't lock the instructions up on the server so they might as well make the binary be freely distributable?

Companies don't commonly give away executable binaries "just because", why'd they start now for these binary blobs that are the models?

Not that I'm unhappy about it! Yay for open data any day, I'm just not understanding why, at least beyond PR in nerd circles


Binaries are source code outputs, they are copyrightable and patentable. Weights are not copyrightable so people can freely extract the weights and run them. If Google patents any of the novel algorithms here releasing it all freely isn't an impediment to making people license it.


Weights are not copyrightable?!

Are you sure that isn't about LLMs' outputs? There I know there have been some court cases that say this, but the model itself is a work created in intricate and somewhat creative ways (I hesitate to use the word "creative" here, but would similarly hesitate to label a routine picture of the moon creative whereas pictures basically always have copyright; the bar for creativity is basically an epsilon amount above zero, afaik)


Because a model like this can't be as easily obfuscated as image processing. Image processing is a bundle of many moving parts, a lot of functions each with it's own inputs and outputs. A model is a single function which can be easily extracted and reused, in comparison


Arguably, but that's not the point. Take image (e.g. png) files on a CD-ROM shipped by a game vendor, which can be trivially copied even by my grandma. That doesn't move the game vendor to release them as freely distributable under the Apache license


Good point but still, why would Google police this model? If they had a restrictive licence on it do you think it would be worth it for them to enforce it? This way they at least buy some good will and mindshare


That makes sense to me. Guess one might say the same for game icons and other such files that lay around in disks, but yeah maybe it's as simple as that


Not quite the same, understandably Blizzard cares a lot about their IP because otherwise private servers leech their users. Maybe a small game designer cares a lot about the small game they made or whatever since that's all they have. A four trillion market cap company can afford to be "charitable".. where it costs them nothing and might cost them more to enforce their rights.


> can't lock down those weights

They could lock them down legally which would prevent commercial use, but they choose not to, and they boast about how many tens of millions of times Gemma models have been downloaded by developers.

So there must be more to the rationale than just local model weights getting hacked out of devices.


But these can't be the same model - the model is far too demanding to be part of regular chrome for most people.


How does this compare to Claude Managed Agents?


Claude managed agents is a general-purpose hosted runtime for Claude. While Twill focuses on SWE tasks.

And so the SWE workflow is pre-built (research, planning, verification, PR, proof of work). Twill is also agnostic to the agent, so you can use codex for instance. Additionally you have more flexibility on sandbox sizing on Twill


I had to double check that this wasn't an April Fools joke. The GitHub project has commits from 2 weeks ago, so it's not.

Looking more closely though, this looks a lot like the Google "AI Cookie" from 2017, which also used Bayesian Optimization: https://blog.google/innovation-and-ai/technology/research/ma...


Google's "Smart Cookie" indeed also used techniques from Bayesian Optimization. For some technical detail, see their write-up here: https://static.googleusercontent.com/media/research.google.c...

Our work on concrete here differs in that the problem is both 1) an inherently time-varying, and 2) multi-objective. See our write-up here for details: https://arxiv.org/pdf/2310.18288


I looked around the repository, and it looks like it's just 3 regexes to strip whitespace or filler words:

https://github.com/ARPAHLS/skillware/blob/main/skills/optimi...


Years ago, my college multi variable calculus and linear algebra courses were both taught primarily using course materials that were interactive Mathematica Notebooks.

We had access to all of the symbolic algebra tools and were even expected to use them regularly for both courses. It was great!

I'm not sure how well this would extend to introductory courses though, especially if the standardized tests still expect integration by hand.


I can't imagine doing proofs like Cauchy's integral theorem without touching a paper.



Oct 2023 is for Space Force, $70m. OP is for NRO, $1.8b.

Yes published budget buckets tend to hide spending reality, but reported as a new contract with a new customer.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: