itchyjunk

Did you watch the video? Do you have opinions on it? On MCP? On A2A? What is a title of youtube video and a link to it supposed to add here? Just start a new submission if you just want to link to it.

Can you provide sample questions that are "LLM proof" ?

Since diamonds are surrounded by danger and if it dies, it loses its items and such, why would it not be satisfied after discovering iron pick axe or somesuch? Is it in a mode where it doesn't lose its item when it dies? Does it die a lot? Does it ever try digging vertically down? Does it ever discover other items/tools you didn't expect it to? Open world with sparse reward seems like such a hard problem. Also, once it gets the item, does it stop getting reward for it? I assume so. Surprised that it can work with this level of sparse rewards.

Since you had already figured out the gist, I was hoping you'd have shared the demo link, so I don't have them ad impressions! But I notice a YouTube link below so I'm going there instead. :)

Came to see this and was checking if someone else mentioned it.

"Models like [..], GPT, Cursor"?

That use of emojis on headings very distinctly reminds me of AI writing.

Superficially lists issue but doesn't feel like the author has explored it?

Were you also trying to edit the same code base as the GP or did you evaluate it on some other criteria where it also failed?

I guess experience is widely varying with tools. I had no idea what WSS was. Cursor searched the web and docs and necessary and built a working program. Then we went over each conceptual blocks and it explained stuff to me. When I was confused about some parts, it decided I need to strip away the RealTime API component and just understand asynco in context of WSS so it created a new file with a toy example.

Could it have written some really, really bad and unmaintainable code and it was just justifying things? Very possible. I asked to strip some 60% of the code after I thought I understood, and it was still working. The original code was too abstract for me and hard to follow. Did I learn in this whole process? I think yes. Would I have learned more if I had worked without AI? I think yes.

In life, some people play games and figure out the best builds after a lot of sweat and tears while others prefer googling the meta and only doing those builds, it seems. Both approach seems fine.

How is that diffusion? What noise is it denoising there?

What is your measure of intelligence?

The level of sophistication for CoT model varies. "good old CoT prompting" is you hoping the model generates some reasoning tokens prior to the final answer. When it did, the answers tended to be better for certain class of problems. But you had no control over what type of reasoning tokes it was generating. There were hypothesis that just having a <pause> tokens in between generated better answers as it allowed n+1 steps to generate an answer over n. I would consider Meta's "continuous chain of thought" to be on the other end of "good old CoT prompting" where they are passing back the next tokens from the latent space back to the model getting a "BHF" like effect. Who knows what's happening with O3 and Anthropics O3 like models.. The problems you mentioned is very broad and not limited to prompting. Reasoning models tend to outperform older models on math problems. So I'd assume it does reduce hallucination on certain class of problems.

The axiom is used to give an alternative construction of real. Everything starts with some axioms somewhere.

Petition to add AI to HN so it lists out the gist of papers and articles posted xD . I also always thought AI would overlap with Q Annealing in some way vaguely because both are optimization process. But maybe this is where QC helps AI and not the other way around.

Wait, I thought GPT's were autoregressive and encoder only like BERT used masked tokens? You're saying BERT is auto-regressive or am I misunderstanding?

"Diffusion Models Are Real-Time Game Engines" was a paper from August 2024. How am I playing a real time AI generated game in my browser in 2024? I did try uploading a random image of a random flow chart I had, and it ended up being mostly noise... but still, how?

Worked for me on Edge if that helps :)

Hmm, if I use chicken broth instead of rice, isn't the end product just chicken soup instead of a rice dish? :D

Is this you guessing the system prompt? Is this the actual prompt the model has? What is the context?

You seem pretty confident about there being "no values here". Just because his action also lends to strategy, does not mean there are no values there. You seem to be doubling down on the sentiment by copy/pasting same comment around. You might be right. But I don't know Zuck at a personal level enough to make such strong claims, at least.

Implication being if life exists in oceans -> surface life must exist in ~4B years? That does seem to be true for Earth I guess.

So is 4o. Problem isn't the absence of model, it's inconsistency.

A single tool must work for ever use case or should the tool be picked based on the use case? Tool X doesn't work for use case Y. Regardless of where you shift the blame, tool X doesn't work for use case Y holds.

Maybe that's what the new vs experienced difference is indirectly capturing.

If true, that's really great that it's contributing to math! I'd heard a lot of negative things about String theory in that it was a waste of time for everyone involved and resources used as well. Contribution to theoretical math is well worth it, in my opinion (I confess I might be biased).

Can you elaborate on why you think it won't? That might be more valuable use of everyone's time instead of the binary X will / won't happen stance comment.

How many humans have you tested this with?

[2nd hand only] Ran some experiments on RAG based system. You can take a pool of candidate, semantic search by keywords you'd like each one to have. Once you have a narrow it down, you can have a more refined system that ranks them based on each criteria met.

Is the idea that the companies won't quickly adapt to this? Or worse, start negatively punishing candidates with anomalies like these? First I heard about this "White on white" trick was 7 or 8 months ago. If you RAG over resumes with fixed criteria's to check, it should bypass this already. I wonder if it's the novelty of "AI" that pushed this to the front page.

Do you have any citation on lack of funding for foundational research? From the looks of things, it seems like there is insane amount of money in AI and while maybe majority of it goes to "projects with high probability of success", whatever that means, fringe ideas seem to generate good amount of capital as well. Even (what I would consider to be) academic projects seems to get good funding [0]. On top of that, it seems like you're not saying that Transformers might be a dead end but Neural Networks itself might be one. At this point, NNs already gives a lot of value. If someone's goal is to just get products out that say diffusion models already helps you achieve, how is that even a dead end? Do you mean dead end for AGI? In which case, there seems to be research on alternatives there as well? [1]

[0] https://www.youtube.com/watch?v=rie-9AEhYdY

[1] https://ai.meta.com/blog/yann-lecun-ai-model-i-jepa

I have been bingeing on ttng for a year now. I thought of the song when I saw the title and didn't realize maybe it was inspired from the essay. Saw the mention of TTNG and it got me excited enough to log in and comment after a long time, hurrah!

Good memory. I wonder how often things like this happen. hmm