kjhughes

Related: Dwarkesh Patel recently interviewed Ada Palmer:

Why Leonardo was a saboteur, Gutenberg went broke, and Florence was weird – Ada Palmer

https://www.youtube.com/watch?v=PAIhVfGbREA

There seems to be a debugging pattern that arises when problems are in proximity to a poorly understood, highly complex part of the system: We tend to think that's where the problem probably is. And we can lose an inordinate amount of time looking for it there. It's like an inverse streetlight effect.

I've seen coding LLMs do it too. I have a well-tested, but complex, subsystem that constantly draws their attention when something non-obvious elsewhere goes wrong.

Dupe: https://news.ycombinator.com/item?id=46150978

Abstract

  We offer a comment on the Centaur (Binz et al., 2025) transformer-based model of human behavior. In particular, Centaur was cast as a path towards unified theories of cognition. We offer a counter claim with supporting argument: Centaur is a path divergent from unified theories of cognition, one that moves towards a unified model of behavior sans cognition.

Here's NVIDIA's blog post on this:

NVIDIA Jetson Thor Unlocks Real-Time Reasoning for General Robotics and Physical AI

https://blogs.nvidia.com/blog/jetson-thor-physical-ai-edge/

I thought they meant that it's a noop as a tool in the sense that it takes no external action. It seems nonetheless effective as a means of organizing reasoning and expressing status along the way.

If you have a valid criticism, express it. Otherwise, withdraw your unsubstantiated slight.

That's the full list. Here's some commentary, including how Anduril took the top spot away from OpenAI this year:

https://www.cnbc.com/2025/06/10/anduril-tops-cnbc-disruptor-...

Does all thinking end up getting replaced by calls to Ask-human-mcp then? Or only thinking that exceeds some limit (and how do you express that limit)?

I'll try to give you credit for more than dismissing my question off-hand...

Yes, it may not need to know with perfect certainty when it's unsure or stuck, but even to meet a lower bar of usefulness, it'll need at least an approximate means of determining that its knowledge is inadequate. To purport to help with the hallucination problem requires no less.

To make the issue a bit more clear, here are some candidate components to a stuck() predicate:

- possibilities considered

- time taken

- tokens consumed/generated (vs expected? vs static limit? vs dynamic limit?)

If the unsure/stuck determination is defined via more qualitative prompting, what's the prompt? How well has it worked?

Cool conceptually, but how exactly does the agent know when it's unsure or stuck?

It got changed (for the worse, in my opinion) away from the original title.

The original title is supposed to be favored here unless it has a serious problem.

This original title had no serious problem, unless accurately summarizing a PhD candidate's thoughtful critique of some questionable AI contributions to scientific research is a serious problem.

G&A = General and Administrative (expenses)

Your framework name suggests that you have an effective method of taking an existing agent that is "close" to meeting requirements by some similarity metric and evolving a new agent that will be better suited than the base agent is to meet the requirements.

If this is true, your post, your repo README file, and your BeeAI Community call presentation (which starts here: https://www.youtube.com/watch?v=5-xqQBv-ccY&t=1294s) ought to be proclaiming such a notable success. Yet, I've seen little to nothing about it in any of those places. Am I missing it?

How exactly does your agent evolution process work?

Suggestions:

- Fix your certs to not lose traffic when Chrome (at least at the default security level) puts up a scary "Your connection is not private" warning for net::ERR_CERT_AUTHORITY_INVALID and requires an override to proceed to your site.

- Fix whatever reputation damage has caused you to be classified as a phishing site: https://phish.opendns.com/main?url=www.calcverse.live

That CNN article is a light-weight review of select food defects allowed, not measured, by the FDA and chosen by the author for their yuck factor. I wouldn't let any of it justify a dismissiveness of heavy metals in dark chocolate.

The Consumer Reports article presents investigative results that measured actual heavy metals levels in dark chocolate. It's worth considering by anyone encouraged to increase their dark chocolate consumption by the diabetes study.

On the other hand, Lead and Cadmium Could Be in Your Dark Chocolate:

https://www.consumerreports.org/health/food-safety/lead-and-...

American art by the late 20th century depicted cows more realistically: https://shorturl.at/aqvDa

Science as a strong-link problem references this piece,

https://www.experimental-history.com/p/science-is-a-strong-l...

discussed here last year:

https://news.ycombinator.com/item?id=35712694