Pinned Tweet
we formalized this further in long horizon research tasks, specifically where there are multiple goodness of fit metrics that need to be considered and optimized for.
one thing we realized is that the loop is most valuable at the accept/reject boundary.
a global aggregate score can improve while the result moves in the wrong local direction, so the external loop has decide if the agent did actually find a better solution, or did it just find a local tradeoff that makes the headline metric look better?
the paper explore's this on a mechanistic ecology model used in NASA’s Carbon Monitoring System.
arxiv.org/abs/2606.11522
1
1
32
15,431
hey dots, can you talk to my muse to check up on my grok bot and ask it to send the repo over to my openclaw so that it can unblock hermes memory upload to my instinct?
3
1
20
1,019
i will accept agi is here when we solve the housing crisis, control inflation, and find a better system than democracy
1
4
205
Dev retweeted
Price discovery for your career / go to market for getting an amazing job.
If you're a swe in sf/nyc there are hundreds of startups that want to hire you. And if you're lit you already know there are people hitting you up but it's across your dms, inbox, linkedin and more.
When you're navigating misc messages, referral requests one by one you're limiting how much liquidity you have. In other words you're trying to price an asset (your time and experience) in a very thin market when there is otherwise a ton of demand.
There's no button on the internet that you can press to get hundreds of startups to start hitting you up when you're ready to explore.
@FonziAI gives you that button. Check it out talent.fonzi.ai/
19
12
295
93,847
this has to be a bubble
Instinct is in talks to raise about $1 billion at a roughly $10 billion valuation, less than a month after raising $250 million at a $2.25 billion pre-money valuation.
The 23-year-old founder Noah Shinn is building a personal AI agent that can operate software and complete tasks like answering emails, negotiating bills and booking reservations.
Instinct now has more than 100,000 users and has already run into compute capacity constraints as demand grows.
Its valuation has gone from roughly $50 million in the spring to $10 billion under discussion today.
Source: The Information
10
432
this entire argument is moot. a common mistake people make is trying to define invention through the lens of human process.
for example, a brute force approach to the solution with no “intuitive jump” would be not be classified as an invention. but that makes no sense. we tend to do this because it is the only concrete manner in which we can define invention: by defining the stages of the process and not outcome due to its open-ended nature.
theoretically llms, or the blanket term “ai” (think of world models, jepa etc.) can get us to invention. and so far they seem to have (navier stokes etc). humans and machines have inherent different strengths from first principles and WILL achieve invention through different means suited to their own makeup.
and all knowledge IS continuous in some capacity. there is no “jump” from a combinatorial sense. it might appear like one BECAUSE of the long and complex relationship between concepts but it is all continuous.
again. think from the lens of discovery not from the lens of how humans conduct discovery. machines will probably not and DO NOT have to emulate humans. they should be operating on the substrate which leverages their strengths. now i’m not saying that llms or “ai” is there yet for all facets of discovery. but scaling up might.
will share a deeper write up formalizing these ideas soon.
Oxford researchers argue that LLMs can never invent anything.
It is mathematically impossible.
They published a paper called “Theory Is All You Need" and it argues against the claim that computational models can generate genuine novelty or new knowledge.
They analyzed the limits of generative ai, and the results are a brutal reality check for the idea that ai will replace human decision making under uncertainty.
Here is why AI is stuck and human cognition wins:
backward-looking vs forward-looking.. llms are probability machines that look backward at existing data. human cognition is forward-looking and capable of generating genuine novelty. human cognition operates theoretically "top-down" rather than "bottom-up" from data.
the "data-belief asymmetry".. the researchers use the invention of "heavier-than-air flight" to illustrate this concept. an ai relies on data-based prediction, which is largely imitative. humans, however, use theory-based causal logic that allows them to hold beliefs that go beyond existing data.
the intervention gap.. humans don't just process information; we use theory to practically "intervene" in the world. we engage in directed experimentation to generate entirely new data. ai-based models are theory-free and place primacy on existing data and prediction.
tldr?
AI uses a probability-based approach to knowledge and ia largely imitative. It can process data and make predictions, but human cognition relies on theory-based causal reasoning.
The decades-old analogy comparing human minds and computers to mere "input-output" devices is fundamentally flawed.
1
161
truly superintelligent ai is OOM worse than nukes.
question is, are we really on that path or is it doomerism or a mix of both?
3
146
