Expert attention map

The Who's Who of AI

What credible people across AI noticed, why it matters, and where the field is converging or disagreeing.

2,397 searchable experts 4,280 tracked across all sources
Build my stream Create my edition
Filter the conversation Who is saying what?

Combine a professional role with a reaction lens. Both must match the same attributed contribution.

Clear all
Active evidence filter

Showing developments with attributable Research & technical analysis reactions.

New Network reaction maps

See how experts are reacting—not just what they shared.

Posts are grouped by conversation and tone. Select a lens to filter the stream; these are never permanent labels on people.

Research & technical analysis
1 live reaction map in this view Jump to the evidence ↓
Reaction converging Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights Research & technical analysis · 7 attributable reactions
Showing signals surfaced by Singularity's Bounty e/cc ×
Debated

The questions experts are actively pulling apart

One card per development. Sources are clustered; reaction bundles describe these posts, never the people behind them.

Established AI research Research 11d ago
LLMs read write neural network weights

Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights

16 experts across 6 network communities independently surfaced this.

Why this matches Research & technical analysis reaction 7 attributable expert contributions · Mark Riedl, Ethan Mollick, Ramon Astudillo
“This is fascinating. Very large models can read the weights of small models and intuit what those models were trained to do, and make direct weight changes to improve them. rikiyat.github.io/blog/weight-...” evidence ↗
16 experts 6 communities 1 sources clustered
Mark Riedl Ted Underwood Ethan Mollick Ramon Astudillo Tim Kellogg Gus Alejandra Caraballo David Marx +8
How the network is reacting 7 experts are independently emphasizing research & technical analysis.
Reaction converging

Research & technical analysis

7 experts

Evidence, methods and technical implications.

“This is fascinating. Very large models can read the weights of small models and intuit what those models were trained to do, and make direct weight changes to improve them. rikiyat.github.io/blog/weight-...”

4 experts discussed this · 5 posts
Ethan Mollick: If this holds up, it is big: "We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x …
Antonino Greco: If this holds up, it is big: "We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x …
Thomas Dietterich: Can someone provide a plausible mechanism for this? Has GPT-6 Astra been trained on neural network weight (or activation) interpretation?
Open the full discussion →
Primary source rikiyat.github.io ↗
More conversations

What experts are discussing without an anchoring article

11 experts · 26 posts · 2d ago
Ethan Mollick Brad Wyble David Manheim Nathan Schneider jake tweety fish Mark Crowley Singularity's Bounty e/cc
Matches Research & technical analysis
“Doing our best over at @inference.coop!” evidence ↗
Multiple readings Building & implementation · 2 Opportunity & adoption · 2 Concern & critique · 1
Ethan Mollick: People who have no contact with coders seem to have no idea how much an entire profession transformed in less than a year due to AI. It is so weird to still see comments here that AI hasn’t changed…
Ethan Mollick: Even I, who has been talking about AI for a long time, has been surprised at how fast things have altered fundamentally in the real world for a major profession. Adoption and transformation usually…
Open the thread →
6 experts · 15 posts · 2d ago
Mark Riedl Thomas Dietterich David Picard Luis Villa Matthew Guzdial Singularity's Bounty e/cc
Matches Research & technical analysis
“I think that overstates his influence. Just my opinion. I’m very online and his public advocacy for open-weights doesn’t ping my radar very often. Anyway, I don’t think NVIDIA gives up on open-weig…” evidence ↗
Multiple readings Concern & critique · 1 Markets & investment · 1 Questions & unknowns · 1
Mark Riedl: Why is this a big blow to AI boom? NVIDIA will continue to make chip and sell them. People will continue to use those chips. Maybe fewer go to Chinese black markets? Maybe it increases availability…
David Picard: Why is this a big blow to AI boom? NVIDIA will continue to make chip and sell them. People will continue to use those chips. Maybe fewer go to Chinese black markets? Maybe it increases availability…
Open the thread →
6 experts · 16 posts · 15d ago
Luke Stark Lilian Edwards Adam L Mary Branscombe Matt Darling Singularity's Bounty e/cc
Matches Research & technical analysis
“And the fun thing is that even with diplomacy and taking the Prime Directive SERIOUSLY Lower Decks still had a lot of fun finding drama and suspense” evidence ↗
Multiple readings Concern & critique · 2 Research & technical analysis · 1 Questions & unknowns · 1
Matt Darling: Just like @liberalcurrents.com tbh.
Luke Stark: the thing i come back to, over and over--which stuff like LOWER DECKS got, and so much other latter-day trek has forgotten--is that the whole idea of trek is that they exist in a genuinely horrifyi…
Open the thread →
3 experts · 18 posts · 7d ago
Ted Underwood Tim Carmody Singularity's Bounty e/cc
Matches Research & technical analysis
“What makes me basically positive about AI is this: we get to design new 𝘸𝘢𝘺𝘴 𝘰𝘧 𝘵𝘩𝘪𝘯𝘬𝘪𝘯𝘨. You can sort of do that with pencil and paper, but it's like writing in sand: readers regress to the mean. …” evidence ↗
Multiple readings Research & technical analysis · 1 Opportunity & adoption · 1
Ted Underwood: What makes me basically positive about AI is this: we get to design new 𝘸𝘢𝘺𝘴 𝘰𝘧 𝘵𝘩𝘪𝘯𝘬𝘪𝘯𝘨. You can sort of do that with pencil and paper, but it's like writing in sand: readers regress to the mean. …
Ted Underwood: "Results are not guaranteed to be positive. Slop, cognitive offloading, dizziness, and nausea may occur. These are not all of the side effects. Don't take Leverofchange if you are allergic to Lever…
Open the thread →
4 experts · 8 posts · 15d ago
Ted Underwood Eli Sennesh tweety fish Singularity's Bounty e/cc
Matches Research & technical analysis
“whatever secret training software engineers are receiving that allows them to evaluate—and change—previous opinions, it's really something we should consider trying on academics” evidence ↗
Multiple readings Markets & investment · 1 Research & technical analysis · 1
Ted Underwood: whatever secret training software engineers are receiving that allows them to evaluate—and change—previous opinions, it's really something we should consider trying on academics
Ted Underwood: it's just very notable that a lot of explicitly AI-skeptical engineers performed a kind of transformation over the last twelve months where they almost "changed," so to speak, "their mind" — though…
Open the thread →
2 experts · 11 posts · 18d ago
Pwnallthethings Singularity's Bounty e/cc
Matches Research & technical analysis
“It's technically incoherent (~ all AIs now use similar model architecture) but I think folks use it because they don't have good language to describe that their real objection is about *use categor…” evidence ↗
Pwnallthethings: It's technically incoherent (~ all AIs now use similar model architecture) but I think folks use it because they don't have good language to describe that their real objection is about *use categor…
Pwnallthethings: Or to put it another way, "Generative AI = bad, ML = good" is just incoherent, but what folks mean by it is that the good categories of AI are mechanical replacement, and the bad categories of AI a…
Open the thread →
2 experts · 6 posts · 19d ago
Singularity's Bounty e/cc apenwarr.ca
Matches Research & technical analysis
“Yup! But of course that's the best kind of scientific breakthrough, because suddenly a whole bunch of new research directions are open. Lots of fun to be had for all. A little retraining won't be s…” evidence ↗
apenwarr.ca: NOT BREAKING: AI still really bad at project estimation
apenwarr.ca: It's slightly wrong wording but proving or disproving a hypothesis is more or less all you can do, so I'll give it a pass. "Prove the XYZ hypothesis" would be unfair instructions because a disproof…
Open the thread →