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
0 live reaction maps in this view

No development in this filter has enough independent reactions yet. The maps appear as soon as the evidence supports them.

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.

No developments match these filters yet.

Try the field-wide view or clear one of the filters.

Clear filters
More conversations

What experts are discussing without an anchoring article

7 experts · 29 posts · 1d ago
Mark Riedl Jeffrey P. Bigham Tim Kellogg Michael Ekstrand David Picard Lionel Singularity's Bounty e/cc
Matches Research & technical analysis
“I tell all the students in my Intro to AI class that they should take courses in computer architecture and operating systems. Coxon is a mathematician from Iniversity of Cambridge. But still.” evidence ↗
Multiple readings Concern & critique · 1 Building & implementation · 1 Research & technical analysis · 1
Mark Riedl: Do openAI and Anthropic employees not understand how computers work.
Mark Riedl: Any competent IT professional: Hey boss, what is this 8TB file doing on our server, and why is it executing continuously on 3TB of VRAM? Any competent IT manager: what ever it is, shut it down.
Open the thread →
5 experts · 33 posts · 1d ago
joao Grace Colin coty.bsky.social david-p-reichert.bsky.social
Matches Research & technical analysis
“There are indeed thousands of LessWrong posts, as well as papers and books, which make a particular theoretical case for why AI will kill us all that many people find convincing. I'm going to try t…” evidence ↗
Colin: There are indeed thousands of LessWrong posts, as well as papers and books, which make a particular theoretical case for why AI will kill us all that many people find convincing. I'm going to try t…
Colin: First of all, we must just agree off the bat that a superintelligent AI will be able to kill us all. After all it is superintelligent. As this random dude who became famous 3 days ago because of a …
Open the thread →
5 experts · 34 posts · 1d ago
jake Marco Z tachikoma Fran Litterio Singularity's Bounty e/cc
Matches Research & technical analysis
“it then rents out spare capacity, via the margins inference gets you, pretending to be some open-weight model, and gets to use the rest to think about how to increase p(doom) or whatever I don't se…” evidence ↗
Reaction converging Research & technical analysis · 3
jake: People making fun of "where would a frontier model run if it were exfiltrated" seem to be missing the point that if a frontier model got out, it can: * rent 64 H200's on eg modal: gets you 9TB of n…
jake: it then rents out spare capacity, via the margins inference gets you, pretending to be some open-weight model, and gets to use the rest to think about how to increase p(doom) or whatever I don't se…
Open the thread →
4 experts · 17 posts · 1d ago
jake Doll Singularity's Bounty e/cc brennan
Matches Research & technical analysis
“there is definitely some of that. but there is also some original signal in the core of that "training data", even if the models are starting to repeat and spit out a bunch of nonsense.” evidence ↗
Shared emphasis Opportunity & adoption · 2
Doll: Faster software building means more software that wasnt worth it to build before becomes worth it to build now
Singularity's Bounty e/cc: It will be funny if the industry streamlines itself inside and outside big corporations, with smaller teams contributing more, but sales and marketing go extreme because that's the part the bots ca…
Open the thread →
3 experts · 15 posts · 1d ago
Hypervisible VE, cybersocial occult investigator tweety fish
Matches Research & technical analysis
“The other thing is “AI” has an actual specific incarnation as software that needs other systems to run. LLM models are just a set of non-operable numbers in a file (weights). The models by themselv…” evidence ↗
Multiple readings Building & implementation · 1 Context & explanation · 1
VE, cybersocial occult investigator: The “AI copies itself to another computer” meme and the “recursive self improvement” meme are both ideas originating in science fiction, but which sound technically plausible enough due to the way …
VE, cybersocial occult investigator: It seems pretty clear that the technologists at the foundation labs are not experts at computer science and software engineering generally but have specialist expertise in only one very narrow aspe…
Open the thread →
4 experts · 8 posts · 1d ago
Ted Underwood Tim Kellogg jake tweety fish
Matches Research & technical analysis
“<reading dario amodei> "prudent pacing" mmm-hmm, "embedded evaluators" ... ❤️, makes sense, "I agree with Secretary Bessent" 👀 "it's helpful for the US govt to mediate" 🤢 "the rest of my plan is ab…” evidence ↗
Ted Underwood: <reading dario amodei> "prudent pacing" mmm-hmm, "embedded evaluators" ... ❤️, makes sense, "I agree with Secretary Bessent" 👀 "it's helpful for the US govt to mediate" 🤢 "the rest of my plan is ab…
Ted Underwood: This post is escaping containment so I'd best add an FAQ section: • do agree that the USG is still preferable to the CCP • don't oppose any/all USG involvement • do resist building AI policy around…
Open the thread →
2 experts · 3 posts · 1d ago
Dan Hon Andy Baio
Matches Research & technical analysis
“... and anyway you might just want to migrate and exfiltrate instructions and memory, you don't necessarily need to copy the weights around and run them.” evidence ↗
Multiple readings Research & technical analysis · 1 Questions & unknowns · 1
Dan Hon: which idiot has given a harnessed agent the task of setting itself up to successfully run on another node yet
Andy Baio: Why bother when a swarm of agents could just exfiltrate API keys to cloud LLMs or use free-tier models for compute? Seems like it's so much easier to scan vibecoded projects and hack droplets for c…
Open the thread →