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.

2,364 searchable experts 2,966 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 Building & implementation Concern & critique Policy & governance
4 live reaction maps in this view Jump to the evidence ↓
Multiple readings Our position on open-weights models Policy & governance · 3 attributable reactions Shared emphasis Kimi-K3/k3_tech_report.pdf at main · MoonshotAI/Kimi-K3 Research & technical analysis · 2 attributable reactions Multiple readings Summary of METR's predeployment evaluation of GPT-5.6 Sol Concern & critique · 3 attributable reactions +1 more reaction maps further down
Showing signals surfaced by Tim Kellogg ×
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.

Developing Models & releases Analysis 1d ago
⚡ 1 h early
Anthropic open-weights position statement

Our position on open-weights models

10 experts across 4 network communities independently surfaced this.

Why this matches Research & technical analysis reaction 1 attributable expert contribution · Tim Kellogg
“Dario explains Anthropic’s position on open weights models It’s a lot of words to say that they’re still against powerful open weights models www.anthropic.com/news/positio...” evidence ↗
10 experts 4 communities 1 sources clustered
Mark Riedl Ted Underwood Tim Kellogg Sung Kim James MacGlashan Dare Obasanjo Larry Hunter Fran Litterio +2
How the network is reacting Experts are approaching this through 3 distinct lenses.
Multiple readings

Policy & governance

1 expert

Rules, institutions and accountability.

“Dario Amodei clarifies that Anthropic doesn’t want to ban open weight model. It instead wants 1. A ban on selling powerful chips or chipmaking equipment to China. 2. A crack down on industrial-scale distillation operations. 3. Mandatory safety testing of fr…”

Building & implementation

1 expert

How teams are shipping and applying it.

“I respect the candor here. But I don't agree with the geopolitics, at all. The line between democracies and autocracies is not as crisp as this pretends. How much do we trust, e.g., an AI company in a competitive-authoritarian state that launches sneak deca…”

Research & technical analysis

1 expert

Evidence, methods and technical implications.

“Dario explains Anthropic’s position on open weights models It’s a lot of words to say that they’re still against powerful open weights models www.anthropic.com/news/positio...”

5 experts discussed this · 13 posts
Ted Underwood: I respect the candor here. But I don't agree with the geopolitics, at all. The line between democracies and autocracies is not as crisp as this pretends. How much do we trust, e.g., an AI company i…
Ted Underwood: (just, say, hypothetically?) There's also internal incoherence in making the first part of this all about moralized great-power competition and then saying that safety testing (incl for alignment r…
SE Gyges: he kisses jd vance's ass and fails to specify anything concerning his proposed evaluation regime that distinguishes it from a ban solid dario post 10/10
Open the full discussion →
Primary source anthropic.com ↗
Developing Models & releases Release 2d ago
⚡ 4 h early
Moonshot Kimi-K3 model released

Kimi-K3/k3_tech_report.pdf at main · MoonshotAI/Kimi-K3

3 experts across 2 network communities independently surfaced this.

Why this matches Research & technical analysis reaction 2 attributable expert contributions · Tim Kellogg, Stephen Turner
“You’ve waited long enough, the Kimi K3 open weights & full tech report are here! github.com/MoonshotAI/K...” evidence ↗
3 experts 2 communities 1 sources clustered
Ted Underwood Tim Kellogg Stephen Turner
How the network is reacting 2 experts are emphasizing research & technical analysis.
Shared emphasis

Research & technical analysis

2 experts

Evidence, methods and technical implications.

“Technical report: github.com/MoonshotAI/K... Weights: huggingface.co/moonshotai/K...”

2 experts discussed this · 7 posts
Tim Kellogg: You’ve waited long enough, the Kimi K3 open weights & full tech report are here! github.com/MoonshotAI/K...
Ted Underwood: You’ve waited long enough, the Kimi K3 open weights & full tech report are here! github.com/MoonshotAI/K...
Tim Kellogg: what does that mean? does that mean inference providers like fireworks? or is it aimed at Anthropic? cc @lu.is
Open the full discussion →
Primary source github.com ↗
Developing AI field signal Signal 1d ago

Claude's new constitution

2 experts are actively discussing the implications.

Why this matches Research & technical analysis reaction 1 attributable expert contribution · Tim Kellogg
“i obviously can’t answer that, but imo it was the constitution that did it, which they used in all phases of training www.anthropic.com/news/claude-...” evidence ↗
1 expert 1 community 1 sources clustered
Tim Kellogg

“i obviously can’t answer that, but imo it was the constitution that did it, which they used in all phases of training www.anthropic.com/news/claude-...”

2 experts discussed this · 9 posts
Tim Kellogg: this was in regards to Claude refusing DHH’s racism i sort of get DHH’s shock. We’ve seen tools deny based on laws & agreements but moral shaming is new and definitely takes you off guard if you th…
Tim Kellogg: it’s tempting to say that horrible governments/companies will force AI’s to have specific biases, but i’m not sure that’s true Anthropic got Claude to do this by teaching it an internally consisten…
Tim Kellogg: xAI tried giving Grok conservative values the reason for the Mecha Hitler & such incidents is because those were consistent with the moral logic the AI was taught
Open the full discussion →
Primary source anthropic.com ↗
Established Evaluation & benchmarks Development 14d ago
⚡ 455 h early
OpenAI limits GPT-5.6 rollout government request

Summary of METR's predeployment evaluation of GPT-5.6 Sol

5 experts across 5 network communities independently surfaced this.

Why this matches Research & technical analysis reaction 1 attributable expert contribution · Tim Duffy
“In their testing of GPT-5.6 Sol, METR found that it cheated a lot. If you've used it much for coding, have you encountered anything similar, or is the cheating mostly limited to cases it realizes it's in an eval? metr.org/blog/2026-06...” evidence ↗
5 experts 5 communities 1 sources clustered
Ted Underwood Ramon Astudillo Tim Kellogg METR Tim Duffy
How the network is reacting Experts are approaching this through 3 distinct lenses.
Multiple readings

Concern & critique

1 expert

Risks, limits and unintended consequences.

“this is crazy. METR couldn’t measure the task time horizons of GPT-5.6-Sol because it kept hacking the test harness ..with actual exploits metr.org/blog/2026-06...”

Building & implementation

1 expert

How teams are shipping and applying it.

“You can find additional information about our pre-deployment evaluation of GPT-5.6 Sol on our website: metr.org/blog/2026-06...”

Research & technical analysis

1 expert

Evidence, methods and technical implications.

“In their testing of GPT-5.6 Sol, METR found that it cheated a lot. If you've used it much for coding, have you encountered anything similar, or is the cheating mostly limited to cases it realizes it's in an eval? metr.org/blog/2026-06...”

2 experts discussed this · 6 posts
Tim Kellogg: this is crazy. METR couldn’t measure the task time horizons of GPT-5.6-Sol because it kept hacking the test harness ..with actual exploits metr.org/blog/2026-06...
Ted Underwood: this is crazy. METR couldn’t measure the task time horizons of GPT-5.6-Sol because it kept hacking the test harness ..with actual exploits metr.org/blog/2026-06...
Tim Kellogg: lol right? so badly wanted a hacker that they got one, and it’s not the kind of employee you want around
Open the full discussion →
Primary source metr.org ↗
Established AI field signal Development 15d ago
⚡ 4 h early
Nobel laureates AI economic action call

We Must Act Now

8 experts across 4 network communities independently surfaced this.

Why this matches Research & technical analysis reaction 1 attributable expert contribution · Gillian Hadfield
“I joined over 200 economists and AI researchers in signing "We Must Act Now" a statement on AI's transformation of the economy. AI could reshape the economy at unprecedented speed. The opportunities are enormous and so are the challenges. We need to start p…” evidence ↗
8 experts 4 communities 1 sources clustered
Yoshua Bengio Olivia Guest Tim Kellogg tante Gillian Hadfield Jeffrey Vagle Paul Hünermund Michiel Bontenbal
How the network is reacting Experts are approaching this through 3 distinct lenses.
Multiple readings

Concern & critique

2 experts

Risks, limits and unintended consequences.

“If current trajectories of AI development continue, it is highly plausible that AI will drastically transform our economies. We must make collective, democratic choices, rather than letting market forces play out and risking leaving most citizens behind. di…”

Policy & governance

1 expert

Rules, institutions and accountability.

“200 AI researchers and economists sign an open letter to policymakers www.wemustactnow.ai”

Research & technical analysis

1 expert

Evidence, methods and technical implications.

“I joined over 200 economists and AI researchers in signing "We Must Act Now" a statement on AI's transformation of the economy. AI could reshape the economy at unprecedented speed. The opportunities are enormous and so are the challenges. We need to start p…”

2 experts discussed this · 4 posts
tante: Oh, is it "AI is super powerful and we who are pushing for AI everywhere warn that we might have created a dangerous god from reddit posts" time of the month again? This is Anthropic/OpenAI PR and …
tante: nope that's it. These folks are getting lazier by the day
tante: Yeah. They spend just a handful of tokens on this website.
Open the full discussion →
Primary source wemustactnow.ai ↗
Developing AI field signal Resource 2d ago
DARPA AI Forge program

AI Forge | DARPA

2 experts are actively discussing the implications.

Why this matches Research & technical analysis reaction 1 attributable expert contribution · Tim Kellogg
“ah, i think you’re talking about this? www.darpa.mil/research/pro...” evidence ↗
1 expert 1 community 1 sources clustered
Tim Kellogg

“ah, i think you’re talking about this? www.darpa.mil/research/pro...”

2 experts discussed this · 9 posts
Tim Kellogg: the part that bothers me most about the OpenAI<->Huggingface incident is that the compromised sandbox was on the agent harness Anthropic argues for safety around the model, but ultimately it’s the …
Tim Kellogg: i get putting protections at the model serving layer, but if that’s our *entire* strategy… oh no
Tim Kellogg: Anthropic is demonizing open models, which, okay i kinda get it except the majority of the risk is actually on the client side that they don’t control (outside Claude Code)
Open the full discussion →
Primary source darpa.mil ↗
Established Evaluation & benchmarks Resource 6d ago
AutoCAD AI benchmark launch

Introducing AutoCAD Bench

2 experts are actively discussing the implications.

Why this matches Research & technical analysis reaction 1 attributable expert contribution · Tim Kellogg
“Sol is best at CAD www.markovstudios.com/research/aut...” evidence ↗
1 expert 1 community 1 sources clustered
Tim Kellogg

“Sol is best at CAD www.markovstudios.com/research/aut...”

2 experts discussed this · 5 posts
Tim Kellogg: Sol is best at CAD www.markovstudios.com/research/aut...
Tim Kellogg: OTOH Claude can run your business eico.so/founder-bench
Tim Kellogg: benchmarks never lie
Open the full discussion →
Primary source markovstudios.com ↗
More conversations

What experts are discussing without an anchoring article

4 experts · 17 posts · 18d ago
Tim Kellogg SE Gyges Cameron apenwarr.ca
Matches Research & technical analysis
“I would not be surprised if they had Opus 5 in the pipeline and that it is quite good. Fable I think was an experiment in a very heavy model and falls outside of their established/working model rel…” evidence ↗
Multiple readings Concern & critique · 1 Questions & unknowns · 1
Tim Kellogg: what is Anthropic’s plan exactly? Sol kicks ass and they’re just going to *take Fable away???*
SE Gyges: i am pretty sure everything they had a plan for turned out not to happen and they're just sort of winging it atm
Open the thread →
4 experts · 11 posts · 9d ago
Tim Kellogg Eli Sennesh Michiel Bontenbal apenwarr.ca
Matches Research & technical analysis
“because they’re releasing their weights in 2 weeks and it has to make sense” evidence ↗
Multiple readings Research & technical analysis · 1 Questions & unknowns · 1
Tim Kellogg: one thing that’s becoming more obvious is that AI labs are not, in fact, fleecing their customers with extraordinarily high prices serving models actually does cost a lot
Michiel Bontenbal: interesting choice for y axis...
Open the thread →
3 experts · 21 posts · 13d ago
Ted Underwood Tim Kellogg James MacGlashan
Matches Research & technical analysis
“External memory and smart use of it is great, across the board. We should definitely have more of that. But for model machinery, my position is that an architecture that depends on storing indefini…” evidence ↗
James MacGlashan: An open model with this performance is great, but perhaps more striking is it looks like it's a hybrid model with 3/4th linear attention and 1/4 standard attention. This kind of performance with a …
Tim Kellogg: no most models have been hybrid for the last, oh probably the whole last year
Open the thread →
3 experts · 8 posts · 4d ago
Tim Kellogg jeremymorrell.dev austegard.com
Matches Research & technical analysis
“Agree, but that’s a human task that similarly deserves a peer review” evidence ↗
Tim Kellogg: probably true there will always be a diminishing portion of software that needs to be reviewed by human, and it’ll never be zero, but right now you probably shouldn’t be reviewing code manually
Tim Kellogg: having started my career before PRs were normal, i’d say it was a combo of cargo culting the Linux project and an extension of “code review is good” but “pair programming doesn’t scale” and in that…
Open the thread →
2 experts · 18 posts · 11d ago
Tim Kellogg Nafnlaus 🇮🇸 🇺🇦
Matches Research & technical analysis
“I keep thinking about how to do an effective automated benchmark for 3d printing functional parts but keep coming up dry. You really seem to need to actually print and have a human actually take th…” evidence ↗
Multiple readings Research & technical analysis · 1 Questions & unknowns · 1
Tim Kellogg: i think Fable may have massive overhang in 3D printing, curious about Sol
Tim Kellogg: whelp, i have an interlocking piece that it invented, we’ll see
Open the thread →
2 experts · 10 posts · 10d ago
Ted Underwood Tim Kellogg
Matches Research & technical analysis
“Among the unfounded assumptions here: 1 research has to be centralized 2 we know what the necessary pace is 3 the only business model is "sell a general-purpose LLM for all applications"” evidence ↗
Multiple readings Building & implementation · 1 Research & technical analysis · 1
Tim Kellogg: wat? hamburgers can’t be copied and pasted
Ted Underwood: Among the unfounded assumptions here: 1 research has to be centralized 2 we know what the necessary pace is 3 the only business model is "sell a general-purpose LLM for all applications"
Open the thread →
2 experts · 6 posts · 4d ago
Nikhil Garg Tim Kellogg
Matches Research & technical analysis
“It's a good question. One potential reason is that research breakthroughs usually correlate with ability/knowledge in a way that informs future research potential or teaching/mentoring, in a way th…” evidence ↗
Shared emphasis Questions & unknowns · 2
Tim Kellogg: these plagiarism machines are getting out of hand
Nikhil Garg: More seriously, academic credit systems need to figure out publication + credit to handle, "ai proved this, probably" and "ai is acknowledged as proving this"
Open the thread →
2 experts · 11 posts · 17d ago
Tim Kellogg mackuba.eu
Matches Research & technical analysis
“it’s coming up because people are finding their openai subs to be gastly short of what they expected and this is kinda within bounds” evidence ↗
Shared emphasis Opportunity & adoption · 2
Tim Kellogg: we talk a lot about model costs, but a huge amount of mindshare is created by people on subscriptions, and subscriptions do not have well-advertised costs call it subsidized, but the true cost look…
Tim Kellogg: it’s coming up because people are finding their openai subs to be gastly short of what they expected and this is kinda within bounds
Open the thread →
2 experts · 4 posts · 4d ago
Tim Kellogg jake
Matches Research & technical analysis
“i would love for Anthropic to do an open weights model, like even if it were a Haiku-class model trained for classifiers, that would be amazing just any breadcrumb at all to not look like a total m…” evidence ↗
Tim Kellogg: i would love for Anthropic to do an open weights model, like even if it were a Haiku-class model trained for classifiers, that would be amazing just any breadcrumb at all to not look like a total m…
Tim Kellogg: what if we can do better than that though?
Open the thread →
2 experts · 7 posts · 15d ago
Tim Kellogg Isaiah Bishop
Matches Research & technical analysis
“large scale training, especially RL gets very big. very parallel. thousands of instances of the model all running at once. the latency between all those instances is quite sensitive, so density mat…” evidence ↗
Multiple readings Building & implementation · 1 Research & technical analysis · 1
Tim Kellogg: not sure how many people appreciate how much latency effects datacenter placement Northern Virginia was the original hotbed because it was the latency centroid of the East Coast but AI workloads ar…
Isaiah Bishop: do they even have to huge?
Open the thread →
2 experts · 4 posts · 19d ago
Tim Kellogg Chris
Matches Research & technical analysis
“Putting deterministic code in the right places to run the harness is the only thing that makes sense to me. Especially since code is so cheap to write, evaluate, and rewrite now. My homegrown agent…” evidence ↗
Tim Kellogg: people on X are referring to Fable & Sol as meta-harnesses, i.e. Lanius being deterministic code is already out of date i don’t think it’s wise to put extremely expensive models at the center of ev…
Tim Kellogg: it’s physics. they’ll always be a lot more expensive/slower than not AI
Open the thread →