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
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 ↓
Shared emphasis z.ai Research & technical analysis · 2 attributable reactions
Showing signals surfaced by John David Pressman ×
Most important

The developments commanding sustained expert attention

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

Established AI field signal Signal 9d ago
⚡ 26 h early

z.ai

5 experts across 3 network communities independently surfaced this.

5 experts 3 communities 1 sources clustered
Tim Kellogg Sung Kim Miguel Alonso Jr. John David Pressman Z.ai
Primary source z.ai ↗
Established Models & releases Signal 11d ago
⚡ 5 h early

Procedural Knowledge in Pretraining Drives Reasoning in Large Language Models

2 experts are actively discussing the implications.

2 experts 2 communities 1 sources clustered
Tim Kellogg John David Pressman

“this is one of my favorite papers, similar method. It convinced me a that LLMs weren’t merely autocompleting they showed that EVEN WHEN the answer to a math problem was present in the dataset, the agent didn’t reference it, but rather looked at problems wit…”

2 experts discussed this · 7 posts
Tim Kellogg: this is only stuff you can do on open models not just open weights, but fully open models with open datasets this stuff is extremely important, and we know next to nothing about it
Tim Kellogg: this is one of my favorite papers, similar method. It convinced me a that LLMs weren’t merely autocompleting they showed that EVEN WHEN the answer to a math problem was present in the dataset, the …
Tim Kellogg: that one paper got me off the AI skeptic route and pushed me a lot deeper into how LLMs actually work and again, without being able to cross reference the training dataset, methods like this simply…
Open the full discussion →
Primary source arxiv.org ↗
Established AI field signal Resource 4d ago
Snickers HUNGR.AI marketing campaign

SNICKERS® HUNGR.AI | Even AI Can Get Hungry

2 directory members surfaced this signal.

2 experts 2 communities 1 sources clustered
John David Pressman René Walter

“In case anyone is curious what the weird dithered Snickers actually does. www.snickers.com/hungr-ai”

“Advertising by prompt injection is a thing now www.snickers.com/hungr-ai”

Primary source snickers.com ↗
Established Culture, work & education Analysis 8d ago
Weave Agent embodiment goodhart devlog

Weave Agent DevLog #2 - Embodiment, Goodhart, and Grounding

2 experts are actively discussing the implications.

1 expert 1 community 1 sources clustered
John David Pressman

“I tried to write about this here but I buried the lede and didn't quite come at it from the angle of "I need you to understand that arrangements of atoms imply discrete outcomes which push continuous neural nets towards discrete program structure. Your body…”

2 experts discussed this · 3 posts
John David Pressman: I should start doing this with AI insights I need to write about at some point. Insight #1: Discrete programs (or as they're often called for LLM agents, "tools") are a form of embodiment because a…
John David Pressman: I tried to write about this here but I buried the lede and didn't quite come at it from the angle of "I need you to understand that arrangements of atoms imply discrete outcomes which push continuo…
Tim Kellogg: i think you’re right, but i also think the explanation is very simple and LLM without tools is basically identical to a human brain completely disembodied and set on a cold table but kept alive
Open the full discussion →
Primary source minihf.com ↗
Established AI field signal Signal 18d ago

AIs are not people

2 experts are actively discussing the implications.

1 expert 1 community 1 sources clustered
John David Pressman

“This essay is actually AI psychosis. "Fake consciousness" rogueresearch.substack.com/p/ais-are-no...”

2 experts discussed this · 4 posts
John David Pressman: This essay is actually AI psychosis. "Fake consciousness" rogueresearch.substack.com/p/ais-are-no...
Singularity's Bounty e/cc: (haven't read it yet) I always love the distinctions "fake ___" which are operationally indistinguishable from "real ___" Like "if you didn't know any better you couldn't tell them apart"
John David Pressman: Mixtral 8x22B base model text vibes.
Open the full discussion →
Primary source rogueresearch.substack.com ↗
Debates gaining attention

What experts are discussing without an anchoring article

6 experts · 9 posts · 6d ago
Thomas Dietterich Ryan Moulton Colin John David Pressman Max Woolf Alex Gude
Shared emphasis Building & implementation · 2
Ryan Moulton: We are not going to de novo invent a whole new language to describe agents rather than reusing the one language for agents we have. Imagine a star next to all the words if you want. Agents don't ha…
Ryan Moulton: Love is a bit unusual in that respect in that the difference between "outwardly displays all the behaviors of love" and "inwardly feels it" matters a lot to people. That distinction doesn't matter …
Open the thread →
2 experts · 7 posts · 1d ago
John David Pressman Doll
Doll: "nobody should care about a piece of work, that nobody cared to create" is nonsense Keystrokes have never been a unit of caring. Not once. Never has been and never will be. You have smuggled “the h…
Doll: Sometimes someone just generates disposable slop. But humans were quite capable of producing work they didn’t give a shit about before we invented language models and using a language model to inte…
Open the thread →