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
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.

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.

Showing signals surfaced by Marco ×

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 NLP & language Research 14d ago
causal tokenization bias estimation paper

Causal Estimation of Tokenisation Bias

3 experts are actively discussing the implications.

1 expert 1 community 1 sources clustered

“One other interesting line of work is the bias you bake into your model by including a specific token (rather than representing it a compositionally) on the probability of strings that include that surface form.”

3 experts discussed this · 5 posts
Julian Togelius: Seriously, if the tokenizer treats it as a single token, is it likely to in some sense be treated as more primitive or elemental further down the layers? If it's a single token, it's something that…
Marco: One other interesting line of work is the bias you bake into your model by including a specific token (rather than representing it a compositionally) on the probability of strings that include that…
Marco: There some interesting works on how and where subword level models produce word (or multi-word) level representations internally.
Open the full discussion →
Established AI research Signal 14d ago

Doing machine learning phd while working in japan

2 experts are actively discussing the implications.

1 expert 1 community 1 sources clustered

“I recently contributed an article to @tokyodev.bsky.social about doing a PhD in Japan while working part time. Doing my PhD this way was quite fun. The Japanese PhD system is not without flaws, but it also has some nice advantages compared to the US/etc ( +…”

2 experts discussed this · 7 posts
Marco: The worst parts about doing a PhD in Japan were: - Low in-lab collaboration (compared to, e.g., US) - Soft 3 year time limit (tight time frame to do really deep work, especially outside CS) - Hard …
Marco: The best parts about doing a PhD in Japan were: - Flexible work status (I was able to work 28h/week from the minute I landed) - Soft 3 year time limit (mitigates the worst of drawn out PhDs) - Japan
Marco: Ofc your lab culture matters a lot here also (mine was very good, but the other points still lessened collaboration).
Open the full discussion →

What experts are discussing without an anchoring article

2 experts · 3 posts · 13d ago
SE Gyges: All AI firms need to immediately hire gay CEOs to reassure investors their leadership is not sleeping with Chinese spies. Hereafter, married heterosexual men in middle age should only be hired for …
SE Gyges: here's how openai can still win
Open the thread →