Issue 1 · The Scaling Bet
Yann LeCun
Issue 1 · Different machine
From the public record

Yann LeCun

LeCun names Amodei, Altman, and Hassabis directly on architecture, regulation, and labor; none name him back — the asymmetry holds because he has no commercial consensus to protect.

Issue 1 cover
In Issue 1: The Scaling Bet
LeCun is one of the ten voices in this issue.
The bet against the language model
Read the issue

Positions

4 published · 1 in current issue
01

The bet against the language model

LeCun's bet is that improvement on the autoregressive-transformer track is improvement on the wrong axis, and that a system trained on video with latent-space prediction will do things the largest LLM cannot.

in Issue 1 · Different machine
ready
02

The Wikipedia argument

LeCun argues open-weight foundation models will out-compete closed models as Wikipedia out-competed commercial encyclopedias, with AI stack value migrating to the application layer regardless of how scaling resolves.

ready
03

The capture charge

On October 29, 2023, LeCun accused Altman, Hassabis and Amodei of lobbying for regulatory capture. The charge can be tested against the rules they proposed, but LeCun has not pressed it in any formal forum.

ready
04

What LeCun thinks the alarmists got wrong

LeCun says fears of AI catastrophe rest on the myth of a hard takeoff, and that AI is just another technological revolution for jobs. Both arguments tell the worriers they heed the wrong experts.

ready
Share Share on X

Feedback

Issue 1 · Draft v3

Please don’t include personal information.

deslop.media editorial. Issue 1, Draft v3, first published October 10, 2026. Reporting current through October 2026. Use the flags on this page to report an error or a disagreement. Changes to the issue are listed on the changes page.

Part of Issue 1 · The Scaling Bet