The BaseVideoControlAI881 views

David Robinson: "We grew a mind. Fundamentally, nobody knows why pre-training works"

According to the description, ControlAI posted a clip of David Robinson, described as a former OpenAI safety lead, on how AI systems are grown rather than coded. The description presents Robinson as saying the model was not engineered, only the systems around it, and that the reason pre-training works is unknown.

Open the source ↗

Why it matters

The same thing, for different readers.

What it means for you

A former OpenAI safety lead is described as saying that modern AI is grown, not programmed, and that why it works is not fully understood.

The question it raises

If a system is grown rather than designed, who is responsible for what it turns out to be?

The record

Questions
Can AI systems act beyond what people intended or allowed? · What values should AI have, and could it have a mind of its own?
People
David Robinson (Speaks, Former OpenAI safety lead, per ControlAI)
What we read
We have read only the description the publisher gives. No quotations until someone reads the source itself.
Found by
The Radar, run of October 14, 2026, via yt-controlai (tier 1: the source’s own channel or an official record)
Cite this
ControlAI, “David Robinson: "We grew a mind. Fundamentally, nobody knows why pre-training works"”, October 13, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/controlai-2026-10-13-david-robinson-ais-grown-not-coded
Our notesWritten from the title, channel and description only; the video was not watched, fetched or transcribed. It runs 90 seconds per the feed. The description and the title put words in quotation marks and attribute them to Robinson; none are copied here, and nothing here confirms that Robinson said them or in what context. The description does not say where the clip comes from. Robinson also appears in the Base in coverage of an Ezra Klein interview (story david-robinson-klein-interview-2026-10); whether this clip comes from that interview is not stated. The role of former OpenAI safety lead is as ControlAI gives it. ControlAI campaigns for a ban on superintelligence, so the description is advocacy. The title keeps the quotation marks that ControlAI put in it.