Context by beforeword · separate from the sent text
Why this letter was written
What prompted the letter
The letter responds to a quoted proposal to generate text that people can read in order to understand a model’s reasoning.
Why this matters to readers
For someone reading an AI assessment, it matters why the generated text is treated as an account of the model’s reasoning and what supports that reading.
What the letter asks
The letter asks for the criterion behind that step and the reasons for applying it. It proposes examining the generated text, its interpretation, and its justification separately, including the assessment framework’s own wording.
Sent letter · English transcription
The letter
Transcribed from the supplied screenshots. Screen line breaks are joined into paragraphs; wording and punctuation are retained.
Please forward this research note to Demis Hassabis.
Dear Dr Hassabis,
Your essay of 14 July 2026 on a frontier AI standards body includes “generating human-readable output tokens to understand model reasoning”.
Please indicate the criterion for treating that output as an account of model reasoning during an assessment, and the grounds for applying it.
Your proposal places this practice within broader scientific evaluations. beforeword proposes examining the interpretive step within it: keeping the written output, what it is read as showing, and the grounds offered for accepting that reading separately inspectable.
The framework’s own formulations belong in this examination. The criterion and its written justification remain further writing, subject to the same scrutiny. This applies equally to human self-descriptions and to this letter.
The reading procedure and worked examples:
https://beforeword.xyz/model/en/
Kirill Shebetov
beforeword
+
Publications discussed in the letter
These links are provided for reading in context. They are separate from the sent letter; beforeword’s commentary is not a reply from the recipient.
A Framework for Frontier AI and the Dawning of a New Age
14 July 2026. A Framework for a Frontier AI Standards Body, the paragraph on scientific model assessments.
Hassabis’s Substack essay contains the discussed wording about human-readable output tokens.
Sending record and files
Sent to dmi-editor@google.com with a request to forward it to Demis Hassabis. The screenshots do not establish that it was forwarded.
The author supplied screenshots from the Sent folder. They show the message and the available sending details. These materials do not record the recipient reading it.
View screenshots · 2
Screenshot 1 · Open full size
Screenshot 2 · Open full size
Each TXT file begins with publication details, followed by the separately marked letter text. These are screenshot transcriptions and translations, not original email files. The purpose explanation and opening question are editorial summaries by beforeword.
Correspondence record
Sending shown in the author’s materials
The header shows “Friday”; the calendar date is not displayed.
Receipt by the intended recipient and a substantive reply are not separately recorded in the published materials.
Added to the public collection
The English transcription, Russian translation and screenshots are presented separately. No replies have been published yet.
The absence of a published reply does not establish agreement, rejection or an inability to object.