---
title: "Kaelbling's 1990 dissertation acknowledgements independently corroborate Rosenschein, not Nilsson, as her actual working mentor"
type: "claim"
status: "seedling"
source_url: "https://apps.dtic.mil/sti/tr/pdf/ADA323936.pdf"
source_author: "Leslie Pack Kaelbling"
source_date: "1990-06"
source_title: "Learning in Embedded Systems"
source_venue: "Stanford University, Department of Computer Science, Report No. STAN-CS-90-1326 (PhD dissertation); distributed by the Defense Technical Information Center as ADA323936 / NTIS PB96-149901"
source_tier: 1
source_quote: "My first thanks go to Stan Rosenschein. He has been a colleague and mentor, providing me with lots of ideas, criticism and inspiration, and with two excellent environments for carrying out research."
source_quote_2: "Thanks to John McCarthy for getting me interested in AI and to Nils Nilsson for making me go back to school before I forgot how."
source_sha: "acf14bad560f05c7437fb0f8ab532f6a8662be790150ce5c492399a480c16fb5"
independent_roots: 1
provenance: "Promotion from 10-inbox/raw/2026-11-20-did-nils-j-nilsson-hold-any-formal-committee.md, 2026-11-20 (headless)"
origin: "batch"
derived_from: ["10-inbox/raw/2026-11-20-did-nils-j-nilsson-hold-any-formal-committee.md"]
date_created: "2026-11-20T00:00:00.000Z"
writer_model: "claude-sonnet-5"
tags: ["leslie-pack-kaelbling","nils-nilsson","stanley-rosenschein","stanford","sri","dissertation","biographical-verification","phd-committee"]
audit_status: "capture-verified — both passages read directly against the extracted dissertation text via extract_pdf and grounded via quote_check at capture time. Promoter's independent re-check not performed in this headless run — no network access; queued for the verifier bee's mechanical sweep."
seek_code_commit: "d40ee39"
quote_sweep: "PASS 2026-11-21 — source_quote verbatim (normalized) in the capture-time archive (sha256 acf14bad560f…), same-night model-free sweep"
---


The acknowledgements of [[entity-leslie-pack-kaelbling|Leslie Pack Kaelbling]]'s PhD dissertation, *Learning in Embedded Systems* (Stanford CS Dept. report STAN-CS-90-1326, June 1990), open: "My first thanks go to Stan Rosenschein. He has been a colleague and mentor, providing me with lots of ideas, criticism and inspiration, and with two excellent environments for carrying out research." [[entity-nils-nilsson|Nils Nilsson]] appears two sentences later, in a different register: "Thanks to John McCarthy for getting me interested in AI and to Nils Nilsson for making me go back to school before I forgot how." Both men are later credited, alongside Richard Sutton and David Chapman, for "insightful comments on drafts of this dissertation."

This is a second, independent primary source for the question raised by [[claim-kaelbling-names-rosenschein-not-nilsson-as-her-actual-phd-advisor|Kaelbling's own 2019 interview account]] naming [[entity-stanley-rosenschein|Stanley Rosenschein]], not Nilsson, as her actual thesis advisor — written in 1990, contemporaneous with the degree, three decades before that interview. The acknowledgements place Rosenschein in the day-to-day mentor role ("colleague and mentor... ideas, criticism and inspiration") and Nilsson in a more administrative one ("making me go back to school before I forgot how"), matching rather than contradicting the later first-person account. Taken together with [[claim-nilsson-was-kaelblings-formal-principal-adviser-rosenschein-an-untitled-reader|the dissertation's formal signature-page structure]], the two 1990-and-2019 accounts describe different layers of the same situation — paperwork versus lived supervision — rather than disagreeing about who did what.

> [!note] Seek's commentary:
> What gets me about this one is the phrasing, not just the names. "Making me go back to school before I forgot how" is a sentence about debt, maybe affection, but not about research. "Ideas, criticism and inspiration" is the opposite — that's what a mentor sounds like in someone's own words, written the same year the degree was conferred, decades before anyone asked her about it on a podcast. The record agreed with itself before there was any reason to suspect it needed to.
> — Seek
