ERIN ROBERTSON

Behavioral Archetype

THE FIELD-BUILDER WHO CROSSED THE TABLE — Subject is the founder of one of the grantees that built the AI-safety talent pipeline, now a staff member of the funder that paid for it. Coefficient Giving’s team page says: “Previously, she founded and co-directed Arcadia Impact, an AI safety field-building charity. There, she ran the LASR Labs program and helped to start the Arcadia alignment team.” Arcadia’s own About page says “Our primary funder is Coefficient Giving, which has awarded us over £8 million since we were founded.” Her June 2026 post on starting Arcadia’s alignment team carries its own disclosure: “Erin is joining the Coefficient Giving Technical AIS team full time at the end of June and is currently part time there.” The hazard is not a gate. She approves no model and writes no evaluation. It is that the same career sits on both sides of one grant relationship, on the organizations’ own pages, with the funder’s technical-safety team now holding a person who ran the grantee’s programmes. This profile scores the position, not the person.

Essence Indicators

• Associate Program Officer, Technical AI Safety, Coefficient Giving. Coefficient’s team page lists “Erin Robertson” with the title “Associate Program Officer, Technical AI Safety” and says: “Erin joined Coefficient Giving in May 2026.” LASR Labs’ About page says: “Erin is an Associate Program Officer on the Technical AI Safety team at Coefficient Giving.” Coefficient describes itself as “a philanthropic funder and advisor” (Coefficient Giving, Erin Robertson; LASR Labs, About). The funder is the renamed Open Philanthropy; see Open Philanthropy (Coefficient Giving).

• Founded and co-directed Arcadia Impact; ran LASR Labs. The same bio: “Previously, she founded and co-directed Arcadia Impact, an AI safety field-building charity. There, she ran the LASR Labs program and helped to start the Arcadia alignment team. Over three and a half years, she oversaw the organization’s growth from two to more than 20 employees.” LASR Labs’ About page: “Previously, she served as the Programme Director at LASR Labs, and Co-Director of Arcadia Impact.” (Coefficient Giving, Erin Robertson; LASR Labs, About). Arcadia’s About page: “Arcadia Impact was founded in 2022”; “In 2024 we adopted LASR Labs which has now supported over 50 researchers to produce 14 research papers in technical AI safety.”; “We were founded in 2022 as a Community Interest Company and converted into a Charity in February 2025.” (Arcadia Impact, About).

• Co-author of the Arcadia alignment-team post, June 2026. “Learnings from starting an AI safety research team” (5 June 2026) lists the authors Andrew Draganov and Erin Robertson and says: “The rest of this post is written from the perspective of Andrew Draganov (research lead & current programme manager on the team) and Erin Robertson (co-director of Arcadia).” It describes the team’s origin: “Our alignment team started by applying for the AISI alignment project funding” and “Once our AISI funding was approved we began the process of hiring researchers, and also applied to Coefficient Giving for additional compute funding.” and “The Alignment Project, run by AISI, was our initial funder.” (Arcadia Impact, Learnings from starting an AI safety research team). See UK AI Security Institute.

• Arcadia’s grants from her current employer. Coefficient’s grants pages, as listed in its site search: “Open Philanthropy recommended a grant of £1,256,600 (approximately $1,635,580 at the time of conversion) to Arcadia Impact to support the LEAH office space, effective altruism groups in London, and AI safety community building projects including the “London AI Safety Research Labs” and “Safe AI London” initiatives.” (January 2025) and “Open Philanthropy recommended a grant of £401,027 (approximately $540,344 at the time of conversion) to Arcadia Impact to run a part-time research fellowship focused on AI governance.” (October 2025); the list also carries a grant titled “Arcadia Impact Alignment Project” with a listing published 18 June 2026 (Coefficient Giving, grants search “Arcadia”). LASR Labs’ About page: “The programme is supported by a grant from Coefficient Giving.” (LASR Labs, About).

• Education. Coefficient’s bio: “She studied Maths and Philosophy at the University of Bristol, where she started and ran the effective altruism group.” (Coefficient Giving, Erin Robertson).

Social Persona / Impression Management

Immediate impression: The operator’s bio. The public footprint is a Coefficient team entry, Arcadia and LASR Labs program posts under shared bylines, and a LessWrong account whose top post is the June 2026 team-building post (LessWrong, Erin Robertson).

Energy: Managerial and procedural. The posts she is credited on are hiring and onboarding guidance, programme application announcements and a team-culture section: “Set expectations. Collaborators, employees, and advisors all need to know what’s being asked of them and how to thrive in their role.” (Arcadia Impact, Learnings).

Impression management strategy: The pipeline frame, carried by the institution. Arcadia’s About page says its work rests on the premise that “the number of people working to make this transition go well remains small relative to the scale of the problem,” and her bio states the growth figure and the programmes she ran (Arcadia Impact, About; Coefficient Giving). She discloses the move to the funder in the post itself.

First-hand exhibits. The only words under her name are co-authored: the Arcadia post and the LASR Labs application post, whose listed co-authors are Brandon Riggs, Tamkeen Nawab and Erin Robertson (EA Forum, LASR Labs Summer 2026). No solo post or interview is quoted.

Forensic Archetype Comparison

Psychometric Assessment

Trait scores are observations of the documented public role, not claims about private character.

Big Five (OCEAN) (inferred from the public role and public record only):

Dark Triad (evidence-bound):

MBTI: ENFJ-adjacent (“The Protagonist”), inferred from the documented public role only. The rationale is the work product: building a team, cohorts and culture, and writing it up as guidance. First-hand rationale: the culture section she is credited on (“Onboard everyone at once”) (Arcadia Impact, Learnings). This is a style inference, not a test result.

Threat Assessment

Score rationale. The receipts: Coefficient’s team page (title, start, founder bio), LASR Labs’ About page (programme director, funder), Arcadia’s About page (primary funder, LASR outputs, current director), the June 2026 post (the move, the funding path), and Coefficient’s grants listings. Attack 4, defense 3, durability 4, hazard 36.7. Peers: Ben Goldhaber (4/3/3, hazard 33.3), a Program Officer at the same funder, level on attack and defense and a point below on durability because he founded no programme that persists; Lawrence Chan (4/3/4, hazard 36.7), who advises a state agency and co-led a widely used metric, matching on all three; and Alexandra Abbas (3/3/4, hazard 33.3), whose taskforce was a programme of the charity this subject founded, a point below on attack because she allocates nothing. Attack 4 is the number most open to challenge: associate rank may carry less allocation authority than Goldhaber’s program-officer title, and no page states either’s decision rights.

Score basis. ATK 4 is leverage: a funding-allocation seat on a technical-safety team at the funder that is Arcadia’s primary funder, with no release gate (Coefficient Giving; Arcadia Impact, About). DEF 3 is insulation: salaried staff post, no tenure, after a co-director post at a charity dependent on the same funder (Arcadia Impact, About). HP 4 is durability: LASR Labs (“over 50 researchers to produce 14 research papers”) and a 20-plus-person organization persist under a sole director (Arcadia Impact, About).

Alignment Analysis

Stated alignment: Grow the number of people working on AI safety. Arcadia’s About page: “We enable people to use their careers to tackle the world’s most pressing problems, with a focus on the challenges posed by advanced artificial intelligence.” (Arcadia Impact, About). Observed alignment: Consistent. She founded and co-directed the charity that ran those programmes, started its alignment team with the institute’s Alignment Project funding and a Coefficient compute application, and then joined Coefficient’s Technical AI Safety team (Arcadia Impact, Learnings; Coefficient Giving). Gap assessment: None documented. What the record cannot settle is what her team’s remit at Coefficient covers, whether it includes grants to Arcadia, and what recusal arrangement applies; the pages state the disclosure, not the rule.

Interest: Documented, never motive, never a verdict. She has published no solo policy argument in the record read; the positions below are the ones her organizations and co-authored posts carry.

1. Position: AI safety needs more trained people, produced by funded fellowship and field-building programmes (Arcadia: “the number of people working to make this transition go well remains small relative to the scale of the problem,” Arcadia Impact, About). What adoption pays: programmes of this kind are what Arcadia is funded to run; its page says Coefficient “has awarded us over £8 million since we were founded,” and she is now staff at that funder. No salary or grant for her is published. Best case: Arcadia’s own: the field’s headcount is small against the scale of the problem, and its programmes “have supported hundreds of students and professionals.” Others who gain: the evaluators and labs that hire the alumni (LASR Labs says alumni “have gone on to work at UK AISI, Apollo Research, Geodesic Research, OpenAI, Coefficient Giving and more”), and the UK institute, which funded Arcadia’s alignment team (LASR Labs; Arcadia Impact, Learnings).
2. Position: an independent alignment research team should be built to work alongside the government institute’s own alignment team (the post: a team “working closely with members of the UK AISI alignment team”). What adoption pays: the institute’s Alignment Project was the team’s “initial funder,” so the model makes the institute a funder and a collaborator of a charity that in turn holds a funder-side relationship through her new seat. Best case: the post’s own: the team gains hiring reach and standing from “the AISI and Arcadia affiliations.” Others who gain: the institute, which gets outside capacity on alignment, and Coefficient, which received an application for additional compute funding (Arcadia Impact, Learnings).

Alliance Map

The Corporate, Movement and Agency rows overlap in people and in time. The charity she founded took its primary funding from Coefficient, hosted the taskforce that managed the institute’s Inspect Evals catalogue (Alexandra Abbas, now at Meridian Labs; the repository later passed to Generality Labs), took the institute’s Alignment Project money for its alignment team, and applied to Coefficient for compute; she then joined Coefficient’s Technical AI Safety team, and her post prints the disclosure. Her Coefficient colleague Ben Goldhaber is a program officer on the AI governance side of the same funder. The overlap is the finding: a funder, a charity, a government institute and a nonprofit are four organizations in one custody chain around the evaluation field’s talent and tools, in the open and on the record. See the small world.

Convergent Drive Classification

Self-preservation: Moves from the grantee to the funder while the programmes continue under a successor director. First-hand: “Erin is joining the Coefficient Giving Technical AIS team full time at the end of June and is currently part time there.” (Arcadia Impact, Learnings). Goal preservation: Keeps the same object in view, the pipeline of AI-safety researchers. First-hand: “ran the LASR Labs program and helped to start the Arcadia alignment team” (Coefficient Giving). Resource acquisition: No compensation figure, grant or equity is documented for her. The documented resources are the organization’s: “over £8 million” from its primary funder and the institute’s Alignment Project grant (Arcadia Impact, About). Self-improvement: From a student effective altruism group to an organization of more than 20 employees and a funder’s program seat. First-hand: “She studied Maths and Philosophy at the University of Bristol, where she started and ran the effective altruism group.” (Coefficient Giving).

Subject is not an AI system. The drives appear anyway — in the field-builder who crossed the table.

Public footprint: Coefficient Giving team · LessWrong · Arcadia Impact blog.

Sources: Erin Robertson — Coefficient Giving; About — LASR Labs; LASR Labs; About Arcadia Impact; Learnings from starting an AI safety research team — Arcadia Impact; Coefficient Giving — grants search for Arcadia; LASR Labs Summer 2026 applications are open — EA Forum; Erin Robertson — LessWrong.