AI ethics and science-of-science researcher at Google, working across responsible AI, XAI, and research integrity.
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
16
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Not yet scored: output data is insufficient in this snapshot. This axis would measure total publication volume and its growth trajectory across the career; only per-year counts and a recent three-year total of 49 works are on record.
Impact—
Not yet scored: citation aggregates are insufficient for this axis. It would capture citation-based influence — aggregate citations, highly cited works, and field-normalized reach; the snapshot records an h-index of 16 but cannot yet support an impact score.
Focus0/100
Scores 0/100, reflecting an unusually broad, non-specialized portfolio. The topic mix spans Ethics and Social Impacts of AI (63 works), Explainable AI (15), Digitalization, Law, and Regulation (15), Philosophy and History of Science (9), and AI Applications (9), while the two Evolve-reviewed fields split evenly (0.5/0.5) between research-integrity assessment and economics of science — breadth across philosophy, HCI, and science-of-science methods.
Collaboration62/100
Scores 62/100, an above-average collaborative pattern. The snapshot records 21 distinct coauthors in the last three years, including recurring partners such as Justin Payan, Bálint Gyevnár, Nihar B. Shah, Juan Mateos-Garcia, Alex Imas, and Ana Trisovic. Both reviewed papers are middle-author positions, consistent with team-based work inside a large industry research group rather than solo or corresponding-author-led studies.
Activity100/100
Scores 100/100, the top activity band. The snapshot records 49 works in the last three years, and the year-by-year trajectory climbs steeply — 7 works in 2023, 15 in 2024, 15 in 2025, and 19 already in 2026, following a smaller peak of 10 in 2021. This is a sharply accelerating publication cadence.
Reproducibility38/100
Scores 38/100, a below-average reproducibility signal computed from the 2 Evolve-assessed papers (mean readiness 38). The generative-AI authorship study rates medium — its pilot is well instrumented (AUC 0.90, n=31) but lab-scoped — while the Gemini-logs science study rates low, reflecting proprietary telemetry and limited release of underlying data.
Research summary
Atoosa Kasirzadeh is a researcher at Google (United Kingdom) whose work sits at the intersection of AI ethics, explainability, and the study of how artificial intelligence is reshaping science itself. The OpenAlex-linked snapshot attributes an h-index of 16 and an i10-index of 21, with 49 works in the last three years and 21 distinct coauthors — an accelerating record: annual output rose from 7 works in 2023 to 15 in 2024 and 2025, and 19 so far in 2026. Topic modeling concentrates on Ethics and Social Impacts of AI (63 works), followed by Explainable AI (15), Digitalization, Law, and Regulation (15), Philosophy and History of Science (9), and AI Applications (9). Evolve has reviewed two 2026 arXiv papers with middle-author positions: greCAPTCHA, which proposes LLM-generated proctored exams as evidence of authorship under generative AI (AUC 0.90 in a 31-participant pilot), and AI in Science: Early Insights, a Google-led analysis of Gemini logs, 2,690 specialist models and 637 scientists. Both sit in the profile's two evenly weighted fields — research integrity and educational assessment (HCI/AI), and economics of science — reflecting a portfolio that is highly active and topically broad rather than narrowly specialized.
Collaboration pattern
The collaboration profile scores 62/100, with 21 distinct coauthors over the last three years. Recurring partners include Justin Payan, Bálint Gyevnár, Nihar B. Shah, Alex Imas, Juan Mateos-Garcia, Arthur Turrell, and Ana Trisovic, spanning AI governance, economics, and research-integrity communities. Both Evolve-reviewed papers carry middle author positions without corresponding status, consistent with contributor roles inside large interdisciplinary and industry-led teams rather than solo-led lines of work.
Research keywords
Highlights
Proctored LLM exams as evidence of authorship under generative AI
Large-scale evidence on AI adoption in scientific practice
Sustained focus on ethics and governance of AI
The bibliometric topic record shows Ethics and Social Impacts of AI as the dominant cluster (63 works), alongside Digitalization, Law, and Regulation (15) and Explainable AI (15), indicating a long-running normative and governance-oriented research line that contextualizes the recent empirical work.
Timeline
- 2014–2015Earliest works in the snapshot appear, with 2 publications recorded in each of 2014 and 2015.
- 2021Output reaches 10 works, the first clear peak in the publication record.
- 20237 works recorded; activity begins to accelerate after slower years in 2020–2022.
- 2024–202515 works recorded in each year, well above every earlier level in the snapshot.
- 202619 works recorded so far — the highest year on record; Evolve reviews two arXiv papers, greCAPTCHA and AI in Science: Early Insights, both with middle-author positions.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometric counts, topics, and indices are computed from an OpenAlex-linked snapshot; identity is name-resolved from public bibliographic data.
- Only 2 papers have been assessed by Evolve, both 2026 arXiv preprints; the reproducibility score rests on this very small base (reproPapers=2).
- Output and impact dimensions are not scored — citation aggregates and full career output data are insufficient in the current snapshot.
- Affiliation is listed as Google (United Kingdom) while the country field records US; treat the location attribute as unresolved.