Ellie Pavlick

Brown University · United States

Unclaimed · AI-structured profileORCID OpenAlex

Brown University NLP researcher of model semantics, multimodal memory, and human-AI collaboration

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-17) 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 scored: the snapshot lacks the year-by-year publication records needed to compute an output score. This axis measures publication volume and productivity trajectory; only summary indicators (h-index 31, i10-index 61) are available here.

Impact—

Not scored: citation-level records are insufficient in this snapshot. Impact would measure how strongly the work is cited and taken up across fields; beyond h-index 31 and i10-index 61, no per-paper citation data supports a score.

Focus100/100

Focus scores 100: the OpenAlex topic profile is tightly concentrated in language-centered AI. Topic Modeling (127 indexed works) and Natural Language Processing Techniques (106) lead by a wide margin over Multimodal Machine Learning Applications (35), Domain Adaptation and Few-Shot Learning (29), and Syntax, Semantics, Linguistic Variation (27). The snapshot's single field tag, human-computer interaction (human-AI interaction), reads as an extension of that core into HCI rather than a pivot.

Collaboration44/100

Collaboration scores 44, a mid-range reading. The recent three-year window holds 61 indexed works with 8 distinct coauthors, and recurring partners include Megan Wei, Melanie Subbiah, Audrey Lee, Annya Dahmani, Dave Edwards, Helen Edwards, Athulith Paraselli, and Etha Tianze Hua. On both 2026 Evolve-reviewed papers Pavlick occupies the last, senior author position, though neither lists her as corresponding author.

Activity100/100

Activity scores 100 on sustained, rising output: OpenAlex countsByYear climb from 5 works in 2014 to 28 in 2021, 24 in 2023, 21 in 2024, and a series-high 29 in 2025, with 11 already indexed for 2026. The snapshot records 61 works across the most recent three-year window, an unusually dense pace for a senior researcher.

Reproducibility55/100

Reproducibility scores 55, but the base is thin: the figure is a mean (reproMean 55) over just 1 assessed paper. Within the Evolve-reviewed catalog the pattern is nonetheless consistent: the context-memory-conflict study carries high reproducibility readiness, while the n=64 ownership survey is rated medium, reflecting judge-based scoring and untested percentages.

Research summary

Ellie Pavlick is a researcher at Brown University in the US whose OpenAlex-resolved profile centers on language-focused AI. The topic distribution is tightly concentrated: Topic Modeling (127 indexed works) and Natural Language Processing Techniques (106) lead, ahead of Multimodal Machine Learning Applications (35), Domain Adaptation and Few-Shot Learning (29), and Syntax, Semantics, Linguistic Variation (27). Summary indicators include an h-index of 31 and an i10-index of 61; annual output climbs from 5 indexed works in 2014 to 28 in 2021 and a series-high 29 in 2025, with 61 works in the most recent three-year window. Evolve's catalog adds two 2026 arXiv preprints with Pavlick as last author: an exploratory survey (n=64) of psychological ownership in AI-assisted everyday tasks, judged insightful but sample-biased with medium reproducibility readiness, and a mechanistic-interpretability study of context-memory conflicts in vision-language models, whose causal evidence the review calls strong and which is rated high for reproducibility. Together they show a senior researcher extending a semantics-and-interpretability core toward human-AI interaction, the snapshot's only field tag. Collaboration scores a mid-range 44 across 8 distinct recent coauthors, including recurring partners Megan Wei, Melanie Subbiah, and Audrey Lee. Output and impact remain unscored because per-paper citation records are missing from this snapshot.

Collaboration pattern

Pavlick's collaboration footprint in this snapshot is moderate (score 44): 61 works over the last three years with 8 distinct coauthors. Recurring partners include Megan Wei, Melanie Subbiah, Audrey Lee, Annya Dahmani, Dave Edwards, Helen Edwards, Athulith Paraselli, and Etha Tianze Hua. On both 2026 Evolve-reviewed papers she takes the last, senior author slot without being the corresponding author, consistent with a supervisory role on the submitted work.

Research keywords

natural language processingtopic modelingvision-language modelsmechanistic interpretabilityknowledge conflictretrieval-augmented generationsemanticshuman-AI collaborationpsychological ownershipcognitive offloadingfew-shot learningmultimodal machine learning

Highlights

Causal account of how vision-language models resolve context-memory conflicts
A 2026 arXiv study with Pavlick as last author finds that VLMs favor provided context for text-based entities but rely on memory for image-based ones. Evolve's review calls the causal evidence strong while cautioning that judge-based scoring and small filtered samples warrant care; reproducibility readiness is rated high.
Mapping psychological ownership in AI-assisted everyday tasks
An exploratory survey (n=64) with Pavlick as last author shows that people's sense of ownership over AI-assisted work hinges on process, voice, and comprehension. Evolve calls it insightful but notes a biased sample and untested percentages, with medium reproducibility readiness.
A steady climb to a series-high 29 indexed works in 2025
OpenAlex countsByYear trace output rising from 5 indexed works in 2014 to 28 in 2021, 24 in 2023, 21 in 2024, and a series-high 29 in 2025, with 11 already indexed for 2026 and 61 works in the most recent three-year window — the basis for an activity score of 100.

Timeline

2014Earliest year in the snapshot: 5 works recorded in the OpenAlex countsByYear series.
2019Output accelerates to 14 indexed works, up from 9 in 2018.
202128 indexed works, the strongest year in the series to that point.
202324 indexed works, sustaining the post-2021 pace.
2025A series-high 29 indexed works.
202611 works already indexed; both Evolve-reviewed papers (AI-task ownership survey; modality in context-memory conflicts) appear as arXiv preprints with Pavlick as last author.

Papers in the Evolve catalog

Frequent co-authors

Megan Wei×1 Melanie Subbiah×1 Audrey Lee×1 Annya Dahmani×1 Dave Edwards×1 Helen Edwards×1 Athulith Paraselli×1 Etha Tianze Hua×1

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.

Bibliometrics come from an OpenAlex snapshot; the profile is name-resolved from public bibliographic data and identically named authors may be merged or split.
Output and impact are unscored: only h-index (31) and i10-index (61) are available, with no per-paper citation records in the snapshot.
The reproducibility score (55) averages a single assessed paper; treat it as indicative rather than settled.
Qualitative highlights rest on two 2026 arXiv preprints in the Evolve catalog; earlier work was not reviewed here.