PeopleofSuperintelligence

Researcher

Dustin Schwenk

Dustin Schwenk is an AI researcher whose work spans visual question answering, image generation, 3D data and language models. He coauthored work on a textbook-science question-answering dataset, CRAFT for video generation from captions, X-LXMERT for text-to-image generation, and Iconary for multimodal communication. His later work includes the A-OKVQA benchmark, the 800,000-object Objaverse dataset, and OLMo, a language model released with open training data and code; in 2025, he contributed to Olmo 3’s model architecture, infrastructure, and training methodology at the Allen Institute for AI.

Timeline

  1. 2017

    Textbook Question Answering

    In 2017, he coauthors “Are You Smarter Than a Sixth Grader?”, which introduces the Textbook Question Answering dataset with 1,076 lessons and 26,260 multimodal questions from middle-school science curricula.

  2. 2018

    Scripts to video scenes

    In 2018, he coauthors “Imagine This! Scripts to Compositions to Videos,” introducing CRAFT, a model that retrieves video segments and fuses them to generate scenes from new captions.

  3. 2020

    X-LXMERT learns to paint

    In 2020, he coauthors X-LXMERT, a transformer extension that generates images from text while keeping question-answering and captioning abilities comparable to LXMERT.

  4. 2021

    Iconary for multimodal communication

    In 2021, he coauthors Iconary, a drawing-and-guessing game built around icon compositions. Its AI models train on over 55,000 games between human players and use language-model world knowledge to handle words unseen during training.

  5. 2022

    A-OKVQA and world knowledge

    In 2022, he coauthors A-OKVQA, a crowdsourced visual question-answering dataset with about 25,000 questions that require commonsense and world knowledge beyond the image.

  6. 2023

    Objaverse and 3D objects

    In 2023, he coauthors Objaverse, a dataset of more than 800,000 3D models with descriptive captions, tags, and animations. Its uses include 3D generation, tail-category segmentation, embodied object navigation, and vision-model robustness testing.

  7. 2024

    An open language model

    In 2024, he coauthors OLMo, a language model released with open training data, training code, and evaluation code.

  8. 2025

    Olmo 3 research at Ai2

    In 2025, he contributes to Olmo 3’s model architecture, infrastructure, and training methodology at the Allen Institute for AI.

  1. Google had the most people shaping AI every year from 2014 to 2025Our estimates of which organizations employed the most people shaping AI from 1961 to 2025, with a year-by-year chart.
  2. The United States employs an estimated 53% of people in our data with a known employment countryOur estimates of where AI careers are concentrated now and how countries’ shares of dated positions changed over time.
  3. 46% of living people in our data with known birth and employment countries work outside their birth countryOur estimate of cross-border careers in AI, from country of birth and the country of each current organization.
  4. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionOur estimates of where the people shaping AI studied and where their careers led, from education records in cited biographies.

Sources and credits

Photo: We have no published photo of Dustin Schwenk yet. Send us a free-licensed photo.

  1. Are You Smarter Than a Sixth Grader? Textbook Question Answering for Multimodal Machine Comprehension openaccess.thecvf.com. Checked 19 October 2026.
  2. Imagine This! Scripts to Compositions to Videos ecva.net. Checked 19 October 2026.
  3. X-LXMERT: Paint, Caption and Answer Questions with Multi-Modal Transformers arxiv.org. Checked 19 October 2026.
  4. Iconary: A Pictionary-Based Game for Testing Multimodal Communication with Drawings and Text aclanthology.org. Checked 19 October 2026.
  5. A-OKVQA: A Benchmark for Visual Question Answering using World Knowledge arxiv.org. Checked 19 October 2026.
  6. Objaverse: A Universe of Annotated 3D Objects openaccess.thecvf.com. Checked 19 October 2026.
  7. OLMo: Accelerating the Science of Language Models aclanthology.org. Checked 19 October 2026.
  8. Olmo 3 arxiv.org. Checked 19 October 2026.