Elman Mansimov created alignDRAW, a 2015 model that generated original images directly from written captions rather than retrieving existing pictures, demonstrated with invented scenes such as “a toilet seat sits open in the grass field”.Mansimov led the team of researchers at the University of Toronto that built the model at age 19, and it is credited as the first text-to-image generation model.,
Mansimov started his undergraduate degree in computer science at the University of Toronto at 15 under Ruslan Salakhutdinov, and later completed a PhD at New York University under Kyunghyun Cho. His doctoral thesis focused on iterative refinement for structured prediction, applied to machine translation and molecule generation. He has worked as a Senior Applied Scientist based in New York City.,The alignDRAW research was published in 2015 in the paper “Generating Images from Captions with Attention”.,,The idea of generating images from written prompts was first conceived in that paper, and took the research field years to develop into realistic models such as DALL-E.In 2023 the artist Roope Rainisto made a three-part AI video series that deliberately echoed the raw, unpolished directness of Mansimov's 2015 experiments, using only each work's title as its prompt.
alignDRAW was exhibited at Paris Photo 2023 at the Grand Palais, in a show presented by Louise Alexander Gallery and Fellowship.The alignDRAW image A Stop Sign is Flying in Blue Skies #3, created in 2015, was minted onto the blockchain in 2023.In February 2024 the Worcester Art Museum announced the acquisition of three alignDRAW artworks.,A red school bus parked in a parking lot, from the series alignDRAW, was included in ArtMeta's exhibition at Art Basel's Zero 10. The exhibition traced the history of digital art from 1950s electronic-signal works to AI and blockchain-based practice.
