Eight surviving recordings on computational taste, machine learning and art, and the design of a life. Each surviving file appears in full, except the segment that is all that remains of a longer MoMA talk.
Can a machine imagine how a recipe will taste without pretending to replace the cook? In 18 minutes, the Synesthetic Cookbook and Gulp Fiction move from sensory association to recipes composed from 160,000 examples. Restored from the original tape.
A 45:36 talk asking whether machines can make judgments about beauty, taste, and value—and whether a more transparent, data-driven art market can turn more of the people who visit museums into people who live with art.
Mike Steib, then Artsy’s CEO, found the recording and posted it himself in 2020.
A 35:29 main-stage keynote on taste graphs, brand neighborhoods, the power of late adopters, and the correlations hidden inside ordinary preferences.
The room left an unusually vivid record. One listener’s only complaint was that it was too short: they could have listened all day.
A 4:02 conversation with Mats Stålbröst after the Webbdagarna keynote. The recording survives under the publisher’s mark; this page’s title is editorial because the original program title has not been recovered.
A 2:07 Swedish television segment in a supermarket, testing what ordinary food choices might reveal about taste. The lower third identifies me simply as a taste researcher.
A 9:45 invited talk at Microsoft Research’s Social Computing Symposium. Through “The Shampoo-Conditioner Industrial Complex” and “A Crowd That Wears Black,” it asks how few choices a well-lived life actually needs—and which constraints make room for attention.
A 2:47 surviving segment of my 2015 talk Computing Taste for the Museum of Modern Art’s R&D salon, “The Way of the Algorithm.” It asks what an Art Turing Test would actually test: the image, the maker, the viewer’s judgment, or the story attached to the work.
A 1:48 filmed statement from New York in 2016, during the ArtAdvisor chapter and before Artsy acquired it in 2017. I speak directly about the work: building machines that understand what people love. The title shown here is editorial; the surviving source file carries no program title.
A recording is not the same thing as a speaking history. These engagements are the strongest selected record recovered from contemporary CVs, calendars, proceedings, and event archives.
Some talks survive only in the notes people carried out. These lines were quoted live and preserved in my Twitter archive.
“The point of personalization is relevance. Time is important. Why waste time on unimportant things?”
“Taste is a game because many times, you’re trying to impress people.”
“AI is a romantic moving target—we’ll call it AI as long as we can’t understand how it works.”
Most of what I host now happens in a room and is not filmed. These are the current formats.