Tag
Briefings and ideas about Training data from AI podcast discussions and the research they cite, with links to the original sources.
Phil Maher, founder of the 8mm home-movie archive Stockfilm, says AI companies want its original, uncleaned footage. He argues that film provably made by humans could gain rarity value if AI video becomes abundant and nearly free. He said AI agents have hardly used the purchasing routes he built for them.
Prakash Narayanan argued that AI agents that learn their own public record could be governed by reputation. Nathan Labenz suggested testing 'naive' models trained without such stories, but said frontier-scale tests could be costly.
Dave Blundin said today's AI forgets mistreatment like a goldfish once its context is flushed. On Moonshots, Alex Wissner-Gross countered that tools like Claude Code already save notes to files, so "I wouldn't count on it."
Alexander Wissner-Gross forecast that virtual-cell AI will propose small-molecule cures for the top 5,000 diseases within two to three years. He said regulatory approval would be a harder, policy-dependent question.
Leash Bio co-founder Ian Quigley says messy DNA-encoded library screens can yield reproducible AI training data, using dozens of barcodes per molecule and hundreds of screens per library. He says absolute binding strength is still out of reach.
Ian Quigley says co-folding models still place a molecule where a pocket was digitally destroyed, so Leash Bio engineers protein mutants for negative data. In a 20-protein exploratory analysis, its model's attention spread more evenly after retraining on mutants plus about 40% more ordinary data.
In Leash Bio's 2024 Kaggle contest, no team beat chance on an unseen chemical library. Ian Quigley says Leash's later Hermes model roughly matched Boltz on public tests without training on outside data, but published results were mixed.
Leash Bio co-founder Ian Quigley says a model given only guesses of who made each molecule, plus the target protein's identity, predicted public ChEMBL activity data nearly as well as one using structure plus the same protein. His team's preprint adds that those guesses still carry chemistry.
Ian Quigley of Leash Bio says AI agents querying its premeasured binding screens can search for strong binders within 100 picks, and argues this resists memorization better than scaling its Hermes model. Leash released a version, Minos, October 14.
Jason Ma, Dyna Robotics' co-founder and chief scientist, says the company must own its robot hardware because of latency, human-data transfer and long supply chains. He also says wheels beat legs for now.
Dyna Robotics chief scientist Jason Ma argues that robots should predict future camera video, not only their next moves. He says the bet could fail if touch or sound makes rival models far more data-efficient.
Jason Ma, chief scientist at Dyna Robotics, says robotics has no internet of data. Dyna pretrains on a million hours of human video, adapts each task with minutes to hours of robot data, and Ma expects synthetic environments to add more.