Jeremy Hsu is a science and technology journalist who writes for publications such as Scientific American, Discover, Wired, IEEE Spectrum and Undark. His recent focus has been on how artificial intelligence techniques such as deep learning could impact society.
Jeremy Hsu
From this contributor
How scientists secure the data driving autism research
Protecting the privacy of autistic people and their families faces new challenges in the era of big data.
How scientists secure the data driving autism research
By
Jeremy Hsu
30 September 2020 | 16 min read
Un ordinateur peut-il diagnostiquer l’autisme?
L’apprentissage automatique (machine learning) présente une possibilité pour aider les cliniciens à repérer l'autisme plus tôt, mais des obstacles techniques et éthiques demeurent.
Why are there so few autism specialists?
A lack of interest, training and pay may limit the supply of specialists best equipped to diagnose and treat children with autism.
Can a computer diagnose autism?
Machine-learning holds the promise to help clinicians spot autism sooner, but technical and ethical obstacles remain.
Explore more from The Transmitter
Why neural foundation models work, and what they might—and might not—teach us about the brain
These models can partly generalize across species, brain regions and tasks, suggesting that a set of machine-learnable rules govern neural population activity. But will we be able to understand them?
Why neural foundation models work, and what they might—and might not—teach us about the brain
These models can partly generalize across species, brain regions and tasks, suggesting that a set of machine-learnable rules govern neural population activity. But will we be able to understand them?
By
Juan Gallego
13 April 2026 | 8 min read
Error equation predicts brain’s ability to generalize
Four statistical measurements of neural network geometry capture how well brains and artificial networks use what they already know to solve new problems, a study suggests.
Error equation predicts brain’s ability to generalize
Four statistical measurements of neural network geometry capture how well brains and artificial networks use what they already know to solve new problems, a study suggests.
By
Natalia Mesa
10 April 2026 | 5 min read
Embrace complexity to improve the translatability of basic neuroscience
Researchers must learn to view heterogeneity as an essential feature of the systems they study and a central consideration in experimental design, not a variable to control for or reduce.
Embrace complexity to improve the translatability of basic neuroscience
Researchers must learn to view heterogeneity as an essential feature of the systems they study and a central consideration in experimental design, not a variable to control for or reduce.
By
Linda Douw, Klaus Eyer, Lara Keuck
9 April 2026 | 5 min read