🤖 AI Summary
Swiss start-up FinalSpark is building “wetware” bioprocessors by turning donated human skin cells into stem cells, differentiating them into neurons, and assembling millimetre-scale brain organoids (~10,000 neurons) that are hooked to electrodes for stimulation and recording. Researchers use small electrical pulses to probe these organoids and read spikes of activity as rudimentary computation; examples include a lab robot using an organoid to distinguish braille letters. FinalSpark and university partners stream neural activity from their incubators and argue biological neurons could one day replace silicon for some tasks because they’re far more energy efficient — the company cites neurons as roughly a million times more efficient than artificial neurons — and can be reproduced in vitro.
The work is significant because it reframes computation as something you can “grow” rather than simulate, potentially addressing AI’s soaring energy costs and chip shortages. But major technical and ethical hurdles remain: scaling from thousands to billions of neurons, reliably encoding inputs and decoding outputs, short organoid lifespans (up to ~six months), unpredictable behaviors (researchers still don’t fully explain certain activity spikes), and persistent questions about sentience and consent. Today wetware is closer to biomedical research than to practical AI infrastructure, but it could both illuminate brain function and reshape future hardware over decades.
Loading comments...
login to comment
loading comments...
no comments yet