EP 37: Neurons: Future of AI Processing

Summary

This Data Science With Sam episode has Sam interview Lena Curtis of FinalSpark about living human neurons as a possible AI processing substrate. Curtis frames biocomputing as a complement to silicon rather than a replacement: neurons are not a power source, but an alternative processor that may eventually reduce energy and cost pressure for suitable workloads. The discussion is careful about maturity: researchers can stimulate and read neural cultures through electrodes, but Living Neuron Computing still depends on understanding learning, encoding, and control in vitro.

Key Claims

• FinalSpark grows human neurons on electrode arrays and rents access to a browser-based neuro platform for external researchers.
• Lena Curtis says neurons should be understood as processors rather than biological batteries.
• The strongest motivation is energy efficiency: neurons are presented as much more energy-efficient than silicon systems for some future computation.
• The main scientific bottleneck is not basic signal access, but understanding how neural cultures encode information and learn when they are outside a brain.
• The platform converts analog neural spikes into digital signals for analysis, and digital instructions back into analog electrical stimulation.
• Curtis expects biological computing to coexist with silicon, quantum, and other substrates because neurons are slow and will not suit every workload.
• Biocomputing Ethics matters because living neurons in machines blur public, regulatory, and philosophical boundaries between human tissue, tool, and computer.

Key Quotes

“processors, not power sources” - Curtis’s correction of the host’s initial framing.

“stored one bit of information” - the episode’s marker of how early the field remains.

Connections

• Data Science With Sam, Sam, Lena Curtis, and FinalSpark - show, host, guest, and company context.
• Fred Jordan and Martin Kutter - FinalSpark founders named in the episode.
• Biocomputing AI Hardware, Living Neuron Computing, Neuro Platform Remote Access, and Biological Processor Energy Efficiency - core technical and infrastructure concepts.
• Biocomputing Ethics, AI Energy Bottleneck, AI Compute Continuity, and Generative AI Use-Case Triage - broader AI-cost, AI-energy, and adoption context.
• Neuroplasticity / 神经可塑性, Population Coding, and Computational Functionalism - adjacent neuroscience and computation concepts.

Contradictions

• No direct contradiction found.
• The source qualifies AI-hardware optimism by emphasizing that FinalSpark is still doing fundamental research, that generative AI is an assumed future application rather than a proven deployment, and that cost and profit claims remain forward-looking.