Breakthrough in AI development using rat brain cells

A biological computing startup, The Biological Computing Company (TBC), has made significant progress in developing AI models using neural patterns from rat brain cells, with Amazon now offering the technology as part of a limited preview. The startup, founded in Baltimore four years ago by neuroscientists Alexander Ksendzovsky and Jon Pomeraniec, claims its AI model is up to five times faster at video generation and significantly reduces inference costs. TBC raised $50 million in funding, including $25 million in March and another $25 million shortly after. The company's focus on video generation stems from the physical layout of its multi-electrode silicon arrays, which makes image processing more efficient. Jeff Dean, a TBC investor, encouraged the startup to fine-tune video models first, as existing benchmarks allow for early scientific validation. TBC's technology involves placing living rat brain cells and human stem cells on silicon arrays to record neural responses, which are then analyzed to improve AI models. Amazon's global director of technology for startups, Deap Ubhi, highlighted TBC's pragmatic approach, emphasizing its alignment with existing generative AI standards. While biological computing offers promising advancements, challenges include maintaining and translating biological data into digital information. As TBC expands its reach through Amazon, the company faces the challenge of maintaining consistency and fidelity in longer video generations, testing its technology's scalability.