Biological AI, Mortal Computation, Anthrobots & AGI | Alexander Ororbia
Alexander Ororbia is a computer scientist and neuroscientist.
AI today is built upon silicon-based computers, but what would artificial intelligence look like if it were based on biological materials and principles?
In this podcast, expect to learn how biological intelligence differs from silicon-based intelligence, the concept of "mortal computers" (computers that die), how cognition extends beyond neurons, examples of living, mortal computers such as anthrobots and xenobots, and more...
00:00 Biological vs Silicon; Mortal vs Immortal
05:06 Mortal Computers vs Standard Computers
09:52 The Importance of Death in Intelligence
18:23 Embodiment and Inactivism in Biological Intelligence
26:16 Limits of Current AI
36:35 Benefits of Mortal Computers Over Standard AI
39:50 Mortal Compter Examples: Anthrobots and Xenobots
45:50 How Good are ChatGPT and LLMs? Efficiency of Mortal Computers
50:50 Will LLMs Ever Get to AGI?
Find Alexander's work:
-Mortal Computation paper: https://arxiv.org/abs/2311.09589
-Survey on brain-inspired learning: https://arxiv.org/abs/2312.09257
-Review of neuroscience machine learning: https://arxiv.org/abs/2403.18929
-Survey on predictive coding / biomimetic learning: https://arxiv.org/abs/2308.07870