A recent work on a spike-based LLM demonstrates the feasibility and desirability of applying biological principles to address the current ‘deathly embrace’ between AI algorithms and GPU-based compute platforms.
Steve Furber (Thu,) studied this question.