The speaker discusses his disagreement with Jeff's emphasis on biological plausibility in learning algorithms and argues that it shouldn't be a limiting factor in their development.