Computational Neuroscience Notes
Human-Level Intelligence or Animal-Like Abilities? by Adnan Darwiche
Momentum of deep learning has caused a shift in research priorities from other areas of AI (traditional ML, knowledge representation, symbolic reasoning etc.)
Modern AI has been mostly driven by improvements in statistical optimization techniques, data availability and computation. This has created a wealth of commercial applications, yet fail to reach the loftier goals of earlier AI research that try to address a new theory of mind.
There are some tasks (speech, vision) that can be solved without a need for a full formulation of the cognitive task occurring biologically.
Counterpoint to this is that many animals have much higher vision and navigational capabilities than modern AI systems (snakes, eagles, cats etc). Yet none of these animals have the cognitive intelligence that humans have.
“How effective will function-based approaches be when applied to new and broader applications than those already targeted, and particularly ones that mandate more stringent measures of success?”