Yes, I think we certainly need to deepen our knowledge of how our memory works. Our current amount of investments in the topic is pathetic. While I'm not going to speculate why that's the case, suffic to say it's much less than that in the 60s and 70s, as measured by results produced. Besides the work by Mozer and colleagues described in this essay, Hopfield Networks is another rare time-tested insight that we got. However, it's still unknown how it can be complemented with the rest of AI, despite it being discovered 50 years ago.
There are three serious gaps in our current scientific understanding of AI:
Due to these basic deficiencies in our science, the probability of it turning into applied engineering in any near future is nil. Besides, history provides ample lessons on the follies of current popular ways of estimating time to AGI. Last but not least, a down to earth review of our societal and economic environment cautions excessive exuberance embraced by the AI community.
Working memory here refers to the ability of recalling past... (read 2546 more words →)
The issue is not that there are missing pieces, it's that there are critical missing pieces and there's lack of awareness of them. With respect to Human Intelligence, it's fine for artificial neural networks to miss the spiking behaviour in our brains, but it's dismal that they cannot ever remember anything due to lack of working memory.