Thanks for the comments Timothy.
You definitely know what you are talking about, and I need to do more and better homework on ABI. I am not so acquainted with connectomics, so I’ll probably spend some time understanding better how you use artificial connectomes to cut off system energy that is normally devoted to (living) functions, as well as how artificial connectomes can improve our search for an AGI.
I knew about Cortical.io, and recently learnt about PROME (great work by the way), but I did not know anything about Monica Anderson. Thanks for pointing that out, I am keen to know more about her and her work.
I agree with you on the difficulty of understanding how ABI works, and also that we are not adopting the correct approach for developing a general intelligence. Although let me suggest two other factors that should be taken into account:
- The sector is driven (as often) by funding. While scientific research is generally funded with governmental/academic resources, AI is currently carried on by VCs, and most of them have “short” timeframe to capitalize their investments. This puts pressure on AI researchers to commercialize their findings, and this means that something “simpler” that works good enough (and soon) is somehow more appealing than a different approach which requires further efforts and commitments. I am not saying this is good, but this is what I am seeing in the market nowadays (and what scares me off).
- Deep learning might not be the final answer to life, the universe, and everything, but it is also one of the reasons why AI is coming back to the top. We need to move forward, I agree, but I am also quite glad that we had the “deep learning phase”, because I am not sure there would have been so many people today working on AI if it wasn’t for deep learning (and big data and higher computational power of course).