State of AI Report 2021

I recommend taking a look at the slide deck on the state of art for AI in 2021, see:

	https://docs.google.com/presentation/d/1bwJDRC777rAf00Drthi9yT2c9b0MabWO5ZlksfvFzx8 <https://docs.google.com/presentation/d/1bwJDRC777rAf00Drthi9yT2c9b0MabWO5ZlksfvFzx8> 

It is lengthy with 188 slides, but gives a good overview across the board.

Self-supervised machine learning using transformers is doing well across many areas, but AI is continuing to be de-democratised with big tech investing in top tier universities only with the rest falling behind. China is leading the world in terms of investment in AI.

I perceive a lack of attention to the limitations of current approaches, especially in relation to human capabilities.  That is fine in respect to work on tasks that humans find hard, e.g. protein folding and the design of new drugs, but is wasteful in respect to replicating human capabilities where there is a lot to learn from studying and replicating human behaviour.

If you are not in a top-ranking AI lab, there are plenty of opportunities for thinking outside of the box of main stream AI.

What do you think about the report?

Dave Raggett <dsr@w3.org> http://www.w3.org/People/Raggett
W3C Data Activity Lead & W3C champion for the Web of things