Of course I am aware of GPT-3 (it's the queen of BERTology) - anyone doing NLU should be aware of this technology.

Now re the capabilities of GPT-3... it might sound impressive for you, but all GTP-3 does after it is given a prompt is continue with information it has seen next to something like your prompt (I mean it has seen in the gigabytes of text it had memorized)

No understanding of language whatsoever. Many have criticized and you look it up so I will not dwell on it too much here.

Finally, my proofs are scientific - they cannot be nullified just by an 'opinion' - so I did not state an opinion in my article, but a scientific proof that ML/DL is not relevant to NLU. If anyone can argue any of the three proofs, I would welcome that.