Andrew Ng – What is the most important problem the AI community should work on?

> Explainable and ethical AI

This IS the most important problem in AI and is a precursor to "Healthcare including Covid-19" and "Combating misinformation". It is also a precursor into achieving Artificial General Intelligence (AGI).

However, Most of what we see about AI is the hype and capabilities it will bring and glossing over the actual issues; especially with GPT-3 which hasn't been extensively scrutinised yet by the experts. It's still a black box which doesn't explain why it has generated the text based on its input. I can definately see GPT-3 being applied into legal text, but only if it can improve its explainability.

Until then, we must keep writing detectors for these models.