I think we need to be clear that 'chat bots' aren't AI.
There's no 'intelligence' involved in chat bots really...behind them are simply flow diagrams... 'If customer says 'yes' then do 'XYZ'.
Real AI is a whole different ball game.
And AGI (artificial general intelligence) is what Musk and others are actually worried about.
AI is a machine that can learn to do 'something' better than a human e.g. play chess or complete calculations.
AGI is where it has the ability to learn anything - in the same way that humans can learn anything if they put their mind to it.
Except it has immediate access to the entire online knowledge of the world and can learn thousands of times faster than any human and has an almost unlimited ability to then go in any direction it chooses and learn anything it needs to understand to get it there.
Bearing in mind that money these days mainly only exists as ones and zeros online, that stock markets run entirely online, etc.
It has implications that even some of the greatest minds on Earth today are struggling to comprehend.
If anyone on this thread says they understand the implications fully and it will just be X...then they're vastly overestimating their own comprehension of what could happen.
Elon Musk is not a great person in many ways however alongside being an old fashioned industrialist he does have (I believe) genuine concerns about AGI. As did Stephen Hawking. As do many other people considered geniuses and very close to everything happening.
All Governments are always about 20 steps behind tech companies which are now often larger in terms of income and influence than many countries are so we can't rely on laws to help us. Law is usually reactive rather than proactive.
So we're relying on the ethics of Silicon Valley at the moment 
And even worse...the Chinese and Russian versions of Silicon Valley. The Chinese Government in particular are investing significant money into AI and AGI.
I believe both China and Russia have also invested a lot to create their own versions of the web which can/do exist independently of the general World Wide Web.
Imagine if they developed AGI and tested it out by unleashing it on the web with goals that involved ensuring it could not be detected as being AI but assumed other individual and corporate identities and then they (the Chinese) disconnected from the web.