"Funny thing is that since I am a programmer who has done a lot of AI, I like and support it.
A lot of people do not realize how AI actually works.
AI actually has nothing remotely like intelligence." R5 #45
I deduce your exposure was early in A.I. inception. Fine. You understand the basics.
If we read between the lines (they're not publishing the code for us to see for ourselves)
one of the important changes to A.I. which in turn results in a change to the threat profile is
A.I. can learn. Some sources call this "recursive", meaning it can experiment, observe the outcome of the experiment,
embrace what worked, and create an alternate for what did not work.
Skip several paragraphs, it might take the worlds top bio-engineers years to create an LD 100 pathogen, a genuine doomsday bio-weapon. BUT !
A.I. may be able to accomplish the same or better in a minuscule fraction of that amount of time.
That is an arms race that benefits the aggressor. CERTAINLY we might A.I. develop a vaccine for such pathogen. BUT ! By then it's too late.
For global conquest:
a) create that precise doomsday pathogen
b) create the vaccine against it
c) vaccinate the chosen surviving population (Aryans?)
d) unleash the pathogen at major seaports and airports around the globe.
The experts clamoring for GOVERNMENT control understand this and also understand, a U.S.-based A.I. organization can refuse to do that BUT !
their refusing to do so does not prevent China, or Russia, or xxx from doing so.
They have ALREADY sounded the alarm.
Idiots like Trump dismiss the warnings as a hoax.