It will be to do with that philosopher he's obsessed with Girard Rene... I imagine it's the same basic kick lots of the "Effective Altruist" pro-AI crowd are on. It goes something like:

1. I must be good and ultimately do good in the long term.

2. Therefore if I poison the water of 50 million people today to build an AI data center that cures cancer in 50 years time (saving hundreds of millions, possibly billions of people) - then I have done the right thing (by poisoning water today) and I'm good.

3. So I can be as evil as I want, as long as I think my being evil will ultimately help humanity advance.

So, you need an ivory backscratcher and to buy another mansion to calm down? Nothing wrong with that because ultimately it will make you more productive so you'll build data centers faster ...or get the important politicians elected faster... Or sell overpriced rockets to NASA faster (it's okay I'm going to help even more future people solve even more problems with the money I'm stealing from NASA).

So really, "the people being the most evil, are actually the best people."

This is the fucked up double-speak, 2+2=5 "logic" they're using. Where you can destroy this world today because you're really good in a theoretical world. They've created a tech religion, and it's very very evil and destructive.