@NeelNanda5 No, that's what you like to say out loud. Interpretability is dual use. You know it. Your employer's goal in relation to AI are evil. You know it. Resign and repent.
You don't need to be an expert to get that AI could kill everyone, but you also don't need to be dumb to NOT get it--to not get it, you mostly just need to never think deeply about it
Allowing AI companies to take creative work without “consent and compensation” would be theft, plain and simple. The PM’s repeated promises about “consent and compensation” - including this one yesterday - must now be upheld genuinely by his government’s actions.
Top of the Democratic agenda on day 1 needs to be a pause on advanced AI development, regulation of this out of control industry, and a ban on artificial superintelligence.
"The only way to defend was to use AI"
No, the main way to defend was to bump up dependency versions.
If you need AI for that, you should not be around cybersecurity, not commenting on it.
The last week in cyber seems like a taste of the future:
• Everything sped up 10x, attackers included
• The only way to defend was to use AI
• Humans could barely follow what was happening