Critical vulnerabilities cannot be resolved by AI

We already have Burp or Qualys that are using AI and efficiently finding vulnerabilities.

But the real question is.

Will AI ever be able to create software without vulnerabilities? Like any other thing "living" in our Universe, perfection is utopia. There is no human, other being or other physical object invulnerable to death or destruction. Software is not an exception.

The most precious vulnerabilities used by APT groups are manually found & exploited by hackers. AI will be blind to this kind of exploits.

Only chance to have them discovered on the defence side will be to have other humans.