But the interview itself is interesting. I made a summary, if you are interested, read it. Especially on mathematics and taxonomics. 00:00:12 - PHP Week (PHP Vulnerabilities Week) Sania talks about how AI models discovered critical PHP vulnerabilities that have existed for many years. The problem, in his view, is not that artificial intelligence has suddenly learned to look for errors, but that no one has previously applied such capabilities systematically to all available code. He expects to discover more vulnerabilities. Fixing individual bugs may not be enough and will require rethinking the architecture of operating systems, applications, networks and routers. Thanks to artificial intelligence, rewriting entire software stacks, once considered almost impossible, becomes a realistic development possibility. OpenAI separately described the discovery of PHP vulnerabilities in its star patch initiative. 00:02:37 - Early days of the internet (early internet) Sania compares current AI with the advent of the mass internet. He sees the main similarity in creative freedom: children are now able to create full 3D games in about ten minutes, whereas in the past even a simple program required significant effort. The new generation will see the creation of software on demand as a natural feature of the computer. He cites a case in point: One day, OpenAI experienced a major codex glitch, and employees suddenly felt how accustomed they were to programming with AI. In his view, traditional manual programming is gradually becoming a historical skill, like working with punch cards or assembly language. 00:07:33 - Cyber ​​Blue (AI for Cyber ​​Defense) Collison asks if it was a mistake to intentionally train AI to attack cyber attacks. Sania admits that the industry has not sufficiently explained to the public why the potential benefits of AI justify the risks involved.

But the interview itself is interesting. I made a summary, if you are interested, read it. Especially on mathematics and taxonomics. 00:00:12 - PHP Week (PHP Vulnerabilities Week) Sania talks about how AI models discovered critical PHP vulnerabilities that have existed for many years. The problem, in his view, is not that artificial intelligence has suddenly learned to look for errors, but that no one has previously applied such capabilities systematically to all available code. He expects to discover more vulnerabilities. Fixing individual bugs may not be enough and will require rethinking the architecture of operating systems, applications, networks and routers. Thanks to artificial intelligence, rewriting entire software stacks, once considered almost impossible, becomes a realistic development possibility. OpenAI separately described the discovery of PHP vulnerabilities in its star patch initiative. 00:02:37 - Early days of the internet (early internet) Sania compares current AI with the advent of the mass internet. He sees the main similarity in creative freedom: children are now able to create full 3D games in about ten minutes, whereas in the past even a simple program required significant effort. The new generation will see the creation of software on demand as a natural feature of the computer. He cites a case in point: One day, OpenAI experienced a major codex glitch, and employees suddenly felt how accustomed they were to programming with AI. In his view, traditional manual programming is gradually becoming a historical skill, like working with punch cards or assembly language. 00:07:33 - Cyber ​​Blue (AI for Cyber ​​Defense) Collison asks if it was a mistake to intentionally train AI to attack cyber attacks. Sania admits that the industry has not sufficiently explained to the public why the potential benefits of AI justify the risks involved.