Deutsche with the note. I have some sympathy with this view. The problem is sequencing which is a policy challenge.
The release of ChatGPT took the world by storm late last year. Within just five days,= it had already reached 1m subscribers, which vastly outpaced the ten months it took Facebook to reach the same milestone. Suddenly, professions such as lawyers or journalists that had once looked fairly safe from automation faced a competitive threat.
Then in March, the release of GPT-4 turbocharged these fears. The new model underpinning ChatGPT was able to score in the top decile of a simulated bar exam, in contrast to GPT-3.5 that was around the bottom decile. It was even able to analyse images, rather than just text.