People make a big mistake about AI thinking the key point of no return is when AI is smarter than the smartest human.
First, if you hire an employee are you looking for someone who is super smart or are you looking for someone who can follow directions? There are probably a handful of people who are hired because they are "super smart" but for 90-95% of people hired the real question is can you follow directions. AI can follow directions very well.
Second, let's think of a person who is super smart, perhaps a lawyer, or Columbo, or Sheldon Cooper. That lawyer who will get the guy off who everyone assumes is guilty, he has a team of people working with him. He has the ideas, but they go and do the research he asks for, gets him the things he needs. He might have a team of ten people working with him when you include private investigators, para legals, and junior attorneys. AI can replace 50-80% of those people.
Third, the improvement in the last ten years has been at break neck speed. Ten years ago, in 2015 AI was in an infant stage. It could play chess, a game with very clear rules. In 2002 we had the first Roomba vacuum cleaner and it was clear, it wasn't going to put any cleaning lady out of business. In 2011 Watson won at Jeopardy and for the first time it became clear, AI would be able to replace many workers, not yet, but soon. In 2017 AI taught itself to beat the world's number 1 Go player, it created new strategies never before seen. That was the point at which we realized AI could train itself to be smarter than humans. Again, Go is a game with very clear rules defined, but it was a signpost that AI was not hype, then in 2020 Alphafold was released solving real world problems that had stumped humans for fifty years. This was a key transition, showing that AI could have super intelligence and that it could be used to solve very significant problems.
At the same time GPT-3 was released and has been used by kids in school to do their school work. Since we have had readily available free versions of AI on the internet millions of people have used them and they have improved dramatically. At first they were likened to a very smart elementary school kid, then a very smart HS kid, then a very smart college kid, and now they are testing at the very top scores for graduate students and Phd's. Depending on your definition in many ways AI has achieved Artificial General Intelligence meaning they have a firm grasp on all the various disciplines in education. Although we still have humans that can outperform the AI on individual tests it is unlikely there is any human that could outperform AI on all the tests and what is much more likely is that given ten tests AI will outperform every single human on at least six. This improvement has taken place in the last five or six years. You could argue this was "born" at the time of Watson in 2011, but that makes AI 14 years old. Also we know that AI can improve with more powerful computers, with more data, with better chips and with better software. Every single year we are seeing improvements in these four areas. The pace of improvement is astounding. If you are starting college this year where will AI be in four years when you graduate? Far ahead of you, that is for sure. If you decide to stay in college for two more years to get a Phd, then where will AI be six years from now? At this point if you are not scoring in the top 1% of humans is it worth it to go to college for the next six years? Obviously it depends on several factors, is the cost of education a factor to you? If not and if your father owns the business and if you are scoring in the top 1% of kids it might be.