An Openai researcher has resigned from his situation after working for four years in the company. The reason for that? Rapid development of artificial intelligence. In a post on X, Steven Adler announced that four years after working on security in Openai, he quit his job in mid -November.
“It was a wild ride with a lot of chapters – dangerous ability evals, agent security/control, AGI and online recognition, etc. – and I will miss many parts of it,” Mr. Adler wrote.
In the next part of the post, he talked about whether he was the most “frightened”. “Honestly, I am very nervous at the pace of development of AI these days. When I wonder where I will increase the future family, or how much to save for retirement, I can’t help, but won’t be surprised: what: Humanity will also make it at that point?
In follow-up posts, Steven mentioned that “AGI race is a very risky gamble” because no laboratory has a solution to its alignment and even if anyone wants to develop technology responsibly If, then the competition among colleagues pushes to give speed to all.
“As is further, I am enjoying a little break, but I am curious: what you see as the most important and neglected ideas in AI Safety/Policy? I am particularly excited: Ways of Control, Schemeing detection, and security matters, ”Mr. Adler concluded.
Some personal news: working on security after four years @OPENAII went in mid -November. It was a wild ride with a lot of chapters – dangerous ability evals, agent security/control, AGI and online identification, etc. – and I will miss many parts of it.
– Steven Adler (@SJGADLER) 27 January, 2025
Not long ago, Jeffri Hinton, often called the “Godfather” of AI, expressed concern that technology could cause human extinction in the next 30 years.
The British-Canadian Computer Scientist, in 2024, was awarded the Nobel Prize in Physics, for its work in the field, “10% to 20%” was the chance that AI could be the result of human extinction in the next three decades. He earlier predicted that he had 10% likely.
Mr. Hinton, a professor at the University of Toronto, also said that human was children as compared to advanced AI systems. “I like to think: Imagine yourself and a three -year -old child. We will be three years old,” he said.
