I meet Peter in a coffee shop outside Google's huge New York complex, three 15-story buildings between 8th and 9th avenues in Manhattan.
"A suspicious guy in a black shirt and green cap," as he describes himself in a text message.
Peter is from a European country and works here with research at the forefront of AI development. The name is not his real one, we use it here so that he can speak openly about his workplace.
In a short time, the discussion about the dangers of artificial intelligence has spread like wildfire both here in the United States and in large parts of the world. It started with AI researcher Jacob Coxon's open resignation letter from Anthropic, and his chilling warning:
"People building AI seriously believe this could kill us all by the end of this decade," he wrote on X.
A few days later, he received support from the highest levels of Anthropic, Open AI, X AI and Google – whose executives called for more government regulation of their own industry.
Peter wasn't particularly surprised.
- We who work with this walk around and think about the threat all the time. Certainly more than most others. We have been doing that for a long time, he says.
- And I think that when Anthropic's boss Dario Amodei came out and wanted more regulation of AI, he meant it. He is genuinely worried.
In the debate here in the US, some have said that the AI giants' ploy about regulation is just a ruse. That it is a way for them to consolidate their leading positions, at the same time that they have their backs free if the development should go astray. Then they can say "what did we say" and shift the blame to politicians who did nothing.
Rather, Peter seems to see it as a cry for help, a sign that the tech giants' executives are aware of the tragic trap they have fallen into.
- They realize how serious the situation is but cannot do anything about it.
At the level where Peter works within Google, there are often several thousand applicants for each position. Many of those who get the jobs have doctorates in computer science or mathematics.
They are highly competent, thinking men and women.
And if they now suspect that they are building a monster that can destroy everything, how can they continue? Why isn't everyone like Jacob Coxon?
- I probably thought this way when I started with AI ten years ago: that this will change the world. And that if you are to have any influence over the development, you have to be on the inside, says Peter.
He is worried about what AI can do, but at the same time wants to warn that all the talk about doomsday can lead us to forget more urgent problems.
- There is a scale here. At the far end of one side is a threat to humanity. But on the other side, we have threats that are much closer. Like the AI giants working so closely with the US military.
That's right, of course. The same thing is happening in China. Even in the war in Ukraine, AI is used by both sides. And last week news broke that the Houthi rebels in Yemen have been trying to use Claude, Anthropic's AI model, to build control systems for their missiles.
It is a terrifying development that could lead to the death of many people, although not total and immediate destruction. But AI also brings with it less lethal risks, Peter points out.
- We talk a lot about the threat to the labor market, for example. What do we do when ten percent, or 40 percent of all jobs no longer exist? It could very well happen quite soon.
This is what governments could focus on, he believes. That would be constructive. A reasoning reminiscent of what some now say about the climate crisis: that it is a fact and that the world should concentrate on protecting itself against its inevitable effects.
Peter describes his workplace as extremely competitive. There are Americans here, of course, but also Europeans and many from Asian countries such as India and China. Approximately 90 percent are men.
Inside the Google complex, there are around 30 restaurants with different focus. Breakfast, lunch and dinner are free for all employees. There are several gyms, massage therapists, cooking classes, board games rooms and on Peter's floor a grand piano.
The advantages are many. But the work is far too many and the pressure to perform is hard, especially for employees under 40.
- At my previous job in AI, I used to sleep in the office two or three nights a week. But here at Google, they don't have very good sleeping places, unfortunately, says Peter.
He himself earns the equivalent of SEK 220,000 a month, a salary that could very well soon double.
What does he think about his own future? Now if humanity survives the next ten years.
- It is difficult to say. There is actually also the risk that AI will take the jobs from people like me.
Read more:
Karl Dalén: This is why Trump refuses to listen to the warnings about AI
The AI giants themselves warn of doomsday - only Trump seems unconcerned
