I met with China’s top AI experts.
They are also panicking.
Last month I attended a major AI conference
Located in Beijing’s bustling high-tech district.
A fascinating demonstration was held
The latest humanoid robot
Session on self-improvement model
Tweak your own code and progress endlessly.
But I learned one important lesson.
The US and China should engage in fierce AI competition
Be on one side and work together.
Let me explain why.
Cybersecurity and related systemic risks
It is too important to ignore when it comes to frontier AI.
The US primarily views China’s AI advances as a threat
both for economic prospects and national security.
impose strict limits on tips
and chip manufacturing equipment.
The US recently ordered Anthropic
to prevent foreigners
Bans the use of Mythos and Fable 5, which are the most powerful models.
As reported by WIRED, the South Korean telecommunications giant
There was particular concern over suspected ties to China.
In response, Anthropic revoked everyone’s access.
The conference I attended reinforced that idea.
Both America and China
And potentially everyone stands to lose.
If AI develops too quickly and recklessly.
As the world’s two largest AI powers,
They are responsible for cutting edge models
and cooperation can help reduce associated risks.
Powerful agent system to carry out cyber attacks
Or you may fail catastrophically.
An expert I spoke to after the conference
I compared the current situation
The way America and the Soviet Union did it
They were ultimately forced to cooperate on nuclear safety.
In that case, the benefits of collaboration are
Ultimately, the national security risks outweighed the risks.
He told me, there’s only one thing that almost everyone in AI can say.
I can agree now
AI doesn’t need an event like Chernobyl.
If you want more stuff like this, please subscribe
Visit my newsletter AI Lab at wired.com/newsletters.
