Bernie Sanders on the Existential Risk of AI

Below is the text of an email I shared with some of my friends and family recently, sharing my thoughts on the dangerous situation humanity is in.

Hi friends,

Senator Bernie Sanders held a panel with Chinese and American experts on the existential threat from advanced artificial intelligence earlier this week in order to discuss the need for international cooperation to slow and pause development. You can watch the recording here: https://www.youtube.com/watch?v=wjBfS3AEk2c

I am moved by his willingness, almost unique among politicians, to speak clearly to the public about the risk from AI. I wanted to share a few of my thoughts about the situation with you.

I know that some of you may be uncertain that the danger that worries me, that of rogue superintelligences causing human extinction, is worthy of our attention. The danger is not posed by AI that exists today, but by the far more capable systems that labs are racing to build. The risk may seem to you to be too hypothetical, too outlandish, to merit action, especially when the world faces many other pressing issues which are clearer to see right now.

As for whether the risk is real, I believe that you should have uncertainty unless you have devoted time to carefully understanding it for yourself. But I hope that either your trust in my own judgment, given the many hours I have spent carefully and critically considering our situation, or better, the judgment of nearly every leading expert in the field of artificial intelligence, including Nobel laureate Geoffrey Hinton and Turing Award winner Yoshua Bengio, will be enough to convince you that this is a risk worth taking seriously.

And the stakes of artificial intelligence are far beyond the stakes of any other issue humanity is facing. Neither nuclear disaster nor runaway climate change, while both would cause extreme suffering to billions, would be likely to extinguish the human species altogether. The universe is young. It is less than 0.02% through the age of stars. If humanity cannot navigate through the technological transformation of the coming decades with maturity, we may lose the potential for life to prosper for a hundred trillion years to come.

What does speaking clearly look like? Bernie does not shy away from stating the extremity of the risk: he uses phrases like possibility of extinction, existential threat, and cataclysmic consequences. This is the only language which honestly represents the situation we find ourselves in. He does not advocate mitigations which merely appear to address this risk, neither investigating lesser dangers like misinformation and mental health, nor asking AI companies to voluntarily disclose their plans for proceeding with caution. One current bill, backed by OpenAI, would grant AI companies legal immunity from catastrophe-related lawsuits in exchange for such cooperation.

Instead, he recognizes that the only sane action is to stop working to build dangerously superhuman machine intelligence, he recognizes that unilateral action from the US will not be enough, and he independently takes steps towards cooperation with China, the other major power in artificial intelligence.

There is a massive amount of work that must be done to shift the Overton window to make a pause in development of advanced AI systems seem like a genuine option available to humanity. I am grateful that he is doing this work.

Finally, if any of this raises questions for you, or if you find yourself wanting to push back, just respond to this email or give me a phone call. I would love to hear from you.

Sending you all my best,
Satya