Carl Robichaud

My concern is that

All of the processing of the information and the interpretation of the information could be done by an AI system in a way that leaves humans essentially as button pushers.

Are you really going to...

reject the conclusions of a system that has proved 99% reliable and that's built on state-of-the-art software and hardware.

And it just really seems to be the best way to support your decisions.

And that's, I think, the slippery slope we might go on.

And there are some efforts in Congress to limit that.

I think that

As with other command and control issues, we are only as safe and secure as the weakest link in the chain.

And so we need to be getting together now with Russia, China, other countries to figure out how can we avoid this slippery slope in which we are essentially delegating nuclear decisions to an algorithm, because that's a really scary world.

Yeah.

I mean, I think you've put your finger on it, which is that these digital systems and these human systems are prone to different modes of failure.

And the problem fundamentally is making high-stakes decisions

under incredible time pressure.

That's the fundamental problem, and that's what I think we need to move back from.

We need to devise a system that allows us to be safe and secure without relying on a decision in minutes that could imperil the world.

Because whether you're delegating that decision to machines or to people, there are these failure modes.

And I don't know which is better, right?

I just reject the premise that we need to accept that.

It seems really implausible at this particular moment, given the height of tensions with