Jeff Klune

I think that's only a matter of time.

And the sooner we can eliminate that tragedy and that suffering, the better.

So if the cost of eliminating the tremendous tragedy and suffering that's produced by diseases

or a lack of resources, a lack of clean water, a lack of food, is that my job as a scientist is a little less rewarding because I'm not the only thing that can cure a disease.

That's a cost that I think I and humanity and all scientists should be willing to pay.

I do understand we will take some of the...

the value and the pride out of the job because we are not the instrument of change.

But if a disease gets cured 10 years faster because we let AI participate in its cure versus humans do it, I'd rather have the disease cured and all of the tragedy that's caused by that disease eliminated rather than letting a few humans be very, very happy that they were the ones to cure that disease.

Yeah, so quickly on the hype front, for 15 years I have been saying I think AI is going to get much more powerful and capable.

And many other people have said it's all hype and it's about to hit a wall and we're just around the corner from seeing it start to fail.

And I have seen repeatedly that

that AI continues to impress, it continues to get better, and I think that it's very clear that the trend will continue and AI will continue to progress.

And many of my colleagues in the field do not see any end in sight, and we expect AI to get rapidly better.

In terms of bias, there's an opportunity with AI scientists to actually do a much better job

than with human scientists in terms of some dimensions of making science more reliable and more trustworthy.

And a main problem with human scientists is the incentive structure.

So human scientists have to get tenure, they have to get grants, they wanna maybe do work that has high impact and gets on the radio.

And so they tend to only publish findings that have big, exciting, sexy results.

If they do a study and it doesn't work, they tend to say, all right, I'm not going to spend the rest of the year publishing how that didn't work.

I'm going to move on to the next thing that might work.