EP 36 - Marc Andreessen and Steven Sinofsky Talk Sentience, Ethics, Job Impacts, and the Future of AI

Every kid is going to grow up now with a friend, right, that's a bot. And that bot is going to be with them their whole lives ... It's going to know everything about them. It's going to be able to answer any question ... As close as a machine can get to loving you, it's gonna love you. That's going to be a thing, and every kid is going to have that.

The entire concept of a video game, I think, is going to turn over completely. The idea that a video game is designed up front is just going to go right out the window. Video games are going to be, basically, created on the fly by AI ... Most books of the future are not going to be a static, single written thing; it's going to be a book that writes itself as you read it.

Steven mentioned this thing of, like, view every edit, basically, box as now an AI prompt. It's going to get very obvious the software that you're using doesn't have this capability, and you're going to kind of wonder whether the developers are asleep, because it's just going to be so clear what the benefits are. And so, I think it's going to hit hard and fast.

People always underestimate the degree to which new technology actually creates new jobs.

In three years, if you're trying to reschedule a flight or book dinner reservations or something, and you're dealing with the future Sydney, and you get a satisfying result from that experience, are you really the next step to debate whether it was sentient or whether it just worked?

Boom, here's ChatGPT ... The chat thing happens, and it's just insane. Like, it's the most exponential thing that we as an industry have seen in the longest time. And that immediately breaks the world into two sets of people: those that can understand exponential and those that sit around and just deny it.

It's actually a lot of the creative jobs that are most directly hit [by AI], I think, the fastest; and then it's the white-collar jobs that are actually less creative that are hit next ... A lot of the blue-collar jobs are actually really hard to hit.

Do we have a method by which we can create something that we know is sentient? Can we create something that's analogous to human brain? No, we can't. We don't have the first clue how to do that. Do we understand how the human brain works? Do we understand how human sentience works? We have no idea. We have no clue.

It doesn't make any sense anymore to have a database of information... about your customers ... Instead, what you're going to do is you're going to have a sales AI, and the sales AI is going to get trained on all customer communication ... And then you're just going to ask it questions.

It may only be the case that new companies can actually do anything interesting in AI, because if you're of any... existing level of size with any sort of existing activist field employee base... It's too much bad press; it's too much investor pressure; it's too much employee pressure; it's too much more pressure. You just can't get from here to there.

Our biggest problem for the next decade is not going to be AI having too much effect; it's going to be AI having too little effect, because there's just these huge swaths of the economy that are just off limits.

There are only two people the history of Time Magazine that have been on the cover in their bare feet: myself and Gandhi.

With a sufficient amount of training data, I think that these bots—already in their current state—are going to be able to surface crimes... that human investigators have not been able to piece together.