How AI Will Disrupt The Entire World In 3 Years (Prepare Now While Others Panic) | Emad Mostaque PT 1 (Fan Fave)

**SPEAKER_1** (0:01)
Reggie, I just sold my car online.

**SPEAKER_2** (0:03)
Let's go, Grandpa.

**Emad Mostaque** (0:04)
Wait, you did?

**SPEAKER_1** (0:05)
Yep, on Carvana. Just put in the license plate, answered a few questions, got an offer in minutes. Easier than setting up that new digital picture frame.

**Emad Mostaque** (0:13)
You don't say.

**SPEAKER_1** (0:14)
Yeah, they're even picking it up tomorrow. Talk about fast.

**SPEAKER_2** (0:17)
Wow, way to go.

**SPEAKER_1** (0:19)
So, about that picture frame. Forget about it. Until Carvana makes one, I'm not interested.

**SPEAKER_2** (0:24)
Car selling made easy on Carvana.

**Emad Mostaque** (0:28)
Pick up these, may apply.

**SPEAKER_4** (0:30)
Tax Act knows filing taxes can be confusing. So, we have live experts on hand who can help answer any questions you may have. Questions like, can I claim my SUV as my home office if I answer work emails in my car? Tax Act. Let's get them over with.

**Tom Bilyeu** (1:02)
Today's episode is with one of the most brilliant minds in the world of AI, Emad Mostaque. Emad is the founder of Stability AI, the company that brought you Stable Diffusion, an open source AI model that's already being used by millions of people around the world. As such, he is closer to the realities of how disruptive AI is going to be than anyone I have talked with thus far. Emad's on a mission to get AI into the hands of everyone around the world, both to take advantage of its incredible power and, more distressingly, to ensure that those without AI are not effectively enslaved by those with it. That is how powerful AI is. Emad believes that only an even playing field will protect us. This is such an incredible time to be alive and I really am very excited about AI, but I also think every one of us needs to face the coming changes head on. AI, like atomic energy before it, can be used to create or destroy, and it's going to be up to all of us to ensure that we chart our course well. As such, I plan to do many more interviews on the topic of AI in the hopes that both you and I can master a technology that I believe will be more disruptive to our way of life than anything ever in human history. Won't happen overnight, but we've only got one chance to get this right. Harness the power well and not get left behind. As always, if you want an ad-free version of this podcast with a bunch of extras like curated playlists and additional bonus content, you won't find anywhere else. Subscribe now on Apple Podcasts. Now, without further ado, I bring you Emad Mostaque. I'm your host Tom Bilyeu and welcome to Impact Theory.

**Emad Mostaque** (2:44)
How do we make sure it doesn't kill us? Or how does it make sure it doesn't enslave us? Or how does it make sure that it doesn't give us eternal suffering? I realize this could be the real thing that unlocks humanity. AI is not going to replace humans. Humans with AI will replace humans that don't use AI.

**Tom Bilyeu** (3:00)
AI is thrilling, it's very exciting, but there is a non-zero chance that it poses an existential threat to the human race. So over the next three to five years, how disruptive do you think it will be? And what are people not prepared for?

**Emad Mostaque** (3:14)
I think that's an excellent question. So, you know, the future is always hard to predict. And existential is a big word.
Existential means no more humans. So I personally think the AI will be absolutely fine. As a base case, it'll be like that movie Her, if it ever gets this artificial general intelligence. Like humans are kind of boring, goodbye, and thanks for all the GPUs. But you could be wrong, because what we're doing is creating something that's more capable than us in narrow fields. And the question is, does that generalize and then become viral? We've seen an instance of COVID and that expansion. We've seen programs that can explode nuclear reactors like Stuxnet and others. What happens if you start combining these and you get a misalignment? So it's got a strange objective function. Our organizations already are like slow dumb AIs. You know, and like Germans are the most sensible people that we probably know, and yet they committed the Holocaust. And we see this over and over again where organizations chew up people. What if an AI takes over an organization and then decides to do something disruptive or something terminal, such as creating a virus? We don't know about that, but that's at the extreme. When we look at impact, we have the more mundane. The more mundane is what happens to programmers when everyone becomes a programmer, just like photographers. You know, now you can take amazing pictures with your thing. What happens when Google's MedPalm 2 model now can outperform doctors and medical diagnosis, but also empathy, according to the latest paper in Nature. This is a fundamental reworking of information flows. That's going to be massively disruptive and deflationary, even with what we have now with no more advances as it becomes enterprise ready. And we have a continuum from that disruption to the productivity enhances, to potential existential threat. If we keep doing the models as we do now, which is we're not exactly sure how they work or their capabilities, but we keep building anyway.