Humanity is approaching a critical moment: when artificial intelligence becomes more capable than humans. This could happen soon—within your lifetime, maybe even within the next decade. And right now, we’re not prepared for what happens next.

So, what happens next?

No one can know exactly, but it could be dangerous.

Like, human extinction level dangerous.

Human extinction from AI? This sounds… Really bad. But also really sci-fi. This can’t actually be likely, right?

I get it. This might sound far-fetched, and it’s certainly not lovely to think about. Yet prominent AI researchers, prominent figures at AI companies, and beyond are warning of this concern exactly. And, upon inspection there appears to be genuinely good reasoning for doing so.

So let’s look deeper at this reasoning, and at what exactly experts are saying regarding why AI may surpass humans in intelligence soon, and how this could be dangerous.

The coming years could bring rapid and unpredictable changes. It’s crucial that we prepare, and the more people who understand the situation, the better our chances of navigating things safely.

Why AI may surpass humans in intelligence soon.

Let’s quickly review AI progress to get a feel for how rapidly this technology is accelerating.

In 1997, AI defeats world chess champion Gary Kasparov.

19 years later, in 2016, AI beats the best humans at Go. Go is, say, exponentially more complex than chess. In that context 19 years is not much time, so this event took the AI community, at large, by surprise

Now jump just 7 years to 2023 and we’re in for an even bigger surprise. When ChatGPT is released and AI goes mainstream. The marked an enormous leap as AI shifts from mastering narrow tasks like Go to exhibiting broad, human-like capabilities across diverse domains. The AI community is like “holy crap” as AI, yet again, soars past benchmarks most experts predicted were a lot further away.

By 2023 or shortly thereafter, AI can now:

• Produce code at the level of a 99th percentile computer programmer.
• Craft imagery and videography from simple text prompts.
• Outwit humans in games involving not just strategy, but deception in ordinary language, like Poker and the board game Diplomacy.
• Score in the 130s on IQ tests, as a consistent average.
• Hold real-time, totally human-like conversations.
• Autonomously operate factories, build construction sites, and drive on our roads. This is robotics, of course, and yes, self-driving driving cars are really, finally a thing.