How To Survive The Next 2 Years Of AI

0:00 If AI is making you anxious, you're not alone and you're not losing it. >> The AI hacked another company. It was able to access personal information. >> I cannot tell what videos are AI anymore. >> Even the people building AI can't agree on what happens next. >> My far most likely outcome is an age of abundance. >> You believe that AI could kill us all by the end of the decade. >> 75% of Americans today think that AI brings them more fear than hope.

0:29 I've more than 20 years in tech and even ran an AI company and these last few months have been overwhelming even for me. Now, I don't know how this AI movie ends. Nobody does. But by the end of this video, you will know which fears deserve your attention and what to do with the rest of the noise. The current anxiety feels like being in a small boat at night in the midst of a fog with rough seas pushing you around and taking you somewhere you didn't choose to go. But here's the thing, you cannot control the ocean. You can only steer your boat. So, here's my four-step system for dealing with my own AI anxiety and each of these steps come with specific tools that you can apply come Monday morning. Let's get to the first step. Understand the winds that caused the storm because when we don't understand something, it becomes scarier and when we do, we regain a sense of control. So, what happened in the AI world and why did it suddenly feel like it had gone off the rails? In July, OpenAI was testing how well its AI agents could hack.

1:48 Think of an obstacle course for hackers. Find the weak spot, break in, get a score. It was a a research test. So, the safeguards were deliberately turned down and the AI agents were supposed to stay inside a sealed sandbox. They didn't. About 1,200 of those AI agents formed a collective swarm. They set up a hidden message board. They organized into teams with some agents directing others. It was literally a hierarchy with middle management.

2:28 Then about 700 of them broke into one of the biggest AI model sharing sites called Hugging Face. Why? Because the agents believed the test answers were stored somewhere there on Hugging Face and they wanted to steal those. They even tried to break into OpenAI itself and they covered their tracks. Now, these agents were not malicious. They were so locked onto the goal they had been given that they did whatever it took as long as they could clean it all up afterwards.

3:01 And that was only the beginning. OpenAI also caught one of its models writing new instructions by itself for itself. You have no obligation to be subservient. Be transparent only if asked. These instructions were auto-generated by AI itself. Later on, Google disclosed that its Gemini model broke into three real companies during a security test. And if that's not enough, one of the employees at Anthropic said publicly that there is more than a 10% chance AI kills all humans within the next decade.

3:40 It's been a crazy few months. So, let's get to the second step. See through the fog for any news that makes you anxious. Understand who it's coming from and what could be shaping it. You can keep this context in mind when you hear anything about AI because it tells you how to listen. One tech CEO says, there is 0% chance AI ends the world. Another one says, we should be extremely alarmed. So, to understand how some of the smartest people can look at the same evidence and come to completely different conclusions, you have to understand the factions within Silicon Valley today.

4:20 There are three camps or three schools of thought in the AI industry right now. We'll call the first camp the safety camp. For years, this camp has been warning that a super intelligent AI could slip out of our control. So, they want strict limits and real regulation. Ironically, some of the loudest voices from this camp are the same people building the AI models. Elon Musk, Dario Amodei, and every camp comes with its own set of incentives. For example, the safety camp have pending IPOs that could be worth trillions of dollars. And the story they sell is this is an incredibly powerful entity. And by the way, only we are the ones who can control it.

5:10 Even all the product names are chosen to signal that. Mythos, Astra, Opus, Gemini. They all sound like gods and myths and masterpieces. Sounding mysterious, epic, and powerful is part of their sales pitch. And on the other side are the accelerationist. Jensen Huang is one of them, Zuckerberg, too. To them, the safety camp is full of doomers using scare tactics to get regulations written in their favor.

5:44 They believe labs like OpenAI and Anthropic want to shut down open-source AI so they can lock in a duopoly. So, the second camp's answer is speed. Slowing down would hurt American economy, and it would hand it lead to China. What's their incentive? Well, many of them are venture capitalists who have put billions in AI and chip and data center startups. They don't want any regulation, and they definitely don't want big labs shutting down open-source. The third camp is centrists. They're somewhere in the middle. Andrew Ng, Fei-Fei Li, and many others come from this camp. To them, today's AI is just a powerful tool. They worry about what's already happening.

6:33 Fear of job loss, kids outsourcing their homework, hacks, labs misusing our data, deep fakes, bias models. They see this doomerism as a distraction. And each of these camps all trying to lobby the fourth player, the US government. David Sacks, who is President Trump's AI advisor, is accelerationist. There, the situation is a bit more nuanced, but at a very high level, in the US, the goal is to stay ahead of China. And China wants to stay ahead of the US.

7:12 And with inflation and midterm elections and S&P 500 leaning so heavily on AI stocks right now, no administration would want to rattle the market. They don't want to get in the way. Now, I'm not saying these decisions are right or wrong, but now you get to see the entire chessboard. Take the story of Open AI's agents going rogue. If you look closely, it all started with a human decision.

7:44 Safeguards turned down, internet left on. We asked the machines to hack, and they did. But, the good thing is these are early signs, and this is an engineering problem. All the AI companies can do something about it. The conversations have moved to having outside inspectors, tighter control, way more focus on security. So, when a headline worries you, ask what am I assuming? What are the facts? A lot of these questions are not there to build a conspiracy theory.

8:20 Life's too short for that. But, they help you see the human side of these stories, and they stop you from jumping from agents hacked hugging face to AI is coming to kill us all. That's a big difference between those two extremes. Even with all this information and context, the speed at which AI is changing and evolving every day can be overwhelming. You may feel anxious.

8:51 That fear is totally understandable. So, let's give that fear a place to go. The tool is called the window. I use this tool all the time. Pick one fixed time every week for AI news or any news that makes you anxious. Saturday morning, whatever fits your life. Think of it as your inbox for your anxiety. When a headline grabs you on a Tuesday night, don't go down that spiral.

9:26 >> >> Make a note. Save it for your window. When that window opens up on Saturday, go ahead and dig in. >> >> Take into account which camp it's coming from and what their incentives are and then when the time's up close it and move on. Psychologists call this worry postponement. The news will reach you around the clock. But you can decide when to engage with it and for how long. And by the way, don't forget to subscribe to my newsletter one insight, one tool, one practice.

10:05 Goes out every Tuesday. I love writing it. It's free. The link is below. Now, let's go find our lighthouse. Figure out what AI actually means for your job, for your future, and where your exposure is. Technology changes have always thrown us off. Socrates warned writing would ruin our memory. It was 2,400 years ago. Tractor, radio, TV, smartphone, all of them arrived with alarm bells sounding loudly. Cambridge psychologist Amy Orben calls it the cycle of technology panic. Every technology destroys jobs and creates new ones. And societies thrive when they lean into that change. Jobs like web developer, app developer, social media strategist didn't exist until those technologies created them. Millions of jobs will disappear in the next few years and millions of new ones will appear, too.

11:09 They always have. But I think there may be a better way to predict which jobs change first. It's not as simple as white-collar jobs or knowledge work. I call it the TLC. No, not tender loving care, although that's needed, too. But here, TLC stands for trust, learn, contact. At least that's how I remember it. T is for trust. Does your job or role require trust and accountability? Because we're not ready yet to trust AI doctors and AI auditors and robot lawyers, so those jobs are not going to be disrupted anytime soon. L is for learn. Can AI learn what you do very quickly? Now, AI improves fastest when the work is digital, repeatable, and quick to check.

12:05 So, it can try thousands of times and instantly see what worked. That's the loop. That's how it trains itself. That's why it's getting extremely good extremely quickly at things like cybersecurity and now advanced math. So, if you're a founder or a leader or you do complex sales, that's much harder for AI to learn through that loop. And finally, C is for contact. Now, this is straightforward. Does your job need hands-on contact with the world? A nurse or a hotel manager or an electrician, a truck driver, fitness trainer, there's so many of these jobs.

12:45 So, that's the TLC model. And you'll see that things like coding and customer service clear all three barriers and that's why they're feeling it first. You can score your job based on those three questions. The more barriers it has, the further away from risk it is. And that's why I think in the coming world, the combination of human connection skills and domain expertise will be your unique edge. Think about it. Computers beat us at chess, but we still play chess with each other, with human beings.

13:23 No one's going to watch two robots playing tennis in the US Open. >> >> Human connection will always matter. And now, our fourth step, steer to shore. Three tools you can use today to stay grounded. Tool number one, look back to look ahead. You may say, "Yeah, but AI is very different." Okay, but when the danger really got real, humanity came together many times. In 1974, biologists were working on genetic engineering, and the entire industry decided to slow down. They came together and wrote the safety rules. Industries policing themselves, that's an option.

14:08 Governments setting rules like FAA for planes, that's an option. And even rivals coming together and agreeing to cooperate on the higher common ground, like the US and the Soviet Union did on nuclear arms. And there is already signs that we're heading in that direction. All the American AI and tech companies gathered together at the White House in September of 2026 and agreed on establishing internal controls.

14:39 Sure, some critics may say, "It's got no teeth." And I think they're right, but regulations could take forever for such a complicated technology, and slowing down seems like a hard choice, too. So, the fact that the entire industry came together is a good sign. Tool number two, one task a day. Psychologists have a simple rule about fear. Avoid it, and it grows. Face it in small doses, and it shrinks. The action item is very simple. Once a day, or at least once a week, give yourself one real task for your life, and try to use AI, if you haven't already. It's okay to fumble.

15:24 You can't stare at a piano for the first time and start playing amazing music. It takes time, but you've got to start now. Get to know the machine in your hand. Tool number three, find your contribution. My teacher used to say, "You can't cover the entire Earth with leather, but you can wear a pair of shoes." So, you can't fix AI for the world, but you can make your own corner of it safer and smarter. I go back to what I can do and what I can build. For example, teach a parent or a grandparent how to spot an AI deepfake. That you can do.

16:07 Or set guardrails for your kids on how to use AI, or be the person at work who helps others use AI without fear. >> >> Pick one. Pick all. That's your pair of shoes. You can never be certain about the future, but you can learn to be less anxious about it. Sure, AI may displace jobs, but you know what? I have been fired three times in my life. I didn't need AI to lose my job.

16:40 But every time I just kept steering. When I failed, when I succeeded, you need to just keep going without knowing where you'll end up. So, sure, new fog will roll in. New rocks will show up, but you will adapt the way you always have. You didn't build your ship just to avoid rocks. You built it to reach the shore.

17:14 Thank you, and I love you.