0:03 Most people are letting AI destroy their
0:05 ability to think, training AI to become
0:07 their own replacement. Tragic, because
0:10 AI can make you dangerously intelligent.
0:13 I went from being homeless to an MIT
0:15 grad and running and advising AI
0:17 companies worth billions. And here's
0:20 what I've learned. The top 1% use AI
0:22 backwards. They don't prompt to get
0:24 answers. They use it to train their
0:26 brain and outsmart [music] almost any
0:28 situation. So, in this video, I'll break
0:31 down a counterintuitive system the top
0:35 1% use to get smarter faster with AI.
0:37 Here is the four-step framework. Step
0:41 one, intelligent laziness. A study in
0:43 Harvard Business Review found that CEOs
0:46 waste 72% of their time in meetings that
0:48 don't move the needle. [music] We've all
0:50 experienced those meetings, haven't we?
0:53 The 1-hour meeting that needed only 15
0:55 minutes to get to a decision, but it's
0:58 hard to stop. So why do some of the most
1:00 accomplished folks feel [music] trapped
1:03 this way? Because we all suffer from
1:06 this biological glitch called completion
1:09 bias. Your brain is wired to seek an
1:12 immediate dopamine hit that you get from
1:14 finishing a task. So we end up treating
1:17 all tasks as equal because we're going
1:19 to get roughly the same amount of
1:21 dopamine when you spend time on
1:24 reddrafting an internal email or a
1:25 million-doll strategy document.
1:28 Everything is priority one. So none of
1:32 it is. So how do you avoid this priority
1:34 blindness? A good way to think about
1:37 tasks is to see two curves. [music]
1:41 First curve has capped payoffs. This
1:43 curve goes up and then flattens out once
1:46 it reaches the zone of diminishing
1:48 returns. So tasks like formatting slides
1:51 or [music] internal emails, expense
1:55 reports, FYI meetings. What happens if
1:57 you spend additional effort to make the
2:00 outcome of these tasks pitch perfect?
2:02 Nothing. There's no upside here because
2:04 the value flat [music] lines after a
2:07 point. Nobody cares if you spend hours
2:10 choosing better fonts or breathtaking
2:12 designs in internal slides that are seen
2:16 for 6 minutes. This curve shows you your
2:19 zone of intelligent laziness. There was
2:21 a Nobel Prize-winning economist and
2:22 computer scientist and his name was
2:25 Herbert Simon and he came up with a
2:28 concept called satisficing which pretty
2:31 much means stop when it's good enough.
2:34 Satisfy and suffice. [music]
2:36 Satisfice. Now our second curve is the
2:39 exact opposite. It has uncapped payoff.
2:42 This curve stays flat for a long time
2:45 but then goes to the moon in a hurry.
2:48 These are tasks like customer
2:50 interactions, product design, pricing
2:54 model, finding a co-founder or a life
2:57 partner. Being 1% better here does not
2:59 yield 1% better result. It actually
3:02 solves the rest of the 99% of your
3:04 problems. Pour your soul into this.
3:07 Johnny IV would obsess for many months
3:10 on even the internal component design of
3:13 iPhone. But, you know, Steve Jobs never
3:16 said, "Hey, this is costing us a lot of
3:19 money." And who's going to pry open the
3:22 iPhone? But Steve knew this was the
3:24 second curve. [music] So, if the first
3:26 curve is your zone of laziness, your
3:29 second curve is your zone of obsession.
3:32 Let's talk about how AI can help. The
3:35 top 1% use AI on zone one or the zone of
3:38 laziness. The more they outsource zone
3:41 one to AI, the more they can focus
3:43 [music] on zone 2, the zone of
3:45 obsession. So how do I decide what to
3:48 outsource to AI and when? So for that I
3:51 use a very simple framework called drag
3:55 framework. D R A G. Four categories of
3:57 work you [music] immediately should
4:00 delegate to AI so you can stay in your
4:03 zone of obsession. First D equals
4:06 [music] drafting. This is the blank page
4:09 problem we all face. It's hardest to
4:12 [music] get from zero to one. Sometimes
4:14 AI can help here tremendously. Actually
4:16 [music] give it a prompt using the AIM
4:18 protocol that I have shared before. Hey
4:22 AI, act [music] in this role. Use this
4:25 input and this is your mission. A IM.
4:27 And that way you get started very
4:30 quickly on that email or code or
4:33 presentation and the first draft from AI
4:36 will be crappy and atrocious, but that's
4:39 fine. Now [music] you have a starting
4:41 point. You're not staring at a blank
4:43 page anymore. Now it'll trigger
4:46 something in your brain and you're off
4:49 to the races. R equals research. This
4:52 helps you solve the information overload
4:54 problem. Today, if something requires
4:56 deep research, [music] it can be
4:58 dramatically accelerated using AI,
5:01 summarization, extraction, competitive
5:04 intel, you know, don't spend time doing
5:06 that kind of research. Let your friendly
5:08 neighborhood AI do it [music] for you.
5:11 When you use the deep research feature
5:14 on Chad GBT or Gemini or Claude, it
5:16 fires off hundreds of secondary search
5:19 queries. It goes out to the web like a
5:22 spider and finds hundreds of sites,
5:24 consolidates the results, even checks
5:26 his own work by asking what's missing
5:29 [music] and follows up on its own to
5:32 finally deliver a rich document to you.
5:34 It's like you just hired a consultant
5:36 for a week-long research project, but
5:39 instead you get there in 10 minutes.
5:43 Third is a for analysis. Let AI take the
5:46 first pass at analyzing, summarizing,
5:49 reasoning, especially if it's all
5:52 unstructured data because AI is going to
5:54 find patterns that we humans aren't
5:56 going to be able to. So use it for your
5:59 advantage. And finally, G is for all the
6:02 grunt work. Tasks like reformatting,
6:05 translating, tabulating, cleaning data,
6:09 and on and on the boring manual work.
6:11 Just give it to AI. So what's the key
6:15 principle behind drag? Apply it only
6:17 when you are in your zone one. That
6:20 first curve. If it requires human
6:23 interaction or judgment or intuition or
6:27 decision-m or tastes, that's curve two.
6:29 That you've got to do it yourself. But
6:33 you know, I found that 70 or 80% of my
6:36 repetitive tasks tend to be in zone one.
6:39 And you might find that too. So be lazy
6:42 when you can use drag. Be obsessed for
6:44 everything else. Step two, the
6:48 intelligent hill. For 300 years, Isaac
6:51 Newton convinced us that universe was a
6:54 clockwork machine, predictable and
6:57 certain. But in 1927, another [music]
7:00 scientist named Heisenberg shattered
7:03 those classical beliefs. He showed that
7:06 our universe exists only as a cloud of
7:09 possibilities. at quantum level. It was
7:11 a profound shift. [music] You and I have
7:14 to make a similar shift when we use AI
7:16 nowadays. The first trick is to stop
7:19 treating AI like [music] a calculator.
7:21 We like to live in a world with clear
7:25 rules. You type 2 + 2 into a calculator
7:27 and you [music] get four always. It's
7:30 predictable. But AI is not a calculator.
7:32 It's [music] a probability engine. If
7:34 you ask the same question to AI again,
7:35 it'll give you a completely different
7:37 answer. It'll happily make things up for
7:40 you unless you ask it to verify. AI is
7:42 brilliant on some days, confused on
7:45 others, but on any given day, it refuses
7:47 to admit that it doesn't know the
7:49 answer. It loves to make things up. So,
7:52 you don't just ask AI the way you ask a
7:54 normal human being. You have to
7:56 architect your questions very carefully.
7:58 Now, most people use a tactic called
8:01 zeroshot prompting. So, for example,
8:03 they would ask, "Give me the best new
8:06 business idea." And of course, AI will
8:08 dish out a response and tell you why
8:11 it's the greatest idea in the world, but
8:13 you're literally rolling the dice and
8:15 looking to win. To get elite results,
8:18 though, you must climb the intelligent
8:21 hill. There are four camps on the way.
8:24 Each camp will show you a different way
8:26 to work with AI. Our first camp is
8:29 called oneshot prompting. When you
8:32 prompt, give [music] one clear example
8:34 so the model doesn't guess blindly. So
8:37 the prompt would look like, write a
8:39 LinkedIn post about remote work. Use
8:42 this specific post as a style guide.
8:44 [music] And so give it a post. Give it
8:47 an example and paste that post in the
8:50 prompt as a reference. And that simple
8:53 act is already an upgrade than rolling
8:56 the dice blindly. Second camp, few shot
9:00 prompting. Now here you give AI three or
9:03 more examples so it can find patterns of
9:06 style and substance and tone that you
9:10 desire. Attach documents, links, data or
9:13 your prior work. This is called
9:15 grounding the model. So basically it
9:18 stops fantasizing and hallucinating and
9:21 gets grounded to reality. Here's an
9:24 example of a prompt. Here are the five
9:27 of my previous presentations. And now
9:29 write a new presentation based on my
9:33 tone of voice on topic XYZ. And here's a
9:36 pro tip. Ask the AI to explain the
9:39 pattern back to you first. That way AI
9:41 is forced to articulate what it's doing.
9:43 And more importantly, you're forced to
9:45 learn how your brain works. How did it
9:47 come up with those patterns? [music] Now
9:50 you're being smart about being smart.
9:52 Now let's move to the third camp. This
9:54 one is called chain of thought
9:56 reasoning. Again, fancy name, but the
9:59 idea is simple. Ask the model to think
10:01 long and hard before it responds. Your
10:05 job is to slow AI down and force
10:08 explicit clarity by asking it to show
10:10 its work. That's all there is. This is
10:11 also a good way to reduce
10:13 hallucinations, of course. So, let's say
10:16 you're working on some report and so you
10:18 attach it and write a prompt that could
10:20 look like this. Do not refine my
10:22 research report yet. List the top three
10:25 most impactful areas of improvement
10:28 after we analyze it. Tell me why you
10:29 think so and suggest how we address
10:33 each. Think step by step. Show me your
10:35 thinking for each step. That last line
10:37 is the most important one. And our
10:40 fourth and final camp is agents.
10:43 According to Salesforce, AI agents help
10:47 drive $67 billion in global sales during
10:48 Cyber Week alone. [music]
10:52 So agents are already here. The best way
10:54 to think about agents is to think about
10:57 who you would hire for a task. So let's
10:59 say if you wanted to hire a researcher,
11:01 an analyst, [music] and a copywriter,
11:04 you can do that with a single agentic
11:07 prompt that looks like this. Do deep
11:11 research on trends on topic XYZ. Analyze
11:13 and cross-reference all the trends to
11:16 find the three most important ones and
11:19 draft a one-page memo summarizing the
11:22 findings. Now, what is actionable? Try
11:24 this framework tonight. Open your
11:27 favorite AI app and take any prompt that
11:29 you were about to use. Just try to get
11:32 to the next camp. That's how you start
11:33 climbing up the intelligent hill.
11:36 Remember when you were dealing with a
11:39 drunk genius? make sure you were the one
11:42 driving the car. So now at this point,
11:45 everything we've done has made you fast
11:47 and efficient. You're delegating better,
11:50 you're prompting smarter, you're moving
11:52 up the hill, and there is less friction
11:55 than before. And that's exactly where
11:57 most people would stop. But here's the
12:00 plot twist. The top 1% go one step
12:03 further. They slow things down
12:06 deliberately. Why is that important? The
12:08 trick that top 1% know is this. They
12:10 know when to shift the gear. Because
12:12 long-term intelligence isn't built
12:14 through convenience, it's built through
12:17 resistance. And that's why we need to go
12:20 to step three, [music] the intelligent
12:24 gym. Most people use AI as wheelchair
12:26 for the mind. And if you sit in a
12:28 wheelchair when you can still walk,
12:31 eventually your legs stop working.
12:34 Atrophy. And today it's happening faster
12:37 than at any point in human history. But
12:39 the top 1% use a very different
12:42 principle. For information task, use AI
12:45 to remove friction. For transformation
12:48 task, use AI to add friction. When you
12:50 go to a physical gym, [music] we all
12:52 know how muscles are built, right?
12:54 Through resistance. You lift
12:57 increasingly heavier weights to
12:58 introduce wear and tear to your muscle
13:02 fibers. So they break and they grow back
13:04 stronger. That is called progressive
13:06 overload. But when it comes to our
13:09 minds, we do the exact opposite somehow.
13:12 You know, we avoid resistance. We use AI
13:14 to outsource our thinking. Write my
13:17 LinkedIn post, fix my resume, summarize
13:20 this book. That's like going to the gym
13:22 and asking someone else to lift weights
13:25 on your behalf. You know, when
13:28 astronauts spend months in zero gravity,
13:31 their muscles and bones atrophy
13:35 dramatically, up to 20%. AI is like zero
13:39 gravity for your thinking. No friction,
13:42 no load, no growth. The intelligent gym
13:44 is not about information. It's about
13:47 transformation. For things where you
13:49 need to be smart and capable, you can
13:52 think of AI as your spotter. In any gym,
13:55 a spotter doesn't lift the weight for
13:57 you. They stand next to you and help you
14:00 lift. They also make sure that you don't
14:02 get crushed when you're lifting the
14:05 weight. So, do the same with AI. Here's
14:09 a concrete example. If you want to learn
14:13 a concept, study it first yourself, and
14:16 then go to your spotter, your AI. Paste
14:19 the concept text and then prompt AI. I
14:21 need to master this concept. Quiz me on
14:23 it. And now comes the most important
14:26 part of your intelligent gym. Ask AI to
14:29 apply progressive overload. Four levels.
14:31 Level one, quiz me like I am a high
14:34 school student. Level two, ask me
14:36 questions like I am a college student.
14:39 Level three, now grill me like you're
14:42 interviewing me for an executive job.
14:44 And level four, now challenge me like an
14:47 iate boss who thinks I'm unprepared. So
14:49 that truly strengthens and deepens your
14:52 understanding on that concept. So now we
14:54 have covered three key steps to learn
14:57 how the top 1% become [music] smarter by
14:59 using AI. But there is one internal
15:01 adjustment that changes everything and
15:04 that is our final step. Step number
15:08 four, the intelligent fool. You know the
15:11 biggest obstacle to intelligence isn't
15:14 ignorance, it's ego. That's why the
15:17 smartest people are obsessed with what
15:19 they don't know. And this is what I call
15:21 the fool's advantage. [music] Let me
15:24 give you an example. Microsoft went from
15:28 $300 billion to 300 trillion in market
15:31 cap [music] with just one mental
15:33 cultural shift. When Satya Nadella
15:36 became the CEO of Microsoft [music] in
15:38 2014, they had missed two huge
15:41 disruptions, search and mobile. The
15:43 cloud race was ongoing but it was
15:45 slipping away from them with Amazon
15:47 becoming the 800 pound [music] gorilla
15:50 and the culture inside the company was
15:53 toxic and political and everyone was
15:55 terrified to admit that there were
15:58 [music] gaps in their knowledge. Satya
16:00 made one cultural move. He told the
16:02 entire company we're switching from a
16:06 culture of knowit alls to learn it alls.
16:09 A complete reboot of Microsoft culture.
16:11 the smartest people in the room were
16:14 finally given permission to say, "I
16:16 don't know," or "I was [music] wrong,"
16:19 and to embrace that beginner's mind.
16:23 Now, Wall Street was skeptical at first,
16:25 but the market cap eventually went from
16:28 300 billion [music] to over 3 trillion,
16:30 and it keeps growing, 10x growth in a
16:32 decade. And here's why this matters. [music]
16:33 [music]
16:36 Neuroscience tells us that our brain can
16:38 rewire all the time. is called
16:41 neuroplasticity. This rewiring happens
16:44 only at the edge of your ability. It
16:46 happens when you are making errors. It
16:48 [music] happens when you're frustrated,
16:50 when you're feeling that discomfort. And
16:53 if you aren't feeling stupid, you aren't
16:56 learning. And aren't you glad that AI
16:58 has just handed you the ultimate
17:00 training ground to be a student again?
17:03 You can bring your beginner's mind to AI
17:05 all day long. Ask questions you would
17:07 never ask your colleagues out of fear of
17:10 embarrassment. AI doesn't roll its eyes.
17:12 Pick one thing that you don't understand
17:14 in your field. Something that everyone
17:17 else thinks you know, but you know you
17:19 don't. And then ask AI the most basic
17:21 questions about that [music] topic that
17:23 you can think of. And then ask, can you
17:26 explain it to me in a simpler way? Teach
17:29 me like I am 10 years old. I ask these
17:31 questions all the time. In fact, I asked
17:33 three times in a row to simplify again
17:35 and again. And sure, I guarantee you,
17:38 you'll feel ridiculous at first. I do
17:40 all the time. But that's the whole
17:43 point. Have the courage to play the fool
17:46 today so you can be the genius tomorrow.
17:48 The trick to mastery is going back to
17:50 simplicity [music] itself. If you
17:52 examine some of the greatest masters
17:54 across human history, you'll see one
17:57 consistent pattern. Every master is a
18:00 student for life. And you can't be a
18:02 genuine student [music] if you're hiding
18:05 behind a mask of mastery. You know, the
18:07 biggest benefit of intelligence is not
18:09 the end of ignorance. It's the end of
18:12 pretending. You know, we're surrounded
18:17 by endless images of flawless people in
18:19 their flawless poses, flawlessly
18:24 photoshopped. But in the end, all art is
18:27 about asymmetry. We're beautiful because
18:30 we're broken. Because the real purpose
18:32 of intelligence, of this thing called
18:36 life, is to travel far and wide only to
18:38 return to yourself [music] and fully
18:42 accept who you are. That is your truest
18:45 intelligence. If you like this video,
18:47 [music] don't forget to subscribe. And
18:48 if you want to use AI to start a
18:51 business, here's another video where I
18:53 walk you through exactly what I would do.
18:54 do.
