0:02 Headline after headline is about yet
0:04 another company laying off thousands of
0:06 workers due to AI. If you're wondering
0:08 what signs to look for to tell if you're
0:10 about to be laid off, I'm going to
0:12 explain step by step how exactly AI
0:14 could take your job. This video is a
0:16 scenario based on the intelligence
0:18 curse, a deeply researched and widely
0:20 circulating report in the AI industry.
0:22 The Federal Reserve chairman Jerome
0:26 Powell says the AI hiring apocalypse is
0:28 real. Job creation is pretty close to
0:31 zero. Let's imagine you're the CEO of a
0:33 firm with a thousand employees, most of
0:35 whom are junior or entry- level. A new
0:37 AI company approaches you with a pitch.
0:39 An AI [music] agent that can do the work
0:41 of a junior employee for 1/5if of the
0:44 cost. The more aggressive firms in your
0:46 sector immediately fire most junior
0:48 employees, but you're cautious. You're
0:50 skeptical that this new AI can replace
0:52 humans, so you treat it as a tool.
0:54 Instead of outright replacing [music]
0:57 humans, everything the AIs do is subject
0:59 to human approval. But within months,
1:01 the AI agents are outperforming most of
1:03 your human employees. This comes as a
1:05 shock to you. But the experts aren't
1:08 surprised. From 2019 to 2025, the length
1:10 of tasks AIs [music] could perform
1:13 doubled every 7 months. This is just the
1:15 trend continuing. And here's the thing,
1:17 you might feel uneasy with these
1:18 developments. But you can't get out of
1:20 this situation because the more that
1:22 your employees rely on the AI tool, the
1:24 faster and better their work becomes. In
1:27 fact, your top performers are now the
1:28 ones who let the AI do everything before
1:31 clicking approve. For basic tasks,
1:33 humans in the loop aren't improving the
1:35 work. They're slowing it down. And if
1:38 you can see that, so can your board.
1:40 Every day, more and more AIs join the
1:42 workforce. The board is happy with the
1:44 extra productivity, but not with the
1:46 payroll. They have a question. Why
1:48 aren't you firing the unproductive human
1:50 employees? You argue to keep them on
1:52 humanitarian grounds. You don't want
1:54 your workers to be replaced by AI. The
1:56 board isn't convinced, but you warned
1:58 them that mass firings would be bad PR
2:00 for the company, and that is enough to
2:03 get the board off your back for now. End
2:05 of quarter, your earnings are up, but
2:08 your competitors, the ones who replaced
2:10 humans entirely, have reported even
2:12 higher earnings at a fraction of your
2:14 costs. In any other economy, your
2:17 results would be a win. But here, you're
2:19 losing market share, and your stock
2:21 price takes a massive hit. So, the board
2:24 delivers an ultimatum. Fire the
2:26 unproductive junior employees or we'll
2:28 replace you with a new ruthless CEO who
2:30 will replace all of your junior staff
2:32 with AI. You know that you can't do any
2:34 good if you get fired. So, saving some
2:36 jobs is better than saving no jobs. So,
2:39 you compromise, firing half your junior
2:41 staff. You hang on to your own role and
2:43 convince yourself that maybe someday, if
2:45 the board allows it, you'll be able to
2:47 hire all your old staff back. After all,
2:49 no one else is hiring them, so they'll
2:51 still be on the job market. But AI
2:53 progress doesn't slow down. At this
2:54 point, your managers are managing more
2:57 AIs than humans. And every month, the AI
2:59 agents improve, doing more and more
3:01 work, while the human employees remain
3:03 [music] stagnant. AI has set a new
3:05 standard that all junior human employees
3:08 are failing to live up to. So, the board
3:09 wants you to fire all of them. You offer
3:12 a compromise. All junior employees will
3:14 be put on a performance improvement
3:15 plan, giving them one last chance to
3:17 prove their worth to the company and to
3:19 prove that they can do the work that's
3:22 just as good as AI. A month later, a
3:23 tiny handful of exceptional junior
3:26 employees who have proven their worth
3:28 are promoted to management, but over 90%
3:31 of them are fired. One of them is Alice.
3:33 She begins searching for a new job, but
3:35 finds that no one is hiring. All other
3:38 firms in the sector are laying off staff
3:40 as quickly as they can get away with.
3:42 From her kitchen table, Alice scrolls
3:44 through the headlines. [music] GDP is
3:45 climbing. The stock market is booming
3:47 under the new productivity gains from
3:50 AI. And yet, unemployment rates are
3:53 approaching 12%. Worse than any point
3:55 since the Great Depression. She's just
3:57 part of the statistic. And it's about to
4:00 get worse. The new generation of AI
4:02 agents doesn't just intend to replace
4:04 the junior staff. It comes for the
4:07 managers. These systems can coordinate
4:09 entire teams of AIS, track every
4:12 decision in real time, and optimize
4:14 workflows with a precision no human
4:16 manager can match. All at a fraction of
4:18 the cost. When you look at your ORC
4:21 chart, it's AIS managing AIS, and the
4:23 numbers prove it works. At your next
4:25 quarterly performance review, the
4:26 numbers are better than ever, but the
4:29 board is furious. To them, the results
4:31 only confirm what they've been saying.
4:34 Human managers can't compete. Your
4:36 reluctance to cut them loose looks like
4:38 weakness. Again, you offer a compromise.
4:40 All human managers are now on a
4:42 performance improvement plan. Prove you
4:45 can be as good as an AI or lose your job
4:46 for failing to meet standards. A month
4:49 later, 90% of your human managers are
4:51 gone. At this point, unemployment has
4:53 passed 15%, [music] rapidly approaching
4:56 20%, with job loss numbers five times
4:58 worse than the Great Recession. Yet, the
5:00 stock market keeps rising thanks to AI
5:02 field productivity. For Alice, now 6
5:04 months into her failed job hunt, the
5:06 message is clear. No one's coming to
5:09 save her. So, she decides to fight back.
5:11 She joins rallies, writes opeds, and
5:14 throws her voice behind one big idea:
5:17 UBI, a universal basic income that would
5:20 guarantee every citizen $1,000 a month.
5:22 It sounds like hope, but the backlash is
5:24 immediate. Critics point out that $1,000
5:26 per month for every citizen would eat up
5:28 twothirds of the budget, sending the
5:30 country further into debt while
5:32 rewarding people for not working. Some
5:34 economists take Alice's side, offering
5:37 justifications for UBI. They say the
5:39 increased tax revenue from GDP growth
5:40 will cover it and inflation will
5:42 effectively reduce the national debt.
5:44 But the debate ends when the US
5:46 administration's AISAR weighs in. His
5:49 verdict is blunt. UBI is a utopian
5:51 fantasy and the administration will veto
5:53 any bill proposing it. Alice is
5:56 devastated, but no one is surprised. The
5:58 AISAR's comments perfectly echo what he
6:00 said back in June 2025. The left
6:03 envisions a posteconomic order in which
6:05 people stop working and instead receive
6:07 government benefits. In other words,
6:09 everyone on welfare. This is their
6:11 fantasy. It's not going to happen. For
6:14 Alice, the fight feels over. For AI,
6:16 it's just [music] beginning. The same
6:19 month the UBI bill dies, AI clears its
6:22 next milestone. In 2025, it could handle
6:24 tasks that took hours. Now the newest
6:27 generation can run monthsl long projects
6:29 without any human input or oversight.
6:31 The board has seen AI replace and
6:34 surpass all junior level employees, all
6:36 middle managers, all senior managers.
6:38 Now all that's left is the executives in
6:40 the seauite. They stop short of firing
6:42 all of you outright, leaving you on the
6:45 org chart in name only. Every day you
6:47 participate in the humiliating ritual of
6:50 coming to the office to review the AI's
6:52 work. And each day with increasing
6:54 horror, you realize that the AI is doing
6:55 a better job than you could ever
6:57 possibly hope to. After all, how could
6:59 it not? The AI has more knowledge than
7:01 you. It has more training than you. And
7:03 because the AI CEO can monitor every
7:05 single decision made within the company
7:07 with perfect accuracy, it has insight
7:09 into operations that no human ever could
7:12 match. Unemployment smashes through 25%,
7:13 [music] breaking every historical
7:15 record, even the worst days of the Great
7:17 Depression. On the same day that the S&P
7:20 500 hits a new all-time high with record
7:23 stock [music] growth,
7:25 Ellis's landlord is taking her to court
7:27 to evict her family from the house she's
7:30 called home for the past decade. So, she
7:32 jumps on a bus to Washington DC and
7:34 joins a 100,000 people swarming the
7:36 National Mall, chanting, "Humans first!"
7:38 and waving UBI banners. But you wouldn't
7:40 know it from the news. TV networks
7:43 barely mention it. On social platforms,
7:45 live stream links vanish from feeds
7:47 under the guise of violence prevention
7:50 and misinformation. The same censorship
7:52 playbook, authoritarian states perfected
7:54 years ago in the pre-ai era. America,
7:57 [music] the world's oldest democracy,
7:59 claims to still have free speech, but
8:01 the online platforms that control the
8:03 spread of information silence the
8:04 protest. Turnout at the next [music]
8:07 rally plummets. Families keep losing
8:09 jobs. healthcare and homes, but the
8:11 story almost fades from public view.
8:13 Alice can't believe what she's seeing,
8:14 but when she tries to talk to her
8:16 neighbors, they brush her off. Keep up
8:18 with the job search. Sometimes the job
8:21 market is hard. Half the people she
8:23 talks to refuse to believe her when she
8:24 says that this [music] time it's
8:26 different. Alice then realizes something
8:28 that has always been true. People's
8:30 politics aren't driven by numbers,
8:32 they're driven by vibes. And the vibes
8:34 during Alice's most recent protest are
8:37 all about the breakup between a pop star
8:39 and her star athlete boyfriend with
8:41 millions of posts arguing over who's to
8:43 blame. As if that mattered more than the
8:45 collapse of the job market. And while
8:47 the public obsesses over celebrity
8:49 drama, the real drama is unfolding in
8:51 the boardroom and it's about to land
8:53 squarely on your desk. You're still the
8:56 CEO technically, but the board calls you
8:57 in to confirm what you've already
8:59 suspected. The AI is doing your job
9:01 better than you ever could. Your company
9:04 is now run by an AI CEO. It makes every
9:07 decision, flawlessly executes them,
9:08 monitors every part of the business with
9:11 perfect precision, and works 24/7
9:13 without pause. It has beaten every
9:15 target the board set. You're only human,
9:17 and that makes you obsolete. The stock
9:20 market keeps climbing, but only AI
9:22 companies selling to other AI companies
9:25 are breaking records. The rest, the ones
9:27 selling food, clothing, or anything
9:30 meant for humans, are drying up. With
9:32 25% unemployment and rising poverty,
9:34 there's simply less [music] profit in
9:36 selling to humans. Politicians take
9:38 note. Most of the government's tax
9:40 revenue, and most of their campaign
9:42 donations are coming from AI firms and
9:44 their shareholders, so the AI economy
9:46 that funds them becomes their main
9:49 constituency. In the years to come,
9:51 public schools crumble. With no need to
9:52 prepare a new generation for the
9:54 workforce, why should public schools be
9:56 anything more than glorified daycare?
9:58 Instead of reading and math classes,
10:01 kids now spend all day watching AI
10:03 generated entertainment served by a
10:05 screen that's cheaper than a teacher
10:06 with an education degree. The few
10:09 remaining universities are exclusivemies
10:11 for the children of elites, hearkening
10:14 back to the pre-industrial days of the
10:16 18th century when less than 1% of people
10:18 received a university education. And
10:20 with AI driving all innovation and
10:22 productivity, upward mobility has now
10:25 been completely eliminated. The only
10:27 wealthy people left are those who
10:29 acquired their wealth prior to the AI
10:32 takeover of the economy or those who
10:34 inherited it.
10:36 This scenario might seem far-fetched,
10:38 but it doesn't take [music] much for
10:41 this dystopia to become a reality. For
10:42 this scenario to work, you only [music]
10:45 have to believe in a few things. AI will
10:48 continue improving up to and past the
10:51 point of human intelligence.
10:53 Running a human level AI will eventually
10:56 be cheaper than paying a human for the
10:58 same job.
11:00 Private companies exist to maximize
11:03 profit and when given the chance will
11:06 replace human employees with AI. Even
11:08 the firms that initially resist will
11:10 eventually be forced by competitive
11:14 pressure to replace humans with AI.
11:17 The media will be or is already largely
11:19 controlled by corporations that have the
11:21 mechanisms to throttle the spread of
11:23 ideas that they see as destabilizing.
11:25 The government is also driven by
11:27 incentives. And in a world where most
11:30 tax revenue comes from AI firms, it will
11:33 largely cater to the interests of AI. Do
11:35 any of those sound implausible to you in
11:37 the next 5 years? Think about how
11:39 different the world was 5 years ago
11:41 before the emergence of AI tools like
11:43 chatbt. You might argue, well, we've
11:45 been here before. Every wave of
11:47 disruptive technology has always
11:49 replaced some jobs, but people have
11:51 always found some new ones. And you'd be
11:53 right. AI isn't the first time we've
11:54 seen a new technology threaten to
11:57 replace human labor. But AI is unique
11:58 because the companies building it have
12:00 [music] explicitly made it their goal to
12:03 replace all human labor. Open AI
12:05 literally made it part of their founding charter.
12:07 charter.
12:08 And to make matters [music] worse, AI
12:11 can accelerate the spread of the paradox
12:13 of plenty. When technology or resources
12:16 make a country richer while keeping most
12:18 of the citizens poorer by concentrating
12:20 wealth in the hands of a few individuals
12:23 and corporations, we've already seen it
12:25 with oil and minerals. Why is Venezuela,
12:27 owner of the world's largest oil
12:30 reserves, enduring near famine
12:32 conditions? Why is the Congo sitting on
12:34 trillions of dollars in mineral wealth
12:36 while its people are among the poorest
12:38 [music] in the world? Or look at
12:41 Nigeria, Africa's biggest oil producer
12:44 with an appalling poverty rate of 47%.
12:47 This is the resource [music] curse. Oil,
12:49 diamonds, and gold can make a country
12:51 rich on paper, yet individual citizens
12:54 can still go hungry. In fact, when a
12:56 country's wealth is more concentrated in
12:59 a single extractive industry, it becomes
13:02 a natural recipe for corruption. Whoever
13:04 owns the oil rigs or diamond mines has
13:06 all the money they need to buy off the
13:08 generals, judges, and media. But how can
13:10 we avoid this? The countries that
13:13 escaped the resource curse were the ones
13:15 that had human-driven industries, where
13:17 the wealth wasn't owned by a single
13:20 person and dug out of the ground, but
13:22 created by human labor.
13:25 Countries like Taiwan or South Korea
13:28 didn't have a natural resource jackpot.
13:30 The only way that these countries could
13:32 become richer and more powerful was by
13:34 investing in their people. Giving people
13:36 access to food, nutrition, and schools
13:38 so that they'd grow up to be more
13:40 productive citizens capable of designing
13:41 better smartphones, [music]
13:43 microchips, and entertainment. When your
13:46 biggest industries rely on human talent,
13:47 the humans have the bargaining power to
13:49 demand [music] better living conditions.
13:51 But now AI threatens to take that
13:54 leverage away. It could do what oil did
13:56 to countries like Venezuela and Russia,
13:58 concentrating power and wealth in the
13:59 hands of a small number of corrupt
14:01 oligarchs at the expense of everyone
14:04 else. A kind of wealth that benefits the
14:06 few while making everyone else poorer.
14:08 In a world where AI can do everyone's
14:11 job for less money, humans have no
14:13 leverage, no bargaining power to demand
14:16 a higher minimum wage. This is called
14:19 the intelligence curse.
14:23 We want a future where AI uplifts humans
14:25 without [music] replacing humans. So,
14:27 how do we prevent AI from replacing us?
14:29 What are the things we should let AI do?
14:30 And what are [music] the things we
14:32 shouldn't let it do?
14:34 We need to keep an eye on three [music]
14:36 things. Whether AI is intelligent,
14:38 whether it's autonomous, and whether
14:41 it's general. We can have a highly
14:43 intelligent AI that is good at only one
14:46 specific thing. Take Alphafold, the AI
14:48 that won the Nobel Prize in chemistry in
14:51 2024. It's good at one specific thing,
14:53 predicting protein structures. It takes
14:56 research that might have taken months or
14:58 years and accomplishes it in hours. It's
15:00 a massive scientific breakthrough, but
15:02 Alpha Fold is only good at one specific
15:05 thing. It helps medical research, and it
15:06 doesn't threaten humanity's future. We
15:08 can also have AIs that are intelligent
15:10 and autonomous. Look at self-driving
15:13 cars. Driving is a complex task that
15:14 requires an advanced AI. to safely
15:16 navigate the roads. And these cars are
15:18 also autonomous, but they don't have
15:20 general intelligence. A self-driving
15:23 Whimo car is only good at one thing,
15:25 driving. The part where AI becomes
15:28 dangerous is when it becomes general. A
15:30 general AI is good at everything. And
15:31 we're getting closer to seeing general
15:34 AI with LLMs and multimodal AIs like
15:37 cloth and GBT. An AI with general
15:38 intelligence could replace human workers
15:41 at practically every job, which again is
15:44 the goal that OpenAI openly admits to.
15:46 But the danger isn't just that the AI
15:48 will make humans irrelevant. A general
15:50 intelligence that's good at everything
15:52 could learn to improve itself or create
15:54 new AIs. It can become more and more
15:57 powerful exponentially until it [music]
15:59 becomes a super intelligence, far
16:02 smarter than even the smartest humans.
16:04 AI scientists are terrified that this is
16:08 not only possible but likely to happen.
16:10 Watch this video next to understand
16:12 [music] exactly how a super intelligent
16:14 AI takeover might happen based on a
