Transcription
There are three possible worlds inside these envelopes. Three outcomes of how your pay stop could look 10 years from now. There's a world where you're not only making more money, but working less, maybe 2 days a week, and life is good and chill.
But there's another scenario. There's one where you're not needed at all, where the AI can take over, so why pay you anything? 300 million full-time jobs are exposed to automation. 40% of global jobs are going to be touched by AI. And that's even worse in advanced economies than in low-income countries. By 2035, AI will either make you richer or trap you in some gig work hell or even shove you completely out of the workforce. So, let's open the envelopes. What kind of future are we building?
[Music]
If you're a white-collar worker, there's a good chance that the tools that you use every day have implemented some form of AI already. But this AI is compared to what we would have in 2035. As you open your laptop for work, your email has been processed. The junk is gone. Cold emails gone. Easy replies answered and summarized. Your assistant has already replied to everything except for those really important four or five emails that absolutely need your human decision-making. Your calendar has been rearranged contextually, prioritizing what's most important. Your focus time as well. Every tedious task that you have to do today is gone. It's taken care of.
An AI that's built in your image is interacting with other people's assistants, not just to coordinate your meetings, but to agree on pipelines, to ask simple questions about work, to fill in those details. This is the World Economic Forum's estimate for the number of tasks that will be automated. Notice how there's this human-tech collaboration category that's bound to grow along with tasks performed just by tech. And much fewer tasks end up in humans. Middle management in this world is gone. We essentially have companies with high-level execs that make educated decisions based on tons of data that has been pre-chewed by the AIs and deep workers who are highly specialized in those tasks that AI can't replace. Creativity, advanced coding, which are also AI-assisted but still led very much by humans.
Workplace productivity has increased to a point where you only need 30 instead of 40 hours to get your work done. So you have much shorter days, maybe 4-day work weeks, maybe even less than that.
>> You know, what will jobs be like? Should we, you know, just work like 2 or 3 days a week?
>> At the same time though, new roles have been created. We still need humans to audit these AI automations to improve on them based on data and on the human touch. In 2035, it's common to find job titles such as AI ethicist or algorithm bias auditor or machine psychologist expert in these large organizations. The 2025 Future of Jobs report estimates that by 2030, 170 million jobs are projected to be created while 92 million jobs are projected to be destroyed. But that's still a net of 78 million new jobs.
Every major tech transformation, the Industrial Revolution, the personal computer, the internet, it's eliminated millions of jobs, but at the same time created millions more. So why should this generation of technology be different? Because we have all this free time now. We are living on this spring of coffee shops and restaurants and physical places where people can spend all that new earned free time. Corporate middle management is no longer a trap. Entrepreneurs are able to pursue their passions across physical or digital spaces. A subset of people maybe still appreciate this non-AI-assisted crafts which have now become a premium. So those who resist change can still pursue those passions and these human-touch artisan creative tasks have value for that subset of people.
The GDP of most countries around the world is growing faster than ever, 7% according to Goldman Sachs, and that is trickling down to all layers of society. Blue-collar work though has not been transformed yet, not significantly though. We still need robots to do that. And we are much longer than 10 years away from doing that. We made a whole video about that. I'm going to link it later. But white-collar work is a whole other story.
Now, I'm a big believer in what you can accomplish by being in the same room with people, but the rest of your work can be done from anywhere. Offices will be above everything. These creative spaces, areas where you can meet a few times per month to do that creative, collaborative work where you need people in the same room. The occasional co-worker will tune in with some better version of augmented reality, while others who still appreciate the human touch will still want to see each other in person.
Now, some reports are even more optimistic than that, expecting that 80% of jobs in 2030 don't even exist today. And and a key for this to happen is education, 'cause 59% of the global workforce will need some form of reskilling or upskilling by 2030. But before I jump into education, let me highlight one of the tools that is leading this transformation. Adio is the AI-native CRM for the next generation of companies. Within seconds of syncing your email and your calendar, you'll already get a CRM that's tailored to your business model with data about your interactions fully enriched. With Adio, AI isn't just a feature, it's the foundation. You can prospect and route leads with research agents. You can build powerful AI automations for your most complex workflows and even get real-time intelligence during conversations. With these customizable templates that you can use to track exactly what you care about during the call, this tedious task of filling in a CRM is gone. As an SDR, your focus becomes client attention, not taking notes, 'cause that's all done automatically. Email follow-ups, automatic automatic delivery by time zone, out-of-office detection, and incredibly powerful reports on the performance of your entire automation stack. We started testing Adio to replace our current CRM, and I was really blown away. It felt like AI doing what it should be doing, and it starts at just $29 a month. Industry leaders like Banola, FlatFile, Replicate, and more are all experiencing what's next. Adio pitched into the production of this video and it's offering you guys a 15% discount if you sign up using the link in the description or this QR code here. You can take your first step towards this 2035 dream and help our channel in the process.
Okay, so let's talk about education now. Education is no longer this cookie-cutter approach. The work of the teacher has now been augmented with tons of creative ideas to teach more efficiently, while separate agents can tutor each student on what areas they're struggling most. Stanford actually worked on a tutor called Copilot, and they tested it with 900 tutors and 1,800 elementary and secondary school students. And students that used this AI versus a control group saw a 4% improvement in their math on average. But when looking at the lowest-ranked human tutors, AI's improvement was actually 9%. The average working week for K-12 teachers is 50 hours. Yet only 49% of that time is spent in direct contact with students. So the work of the teacher can become focused on interactive projects, on mentoring, and social or emotional development. Memorization has stopped being the priority in the classroom. It's more about creative thinking. It's discovering talents that AI can't replace and nurturing them from a young age.
There is a direct correlation between education and economic output. But if education becomes accessible to everyone at a much higher quality, there is an exponential possibility to transform society. This idea of unskilled labor is starting to disappear. Also, elite universities lose a lot of their force because knowledge and learning have become commoditized and available to everyone. The university is more of a networking ground, but automation allows everyone to access it without going into debt for the rest of their lives.
[Music]
A higher economic output means more tax revenue for countries which is also not wasted in inefficient governments. We start to see early discussions of legislation maybe around universal basic income funded by this surge in productivity and tax revenue. While still overseen by doctors, we're starting to trust AI to diagnose diseases based on an amount of data that a human could never be able to process. In this world, work-life balance has stopped being some slogan. It's just the standard. The time affluence created by all this AI productivity has really allowed individuals to be more than workers. They're also dedicated parents, artists or learners or active citizens, drastically improving other metrics around mental health and happiness for society.
Setting the ground for this to happen would have to start soon. Stronger caps on overworking employees, laws like the right to disconnect, which is actually gaining popularity around the world. So that instead of using AI agents to make people work more, we use AI agents to allow people to log off. We can't trust ourselves employers not to try to extract more value out of the same hours if this AI comes to be. So the standard for shorter work weeks needs to be enforced through policy and that is up to people to pursue. Traditional education models need to be scratched. K-12 curricula, for example, needs to be updated to include not just digital skills, but also emphasize creative thinking, problem-solving, adaptability, any skill that a human would need to live in an AI-rich world and to stand apart from these AIs. Coding and AI literacy need to become as fundamental as reading or math, and we need to live in a society that leaves us some dedicated time to learn all the time 'cause things move too quickly to fall behind. The additional GDP needs to trickle down to society somehow. And we've explored this concept and there are fair share of papers and even real-life experiments with stuff like universal basic income or at least some universal productivity dividend similar to how sovereign funds end in society. We also would need to stop the US or the AI from profiling or social scoring people. Algorithm auditors must be enforced to ensure that models aren't biased. It's not impossible, but the political will needed to make all of this happen truthfully is just not there.
Let's look at a middle-ground scenario. And for a premise, let's assume that AI just doesn't get much smarter than it currently is. There is a real hard limit to how smart AI models can get. It has to do with the way LLMs work today, which at their core, they're mostly models that find the most predictable word to follow in a sentence. For years, we've theorized that adding more training data, more parameters, more computing power to these models that they'd become significantly smarter, but they aren't. OpenAI released ChatGPT 4.5 in February 2025, and it was massively expensive, and overall just not really smarter than GPT 4.0, so they discontinued it. For context, OpenAI and a bunch of other companies, they've been able to make these models more useful, smarter in theory, not by making the model itself bigger, but by making them reason or by giving them access to some model, multimodal stuff. We have a full very geeky video on how they solve this predicament, but that's for another day. But the point is that this might be foreshadowing a cap on how much AI can truly help us with day-to-day tasks. Call it a point of diminishing returns for all this AI craziness and all of this money spent.
McKenzie estimated that six out of 10 jobs, around 30% of tasks, could be automated. But replacing a job completely is much harder. And this balance or imbalance could be devastating to the workforce for a whole different set of reasons. Because this middle-ground world makes employers expect more work, but in practice, it's still humans that need to do the hard work. A McKinsey report found that 47% of execs thought that their companies were moving too slowly to build AI tools. Google, which has some incredible in-house AI tools, I've been able to see, is pushing employees to work 60-hour work weeks now. And that connects to this rough reality that we've been seeing for decades where better productivity is real. It's tangible. It's quantifiable, but it's not translating into better wages. On the contrary, a survey of 10,000 workers by the workforce lab at Slack found that 81% of them think AI makes them more productive. Now the question is what percentage of that productivity, in other words, the money that these people have created, the extra money, how much of that is ending up in the worker's pockets versus the investor's pockets? Like the crisis danger of losing your job to an AI, the crisis of inflation is great to squeeze more work out of your team. It creates this vicious cycle that ends up in efficiency and economic value for the companies but not reflected for the employees. And this is why this productivity line grows but wages have stayed stagnant. We're already seeing employee burnout at an all-time high.
But in this world, in this 2035 world, the fact that some work has gotten a little automated or at least with the illusion of automation converts a lot of full-time workers into gig workers. And that breaks social security, breaks pension systems, and widens the income gaps. Because in this plateau, you have a job, but it's a patchwork of gigs instead of a steady income. It's stressful, it's unstable, and it absolutely does not let you plan for retirement. This trend is real, by the way, and it has accelerated with AI. And even if these models don't improve too much, they are perfectly capable of facilitating that gig work approach for employers because company investors and the C-suite are going to see their salaries grow with all this extra productivity. They're extracting more value out of their teams, middle management and actual employees. They're just getting more work. They're getting not so great automation and no salary changes. In this potential world, the biggest punch is inequality. Wider income gaps pushed further by this slight increase in unemployment. 50% of Americans think that this is the path that we're headed. An even wider gap because company leaders don't realize that their employees are actually using AI. A McKinsey study from 2024 found that they underestimate how many employees are using AI by 3x. So those who have a grip on prompting or automating some of their work, they have this huge advantage against everybody else. Maybe they're working less. Maybe they're getting two jobs. They might even get away with chatbots on their Slacks or on their emails without their co-workers even noticing. And this might prompt employers to implement maybe some AI surveillance to enforce work hours on their companies, making work shittier for everybody. The CEO of Ford estimated that 50% of white-collar jobs could disappear by 2030. New jobs might be created, but we need bright minds to take them.
So, let's talk about education. In this 2035 scenario, education is stuck. Just as employers underestimate AI use, teachers do as well, while kids basically are prompting their entire homeworks. If the school system and the approach to teaching doesn't change, this is not too far from reality. So, you'd have the smart kids. Smart kids who are leveraging AI for the best to learn more, to complement their teachers' work. You would also have teachers who are leveraging AI to maybe spend more time working with students. Maybe they're doing that behind their boss's backs because they haven't really approved AI for white use. However, those teachers that are trying to take advantage of what we have, they're reserved for maybe private schools or for well-funded school districts. The rest of the kids, the rest of the teachers, they're just using AI to do less work. They're creating these lazy brains. It's the calculator in your pocket analogy all over again. Like look where having calculators in our pockets has gotten us to as a society. But we need to learn stuff on paper first.
In college, denial might make people continue to pick careers that are about to disappear. Journalists, lawyers, software engineers, and those are some of the biggest jobs that this Pew survey found at risk of being snatched by bots. We basically end up with a society that has a hole in the middle. Those at the top are disconnected from everything else. They're imagining AI is doing everything while their employees are just grinding with useless AI assistance. The middle of society is broken. It has badly automated middle management. It has really bad unemployment, but not bad enough to trigger a strong reaction from our governments. At the bottom you have either people that are grinding with three AI-assisted jobs to make ends meet and doing a poor job at all three of them or people with just no job at all. You also might have white-collar workers that came from this middle of the society but now are just ending up at the bottom transitioning to underpaid blue-collar work for unskilled workers. And what do we need to get to this? Honestly, not much. Like this more or less follows the current trend of policy of inequality of wealth caps which might also put us on a path to the third scenario.
In our first scenario, we looked at a smarter AI that made the world better. And in the second one, we looked at a not-so-great AI that went unchecked, no guardrails to society. And this third and last forecast is is a combination of both of those things. And I think that that's why it's so scary. Axios called it a white-collar bloodbath. One that starts with junior workers who are the easiest tasks to replace as AI continues to get smarter and to train itself in doing these junior tasks. AI could wipe half of all entry-level white-collar jobs and spike unemployment to 10 to 20% in the next 1 to 5 years. This is from Dario Amodei, the CEO of Anthropic, who's helping build this, but who also insists that we don't need to sugarcoat it to people. As we approach 2035, the replacement moves from junior to middle management at a pace that new job ideas just can't replace. It also starts to take some skilled white-collar work because AI can now produce first drafts or of legal contracts, write basic news articles, generate marketing copy, and even produce software code at near-human levels. And this means that fewer lawyers, journalists, programmers, or marketing staff are needed at companies. Senior execs and creative visionaries. They still have jobs, but they've replaced their think tanks for debates with a bunch of AI agents. Imagine that 50% of the US white-collar workforce is replaced by bots. That's 35 million people without a job. All junior and mid-level tasks for lawyers, journalists, programmers, marketing. They're all done by AI agents supervised by still crucial humans that are clinging on to their jobs while they can. They've lost all leverage. Forget 40-hour work weeks or collective bargaining for higher wages. This is no man's land. Goldman Sachs goes even further. They estimate that two-thirds of today's jobs are exposed to AI.
>> I mean, will we still need humans?
>> Uh, not for most things. Uh,
>> In this scenario, because we don't assume robotics are a thing yet, blue-collar work still requires some humans. But all this white-collar unemployment has completely devalued those jobs which are now paid the federal minimum wage, which of course has not increased in decades. Actually, the federal minimum wage is 42% lower than what it was in 1968 and 30% lower than what it was in 2009. Humans are still needed in this world for hand labor or physical presence stuff, but there is an oversupply of workers which has basically commodified labor. Now, productivity per worker is up. Of course, you have fewer people working and all of this automation, but serious unemployment means that the country's GDPs are stagnant, which leaves no flexibility for welfare or for assistance to this unemployed growing percentage of the population. Informal and uninsured gig work is the only means of substance for millions of people, which basically belong to this useless class. Now, it's an entire young generation that never got a chance to enter the workforce, and they're still drowning in college debt. The unemployment crisis trickles across all industries and it's an unstoppable snowball. Unemployment reaches levels not seen since the Great Depression. These workers are educated. They have careers or they had, and they have college debt. Workers with mortgages that they haven't paid now basically surviving on public benefits, charity, or informal work. Unemployment is directly correlated to depression and to addiction. A 2017 paper found that a 1% increase in local unemployment translates to a 3% increase in the opioid death rate. Neo-synthetic drugs or opioids are really the only relief for many, and others might escape this with whatever VR world we've developed by then.
The biggest problem with education in this world is that it didn't transform itself fast enough. Most schools and universities just continued with business as usual until it was too late. By the time curricula changed, a generation had already graduated into a disrupted economy. A generation that was educated by teachers who wasted time trying to ban the use of AI or regulate it instead of focusing on teaching or reskilling people around it. The only winners in this world are the elites, the investors, and the execs of these companies. They hold more political power than ever because politicians are fearful that their jobs could be taken as well. We'd probably see the first trillionaire individuals at this point. Actually, the wealth of the five richest people in the world has doubled since 2020. Wealth and power becomes more wealth and power, and taxes for these gargantuan enterprises gets slashed because they control everything. A plutocracy, basically policies are tailored to preserve the elites' wealth. Tax breaks, anti-labor laws, weak antitrust enforcement, and walls begin to be built to separate this extreme inequality from the billionaire elites. Almost like back to feudalism, they retreat to secure enclaves or islands where riots can't reach them. For them, the future of technology is about only one thing: just escaping from the rest of us.
What leads us to this scenario is not only the lack of action, but a subset of policies that benefits billionaires and tech company owners. This scenario ultimately happened because businesses could replace workers with AI as quickly as the tech allowed with no adjustment period or no oversight, and because governments weren't able to collect the benefits of that progress. Humans provided the training, governments provided the platform, but AI companies turned their back on both of them in favor of profits. Companies aren't inherently evil for prioritizing profits, but if it's companies and investors that ultimately rule legislation, then there is truly nobody to stand for the rest of the humans. If AI ends up being the transformative force that it's starting to become, failing to enact policies to cushion workers or to distribute AI's benefits is what could absolutely put us on this third path. We've given you some evidence and some studies to back either direction. So, which one do you guys think is the most likely? Drop it in the comments and check out our video on why the AI can't actually get any smarter. Catch you in the next one.
[Music]