Most likely AI will change your job rather than remove it, and it will reward people who use AI over those who do not. AI is strongest at repetitive, well-documented tasks, so roles built mostly on those are the most exposed. Work that depends on judgment, taste, and deciding what to build grows more valuable. As Perplexity CEO Aravind Srinivas argues, the bigger story is a wave of new companies built by very small teams, which creates opportunity rather than simply erasing it.

Will AI create new jobs?

Most founders close to the technology think so, though not in a one-for-one swap. The clearer effect is that AI lowers the cost of starting a company, so more companies get built, each with a smaller team. Perplexity's Aravind Srinivas describes new companies reaching multi-hundred-million-dollar valuations with 20 to 30 people and adding trillions of dollars of new GDP. That shifts where work happens rather than deleting it: from doing tasks by hand to directing the tools that do them.

How do I keep AI from replacing me?

Become the person who uses AI rather than the person out-produced by someone who does. NVIDIA CEO Jensen Huang's line captures it: you are more likely to lose a job to someone using AI than to AI itself. Practically, automate the repetitive part of your own work, then spend the time you free up on judgment: choosing what to build, talking to customers, setting the standard, and reviewing output. Getting fluent with one strong agent workflow is the highest-leverage move you can make.

Will AI Take My Job? An Honest Answer

Q: Which jobs are most at risk from AI?

The jobs most exposed are the ones made up largely of repetitive, well-documented tasks a model has effectively seen many times, including parts of entry-level white-collar work. Anthropic CEO Dario Amodei has warned that AI could eliminate a large share of entry-level white-collar jobs within one to five years. Roles that center on judgment, relationships, physical presence, or deciding what to build are far less exposed, because those are the parts a model cannot supply on its own.

If you are worried that AI will take your job, you are asking a reasonable question, and most of the answers you will hear are useless. One camp says the machines are coming for everything. The other says nothing will change. On the 20VC podcast with Harry Stebbings, Perplexity co-founder and CEO Aravind Srinivas gave an answer more useful than either, in a clip titled "Everyone's Wrong About AI and Jobs." His point: the "all your jobs are going away" panic misses what is actually happening, which is a wave of new companies getting built by very small teams. Here is the honest version for founders, and what to do with it.

The short answer

AI is unlikely to take your job outright. It is very likely to change what your job is, and to reward the people who use it over the people who do not. The tasks most exposed are the repetitive, well-documented ones a model has seen a thousand times. The work that grows in value is the part that needs judgment, taste, and knowing what to build. If you are a founder, the shift runs in your favor: AI lets a tiny team produce what used to take a large one, which means more companies get started, not fewer.

Why "AI takes all the jobs" is the wrong frame

Srinivas runs Perplexity, an answer engine built to fact-check claims, so he has a stake in cooling the panic. But the structural point stands on its own. His argument is that the fear-mongering framing, "all your jobs are going to go away," describes the wrong future. What he sees instead:

"There's going to be lots of amazing companies that are going to get built with far fewer people getting multi-hundred-million-dollar valuations with like 20, 30 people and propelling trillions of dollars of new GDP."

Then the part founders should underline: "Let's talk about how to enable that. Let's talk about how to build that and create a more positive future." The jobs story is not a fixed pie that AI eats. It is a growth story where the unit of value shifts from headcount to output, and small teams get to capture value that used to need hundreds of people.

This is not a lone opinion. The same pattern shows up across the founder conversations we track: small teams wired with AI into every process competing with companies many times their size, and reaching high valuations with very few employees. The metric that captures it is revenue per employee, and it is climbing fast.

The honest part: some jobs really do shrink

The optimistic case is not the whole case, and pretending otherwise is its own kind of fear-mongering. AI is genuinely good at the repetitive slice of knowledge work, and that slice is a real part of many jobs. Anthropic CEO Dario Amodei has been blunt that this has a cost: he has warned that AI could eliminate a large share of entry-level white-collar jobs, on the order of half, within one to five years. You do not have to accept the exact number to take the direction seriously. Roles that are mostly well-documented, repeatable tasks are the most exposed, and "my job is safe because a computer could never do it" is a weaker bet every quarter.

The honest answer, then, is not "relax, nothing changes." It is that the work is being repriced. The repetitive part gets cheap. The judgment part gets more valuable. Where you sit on that line is the thing to manage.

The real dividing line

The most useful way to think about it comes from NVIDIA CEO Jensen Huang:

"You may or may not lose a job to an AI. But you will absolutely lose a job to someone who uses AI."

That is the practical version of Srinivas's optimism. AI is not a wave that hits everyone equally. It is a tool that widens the gap between people who wield it and people who do not. The founder who ships with a fleet of agents outruns the founder who does everything by hand. The same is true one level down, for every role in the company. The question is not "will AI replace me" in the abstract. It is "am I the person using it, or the person being out-produced by someone who is."

What actually happens to founders and small teams

For founders specifically, the jobs panic gets the story backwards. AI does not shrink your opportunity, it lowers the cost of starting. Garry Tan, YC's CEO, frames it as using AI to enhance what humans can do rather than replace them: he describes directing agents to ship, as one person, on the order of the work of a 400-person engineering team. That is not a job disappearing. It is a job getting a force multiplier.

The knock-on effect is more companies, not fewer. When one experienced person can build what used to need a team, more ideas become viable, which is the whole premise behind the solo founder who now builds like a team. Fewer people per company does not mean fewer companies or less work overall. It means the work moves: from doing the task to directing the machines that do it, and from filling seats to choosing what is worth building.

What your job becomes

If AI takes the repetitive part, what is left is the part only a person supplies. Andrej Karpathy, a founding member of OpenAI and the former director of AI at Tesla, makes the point that taste and judgment stay essential precisely as agents handle more of the technical details. The model can write the code, draft the copy, and run the analysis. It cannot decide what is worth doing, what "good" looks like for your customer, or which of a hundred plausible directions is the right one.

So the job that survives, and the job worth training for, is the one built on judgment: choosing the problem, talking to the people you are solving it for, setting the standard, and reviewing what the machines produce. As a CISSP, I would add one more: knowing where not to let an agent run unattended, especially anything touching money, customer data, or security. That judgment is not going away. It is becoming the whole job.

What to do this week

Pick the most repetitive task in your week and hand it to an AI tool. That is the part most exposed, so get ahead of it by automating it yourself.
Find the part of your work that needs judgment, the choosing and the reviewing, and spend more of your time there.
If you have an idea, price out the small-team version. The cost of starting is lower than it was a year ago, which is the real message under the jobs headlines.
Get fluent with one agent workflow, well enough to out-produce someone doing the same job by hand. Be the person using AI, not the one being out-produced.
Draw your own no-fly zones: the decisions and systems where a human stays in the loop no matter how good the model gets.

The jobs question has an honest answer: AI reprices the work, rewards the people who use it, and lets small teams build things that used to need hundreds. Learning to run your company that way, as one person plus a fleet of agents, is exactly what the AI Operating System for Startups is built to teach.

Sources

Everyone's Wrong About AI and Jobs (20VC with Harry Stebbings), the clip this article distills, featuring Perplexity's Aravind Srinivas.
Background on Aravind Srinivas and Perplexity: Observer and his LinkedIn profile.
The "lose a job to someone who uses AI" line is from NVIDIA CEO Jensen Huang (Sequoia Capital).
Dario Amodei's warning on entry-level white-collar jobs: Bloomberg Originals, "Inside Anthropic", and widely reported.
From our video knowledge base: Tokenmaxxing: How Top Builders Use AI To Do The Work Of 400 Engineers (Y Combinator, on AI enhancing rather than replacing people) and Andrej Karpathy: From Vibe Coding to Agentic Engineering (Sequoia Capital, on taste and judgment staying essential).