*Note to readers: This month, we’re stepping away from our usual focus on industry updates and case studies to lay out a framework exploring labor market disruption. We hope you enjoy it.*

Have you heard that AI is taking your job?

Oh wait…we aren’t supposed to say that out loud. Oops!! Instead, we can say “AI won’t take your job, but a person using AI will.” We can also talk about scenarios where AI provides leverage in place of labor. Cue Sam Altman’s prediction of a one-person billion-dollar company from February this year, and Vinod Khosla's decade-old essays about algorithms replacing teachers and doctors.

I wonder if, perhaps, we should just say the quiet part out loud: Yes, AI is going to take many of the jobs currently held by humans. AI is inherently better, cheaper, and faster at certain tasks. Would you pay a human to analyze traffic data and recommend a route for you to drive? Of course not. Google Maps is great at that (and free). 

So instead of ignoring this reality, let’s try to understand the types of jobs where humans will be better than AI, and vice versa.

A framework for jobs.

What's a simple way to categorize the millions of jobs that humans do today? That's a tough question. But I would posit that we can fit most, if not all, of today's jobs into four broad categories:

Are AI's or humans going to be better suited to each of these types of jobs?

1. Processing

These are repetitive, well-defined tasks—examples include accounting, recruiting, sales, fulfillment, and customer support. AI excels at processing inputs and following defined steps to produce outputs. In the physical realm, computer vision and robotics are advancing rapidly, with robots soon expected to handle repetitive manual work. Waymo and Tesla are leading the way in autonomous vehicles, and BMW is testing humanoid robots on its production line. It’s becoming clear that AI will soon handle most processing jobs.

Verdict: AI, now. 

2. Evaluating

This type of work is less defined, requiring the ability to locate relevant information, ask incisive questions, and navigate ambiguity—challenging even for humans. For AI to handle this, it must “think”: evaluating potential solutions, connecting to systems or people for information, and then organizing and synthesizing that data. This is possible for AI, but not imminent. Dario Amodei, CEO of Anthropic, recently described a future where AI autonomously completes complex tasks over days or weeks, much like a smart employee. Amodei suggests this could happen as soon as 2026. We’ve already seen progress, such as OpenAI’s o1 model hypothesizing and reasoning through solutions. However, several technological leaps are still needed, so for now, humans remain central to this type of work.

Verdict: Humans, for now.

3. Deciding

The classic executive role involves using expertise, data, and insight to make complex, high-stakes decisions. For AI to take on this role, it would need deep context, specialized domain knowledge, the ability to set objectives, and the capacity to make well-reasoned decisions without hallucination. Despite advances in specialized AI agents, this remains unlikely for now. Even if it becomes technically feasible, it’s unclear whether humans would trust AI with such responsibility.

Verdict: Humans.

4. Relating

Relating tasks depend on emotional intelligence, trust-building, and interpreting subtle human reactions—skills that are deeply interpersonal. While AI has advanced in simulating empathy, it still lacks genuine emotional intelligence. These tasks remain uniquely human, requiring understanding and sensitivity that AI cannot replicate.

Verdict: Humans.

Implications for Business Leaders

If the above analysis is correct, AI can and should already be doing all processing tasks. And soon, AI will augment or do evaluating tasks. Humans will specialize in deciding and relating tasks, while learning a new skill: leveraging and supervising AI. 

In previous newsletters, we’ve highlighted businesses using AI for processing, though we hadn’t labeled it as such. As companies have success in using AI, we can expect an “arms race”. Once the USA had nuclear weapons, competing countries had to develop nukes (or no longer be able to compete militarily). Once western companies began outsourcing to access cheaper labor in China, competitors had to outsource as well (or no longer be able to compete economically). And so it will be with AI. The clearest example has been in customer support, leading to cogent analysis of call center obsolescence. And so for business leaders, the question is when to adopt AI: 

5. segments of the technology adoption lifecycle

Source: Crossing the Chasm, Geoffrey Moore

Early adopters will gain higher margins, resources for future investments, and a head start in AI management skills—compounding their advantage as AI advances. Those who wait must be ready to follow fast or lose an edge to competitors.

Where will ROI come from?

Yes, you should probably get at least some team members that $25/month subscription for secure LLM access, but that alone won’t drive large, measurable ROI. The biggest returns come from the reality that AI can replace humans for processing tasks. History provides a useful precedent: in the 1970's, major banks employed humans, armed with calculators and typewriters, for bookkeeping. When computers were introduced, banks did not simply give each human a PC to replace the calculators and typewriters; rather, they transitioned all of their bookkeeping to computers. 

What about us humans?

To be clear, I am not suggesting that humans will be out of jobs; rather, that humans will do different jobs. There will be wrenching job disruption in sectors including call centers, sales and logistics. But as history shows, humans are incredibly adaptable to new technology. The banking sector is still employing millions of Americans; we just aren't sitting in back offices doing bookkeeping on typewriters anymore. 

Conclusion

Let's face the hard truth: AI is coming for human jobs as we have known them. Processing (and soon evaluating) jobs will be done by AI. Humans will lean into their innate advantage in relating while developing skills around orchestrating and managing AI. Businesses will be compelled to make the transition to AI, because capitalism is an arms race. The only question for each business is when, not if.