AI Isn't Killing Jobs, It's Creating Weird New Ones

Dr. Ryan Ries here, back in your inbox.

For two years now, the loudest story about AI and work has been the same one: the machines are coming for the jobs, it’s just a question of when.

This week served up a pile of evidence that the reality is stranger and more human than the doom and gloom headlines suggest. So let’s take a look.

The Data Says the Opposite

A new Ramp study is the first to match real, firm-level AI spending against actual headcount across 21,599 U.S. companies. The study found that firms that adopted AI most aggressively grew their workforce 10.2% over the following two years. Entry-level roles, supposedly first on the chopping block, grew even faster, up 12% at the biggest spenders.

Honestly, those gains cluster among companies that were already large, well-funded, and engineering-heavy. But I think this study definitely shows the cracks in the blanket “AI shrinks headcount” story.

Everyone’s Hiring… to Deploy AI

Last week Microsoft launched the Microsoft Frontier Company — $2.5 billion and 6,000 specialists whose whole job is to embed inside customers (Unilever and Novo Nordisk first) and make deployments actually work.

Microsoft isn’t alone in this.

• AWS stood up a $1B internal deployment org two days earlier
• OpenAI spun out a Deployment Company backed by $4B+ from private-equity firm TPG
• Anthropic teamed with Goldman, Blackstone, and Hellman & Friedman on a $1.5B venture to embed engineers in mid-sized firms

The AI economy’s hottest job is helping other people use AI.

The Great AI Rehiring

Remember the companies that proudly cut staff for AI? A lot of them are quietly hiring those people back.

Staffing firm Robert Half found that 32% of managers who eliminated a role for AI later rehired for the same or a similar one. Gartner expects half of all AI-blamed cuts to be reversed by 2027.

The pattern is being described as a “60/40 gap”: AI handles about 60% of a repetitive workflow beautifully, then faceplants on the 40% that needs judgment, context, or accountability.

The Strange New Jobs

AI companies are loading up on philosophers and ethicists. While unexpected, in hindsight, this makes perfect sense.

Anthropic and Google DeepMind each keep a half-dozen on staff to wrestle with what a model “believes,” what “honest” means for a system that can bluff, and whether a machine that reports distress deserves any moral weight.

Here’s a wild stat: philosophy grads currently have a lower unemployment rate than computer science grads. Meanwhile a cottage industry of freelancers has popped up doing something very 2026: getting paid to clean up the “AI slop” a model generated in seconds.

Even the Robots Brought a Chaperone

I’ve mentioned this before, but here’s the most straightforward example of how the machines still need us.

Look at China’s humanoid-robot rental boom. Over 153,000 rental businesses.

Then read CNN’s report on what you actually get: most rentals ship with a human operator to control and babysit the machine, and even UBTECH’s most advanced models top out around 80% of human productivity… even on narrow tasks like box-stacking! Good luck trying to get the robot to actually do your dishes.

The robots are impressive. But… they also still need a person standing next to them.

But Let’s Be Honest

At the end of the day, still none of this means AI isn’t displacing anyone. Tech companies have announced roughly 139,000 job cuts in 2026, with AI cited as the top reason four months running, and Meta itself shed about 8,000 roles in its AI pivot. My point isn’t that the machines create no losers. It’s that the tidy “AI replaces humans” headline isn’t so simple.

We’re in the awkward middle of this transition.

I saw a LinkedIn post the other day from Liz Melton, Head of AI/ML Enablement at AWS that summed this up perfectly:

AI removes some work, creates a surprising amount of new work, and routes almost all of it back through a human somewhere.

The company that fired its support team learns about the 40%. The company that “deployed AI” discovers it needs 6,000 people to deploy AI.

My advice: I’d be skeptical of anyone selling you AI as a headcount-reduction plan right now. Organizations shouldn’t be shrinking to make room for AI. They should focus on staffing up to use it well, and pointing their people at the judgment-heavy 40% the machines can’t touch.

One More Cool Thing

Chinese scientists demonstrated a sunlight-only desalination device. It’s a 0.75 m² unit built on a new 3D nanostructure that produced more than 20 liters of clean drinking water a day with zero utility energy, at a projected cost below bottled water.

We love an AI for good story!

Let’s Talk

If your team is trying to figure out where AI actually fits (which work to hand it, which to keep human, and how to deploy it without triggering the boomerang) that’s exactly the conversation we have in our Mission Cloud AIM sessions. We help you sort the hype from the fit and hand you a plan you can act on.

Reply to this email or reach out to our team here.

Until next time,

Ryan

Now, time for this week’s AI-generated image (in honor of the Young Washington movie) and the prompt I used to create it.

Create an 18th century style portrait of me. Use my reference photo. My portrait should show me in attire typical for the time period. There should be a dog sitting at my feet.