Skip to content
Missing Link — home
← All articles

Article

When AI Hiring Went Wrong

“Head of AI” can mean four very different jobs and most failed AI hires happen because companies hire before deciding which one they need.

Miklos Blasko

Miklos Blasko

November 16, 2026 · 6 min read

A founder of a mid-size developer (+$50M) called me back last week wanting to hire a Head of AI. Reasonable ask. So I asked the obvious follow-up: to do what?

20 minutes later, we had established that what he needed was a production engineer who could wire generative tooling into an existing art pipeline. The six-figure leadership hire they had budgeted for would have likely spent a year with nothing to lead, writing strategy decks for a team that did not exist and benchmarking vendors in a market where new leaders emerge on a monthly basis.

We have some version of that conversation almost every week. The volume of studios wanting to hire for AI has exploded. No other way to describe it. Yet the clarity about what that means has not moved an inch.

Titles tell you nothing

Director of AI. Head of AI. AI Chief of Staff. Forward Deployed Engineer. Sometimes these are simply seniority variations on the same job. Often the focus, internal versus external, is completely different. More often, the title means nothing at all because the job description was written by AI, which is a poor authority on what your company's current stage actually calls for.

For example, Netflix advertised a Director of Generative AI for Games asking for a decade-plus of gaming experience plus deep AI/ML expertise, with a disclosed band of $430,000 to $840,000. Elsewhere, roles with generative AI in the title get posted as hourly contract work.

This matters more than it sounds. Games-industry AI talent largely does not advertise itself with clean, searchable titles. The best candidates are usually inside another function: a technical director who built the pipeline tooling, an analytics lead who shipped the moderation model. They are not labelled experts. That’s why a keyword search finds you people who optimized their LinkedIn headline. And that’s a different population from people who did the actual work. It happens to be the same population that demands a bombastic AI title.

Head of AI is four different jobs

The responsibilities problem is even worse than the naming problem since surprisingly many companies offer almost a blank slate for their AI hires. In my experience, the misalignment of expectations by not setting expectations leads to costly terminations within a year.

I’ve learned that a “Head of AI” can mean roughly four things:

  1. A research leader pushing model capability
  2. An internal transformation leader changing how the company operates
  3. A creative-pipeline leader embedding generative tooling into art and content
  4. A live-ops leader applying ML to player data, moderation, and matchmaking.

If you’re confused, you should be. That’s four different people. Four different backgrounds. Four different definitions of success. And a high chance of hiring the wrong person. A CEO who has not decided which mandate they are buying can end up with a leader optimizing the wrong thing. And they will realize this in month eleven.

We actually had a case like this: A public gaming company hired a head of AI to take their studios into the new era. The leadership believed they were hiring an internal transformation leader, yet what the studios needed was a live-ops leader. They were all bought in. They needed direct execution support. That journey ended inside the first year with the head of AI looking for another job.

Here’s a thing to keep in mind: the adoption data says the value is landing first in the operational layer. This means crafts like coding, QA, marketing, analytics, and art. Square Enix has publicly committed to automating 70% of its QA and debugging by the end of 2027.

The role you need first is usually the unglamorous one that makes your existing pipeline faster. Not the moonshot hire that you hope will make your board stop asking about “how you’re using AI in your company?”

Head or Team First?

Based on what I just said, you must think that I’m always suggesting individual contributors (ICs) over managers and heads. So let me insert a twist.

Hire team first when you have a clear, bounded use case. Automate the QA pass. Accelerate the texture pipeline. Build the live-ops analytics layer. If you already have engineering or production leadership credible enough to direct that work, a senior AI leader is premature. Get ICs in, ship something real, learn what the actual problems are. Create proof before a strategy function.

Hire the head first when AI is cross-cutting and politically contested across teams, when the failure mode you are avoiding is fragmentation, or when you are at a scale where a transformation mandate needs an owner with authority to make tradeoffs and kill bad projects. At that point, a strong leader is the thing standing between your company and an expensive mess.

Both hires share one root error: buying capability with no judgment attached or buying judgment with nothing to apply it to. The right sequence depends entirely on which one you are currently short of. Most companies do not know which that is until someone maps it out for them. And they think that buying judgment is the same as buying mapping and capability in one package. It’s not.

The EA story is the clearest public case of AI strategy with nobody owning the execution. Plus the slower mistake most studios are making right now with their juniors.

Where, why, how long, and how much will it cost?

Every founder, executive, and hiring manager I talk to about AI hiring believes their problem is finding the person. It almost never is. Trust me, the market has people. What I find lacking are companies that can articulate what the person is and isn’t for.

This makes AI hiring a mirror. The clarity of your job description is the clarity of your strategy. If you cannot describe the role without using the word AI, chances are, you have not decided anything yet. You have just noticed that other people are moving and want to be seen moving too. And so you hire someone to take you somewhere. But you don't quite know where, how long it will take, and how much it will cost you. But the fear of missing out is strong.

The hires that we’ve seen work are boring to describe and specific to death. The hires that fail sound impressive in a board update. That’s actually always been true of hiring. AI just made the failure mode more expensive.

This article was written in collaboration with Deconstructor of Fun, our partner for connecting the top 1% of operators with the companies shaping the future of games, tech, and consumer apps.

Miklos Blasko

Written by

Miklos Blasko

Partner

Hiring in this space?

Missing Link runs executive and IC searches for AI, Web3, Frontier Tech and Dev Tooling companies: roles too specific for a generalist agency and too high stakes for a job post.

Start a search

Keep reading