Who do you hire when the machines move in?
Read a job ad with me. It isn’t a real one. I’ve pieced it together from many I’ve come across in recent months. But you’ll recognize it. “Head of AI. Your role: you drive the AI transformation across the entire company. Your profile: STEM degree with top grades, deep knowledge of machine learning, ideally a PhD. Strong communication skills round out your profile.”
That last sentence is my favorite. The ability to bring people along rounds things out. Like a garnish.
I understand the thinking behind it. It’s the obvious one. Many people carry the idea around: AI is software, software is technology, technology is hard.
But I think that falls short.
The numbers point somewhere else
Prosci asked 1,107 professionals and managers about the biggest hurdles to adopting AI. The result: 63% of the hurdles are human in nature, and 38% alone come down to a lack of user skills and training. Technical problems? 16%. It’s a vendor study, and that needs saying. But the order of magnitude matches what I know from many conversations.
And if nearly two thirds of the hurdles are human, the job profile really ought to reflect that.
Two needs you have to keep apart
One is in-house development. Your own models, your own data, deep integration. That takes technical depth, people who have done it before. The catch: anyone who can really do this gets bought up by the market. Europe trains these people and loses many of them to the big labs, mostly in the US. Post that job and you’re competing with San Francisco.
The other is bringing existing AI tools into the company and making them work. That takes an IT perspective. But existing structures can often cover it.
The real problem isn’t in any job ad
In many organizations, people simply don’t want to. There’s resentment: employees who reject AI, ignore it, work around it. And there’s the other side, those who have been using it for ages, just quietly. The keyword is shadow use. I hear both across industries and professions. Not from one company. From many. And both fall under “AI transformation.” But we almost never look for anyone to handle them in the job profiles.
We look for software skills. We look for ML knowledge. We look for degrees. We rarely look for someone who can truly bring an organization along. Understandable. An adoption score isn’t as tangible as a finished, usable result. Only, it should matter to whoever is supposed to use that result in the end.
The question worth asking
Technical skills can be built within a team, up to a point. Cultural change can’t be bought in later. It’s slower, harder to measure, and it doesn’t appear on any résumé as a certificate.
Which is exactly why the question is worth asking before the next position goes out. Who do you really need?