The Stuff AI Can’t Take From You

Palantir’s Alex Karp recently said something deliberately spicy: in the AI era, the people who’ll thrive are either those with hands-on vocational skills or those who think a bit differently.

The headline is built to grab attention, but underneath it there’s a point worth sitting with.

Trades stay relevant because they’re grounded in the real world. They’re physical, situational, human. AI can summarise your meeting, but it can’t rewire your house. It can label your entities, but it won’t shimmy under the sink to sort a leak.

And on the neurodivergent side, Karp’s not claiming some magical advantage. He’s talking about the mindset that sometimes shows up alongside it: seeing patterns others miss, taking leaps that don’t look sensible on paper, building the thing no one else thought to build. As someone who talks openly about ADHD, hyperfocus, and the non-linear way my brain works, that bit resonated. You can’t template that. There’s no connector for it.

But then you’ve got the counterargument from Microsoft’s chief scientist and Anthropic’s Daniela Amodei: critical thinking, adaptability, communication. The ability to reason, to explain, to stay curious. Basically: the liberal arts aren’t dead; they’re probably more important now.

Across all of this, the real takeaway isn’t “which group wins”. It's simpler:

Whatever you bring to the table needs to be genuinely hard to automate.

When I think about the Power Platform, this lands hard. Most of the work that matters isn’t the flow itself. It’s the empathy in understanding what someone is really asking for. It’s the ability to translate messy human needs into something structured. It’s choosing clarity over cleverness. It’s telling a stressed-out client: “Don’t worry, we can sort this,” and then delivering.

AI is brilliant at acceleration. It’s not brilliant at care, intuition, or context.

This is why I still believe the most important skills in our world are:

• Asking better questions

  
  

  A good solution starts with understanding the human, not the system.

  
  

• Seeing the edge cases

  
  

  Neurodivergence often helps here: noticing the weird angles others miss.

  
  

• Keeping things simple

  
  

  Complexity looks clever until you have to maintain it.

• Working with people, not just tech

  
  

  Automations don’t stick unless the humans using them feel heard.

  
  

• Telling the truth kindly

  
  

  Whether it’s about timelines, scope, or what the platform can’t do.

  
  

AI will eventually write flows, plugins, expressions, documentation… but it won’t replace the ability to connect dots that aren’t obviously connected. It won’t replace the emotional labour of helping someone go from stuck to confident. It won’t replace the creativity that comes from being wired a bit differently.

For those of us in tech, especially in automation, the question is no longer “Will AI take my job?” It’s:

What part of my work is deeply, unmistakably human — and am I leaning into that enough?