Thought Piece: The most important AI skill

Hi there,

The tech industry has spent the last year chasing the wrong skills. The obsession with "prompt engineering" assumes that clever commands are the key to good AI performance. A new paper studying over 600 users shows this is a myth.

The study found almost no correlation between how well someone performs a task alone and how well they perform with an AI assistant. The one skill that does predict high performance is Theory of Mind. It is the ability to model another agent's beliefs, goals, and blind spots.

High-performers treat AI less like a search box and more like a non-human collaborator. They have a good sense of what the AI knows, what it is likely to misunderstand, and how to guide it. They are not just giving instructions; they are building a partnership.

This reveals an uncomfortable truth: most companies are treating AI like the same old software that has failed them for years. They see it as a rigid tool to be operated, not a system to collaborate with. This is the same broken thinking that leads to a 70% failure rate for digital transformation projects.

Here's how to get it right

1. Hire for cognitive empathy, not technical tricks. The most valuable people will not be the ones who memorize prompt formulas. They will be the ones who can anticipate the AI's "thinking" and steer it. It is about knowing when to trust the AI, when to question it, and how to frame a problem so it can actually help.

   

   There is a direct link between Theory of Mind and performance. The higher a person's ability to understand the AI's perspective, the more effective they are as a collaborator.

2. Make your team better, not just faster. The study shows the biggest gains come from pairing human expertise with AI to create something new. This confirms a core belief at Future Works: technology should adapt to people, not the other way around. At Maersk, they did not replace their best ship captains. They built AI that learned from their expertise, scaling that wisdom across the entire fleet and cutting fuel costs by 15%. This creates a competitive advantage that is hard to copy.

3. Treat AI as a people problem, not a technology problem. A new AI tool is useless if it doesn’t fit into the messy reality of your team’s workflow. This is where most initiatives fall apart. They focus on the technology while ignoring the human systems. If you treat AI as just another tech rollout, it will end up on the shelf with all the other expensive, unused software.

The companies that win will not be the ones that buy the fanciest models. They will be the ones that master the human-AI partnership. They will build a culture of collaboration, not just command. Being AI-native means mastering the partnership, not just buying the tools.

Much love,
Matt

At Lighthouse, we love featuring fresh perspectives from our community of AI, tech, and innovation leaders. Got insights to share? Just reply to this email—I’d love to hear from you!