AI and intelligence ·

Beyond the Faster Horse: AI Doesn't Need Our Ideas

A follow-up to AI Is Not a Tool: Reflections on ICAIH 2026 in Lebanon. First posted as a reply on LinkedIn.

After I published my ICAIH reflections, a fellow panelist pushed back. The comment called AI a machine: a reasoning amplifier made from human language, more than a tool because it proposes and challenges, but not a partner, because a partner shares responsibility for the outcome and AI does not. It called AI creative only by recombining existing ideas, and innovative only through us, because innovation means choosing which idea matters, committing to it and answering for the result. It recalled the line attributed to Henry Ford, that had he asked people what they wanted, they would have said a faster horse, placed AI closer to the customer than to Ford, and allowed that autonomous robots may change that. On the law, the comment is right: authorship still needs a human. Where I'd push back is the faster horse, and the idea that AI needs our ideas.

What most people know as AI is large language models, and so far those are trained on our thoughts. That's the part that looks like the customer asking for a faster horse. But AI isn't only reading what we wrote anymore. It's in the world: robots, cars and drones moving on earth and in space. Robotics researchers like Rolf Pfeifer have long argued that intelligence comes from a body acting in the world, and that moving through it shaped how brains evolved. Now AI has bodies too. It's building its own experience, and the ideas it forms that way won't just be ours.

We've already seen what happens when AI doesn't follow our thinking. In 2016, AlphaGo played Move 37, a move some experts first took for a mistake. AlphaGo estimated a human would play it about 1 in 10,000 times. It won the game. That's not a faster horse. It's a new way of seeing the board. Even on our own text, AI already outscores experts: LLMs predicted the results of neuroscience studies better than neuroscientists did.

AI is also solving science. AlphaFold predicted the structures of virtually all of the 200 million known proteins, a problem biologists had worked on for 50 years, and its creators shared the 2024 Nobel Prize in Chemistry. AI will accelerate discovery and redefine the confines of our understanding.

And on responsibility, the ground is already moving. As Peter H. Diamandis discussed on Moonshots, Argentina's government has proposed "non-human corporations": companies run by AI agents. Under the draft, no human shareholder is required, and such a company could own property and sign contracts. It isn't law yet, it will face pushback, and even its most autonomous format still requires a human who answers for the company's obligations at formation. But whether AI can answer for an outcome is no longer only a philosophical question.

So I wouldn't assume AI needs us for its ideas. We're no longer alone at the apex of intelligence. Right now AI can explore far more possibilities than I can, work through many different thinking models, and process in minutes what would take me ages. I still guide it, and I still play my role. That doesn't make it any less of a partner. It's a thought partner, not a tool.

Michael Joseph is an AI expert and consultant with operations across the United States and EMEA. He is Founder and Chief Strategy Officer of Epirroi and AI Advisor to the President of the Holy Spirit University of Kaslik (USEK).

Sources

Thinking about what AI can do for your organization?

Start a conversationStrategy, governance, and foresight Executive AI trainingFor leaders, boards and faculty

← More insights · AI Is Not a Tool · Next: Three Friends at Dinner