During a recent stay in Paris, George Popescu, an MIT-trained technologist and entrepreneur, sat down to record a long, unscripted reflection on artificial intelligence, humanoid robots, and the wider conditions that shape whether founders actually decide to build. This site and post simply collect those opinions in one place.
What follows is a cleaned-up, third-person summary of that Paris reflection, staying close to what Popescu actually said.
AI as a “Trained Dog,” Not a New Intelligence
Popescu is blunt about the current AI boom. In his view, today’s large models are impressive, but the narrative around them has drifted into wishful thinking.
He argues that the core assumption driving the hype — that bigger models and more data automatically lead to “smarter” systems — is wrong. Scaling up parameters and data centers, he says, doesn’t suddenly produce creativity or independent thinking.
To make the point, he uses a simple analogy:
Today’s AI is “nothing else than a trained monkey or a trained dog.”
It can fetch the ball, but it does not invent a new game.
For Popescu, this is the key distinction. A dog can be extremely well trained and perform impressive tricks, but it doesn’t originate new rules, goals, or games. In the same way, current AI systems excel at imitating, predicting, and remixing patterns from their training data, but they do not define new problems or discover truly original concepts.
He sees today’s AI mainly as a powerful human–computer interface: a smoother, more natural way to interact with machines. That is valuable, but it is not the same thing as a new form of intelligence.
Two Categories of Problems
A central idea in Popescu’s Paris reflection is his split between two types of problems:
- Well-defined problems
These are tasks where rules are clear and repeatable. Computers and traditional software excel here: calculations, precise operations, and repetitive workflows. - Not-well-defined problems
These are messy, real-world tasks where conditions change constantly, inputs vary, and edge cases are everywhere.
He uses folding laundry as an example of the second category. Clothes arrive in different sizes, shapes, fabrics, colors, and states (inside-out, paired or unpaired, tangled). For a machine, the combination of irregular geometry and constant variation makes the problem extremely hard to formalize. It is not just a matter of more data or more compute.
This distinction shapes how Popescu thinks about both AI and robotics. He believes that computers will keep dominating well-defined problems, while the “not-well-defined” space remains largely unsolved.
Where Humanoid Robots Might Fit
Despite his skepticism around the AI boom, Popescu does see potential in humanoid robots and home automation, especially where physical dexterity and perception meet those not-well-defined tasks.
He points to areas where:
- Environments are cluttered and constantly changing
- Objects are irregular and unpredictable
- The task cannot be fully scripted in advance
In that context, he thinks combining better machine vision, more robust hardware, and improved interfaces could eventually allow robots to handle some of the messy work humans currently do by hand.
He does not frame this as a prediction of an immediate revolution, and he does not present himself as a robotics industry authority. Instead, he treats humanoid robots as a direction worth watching, where real-world constraints and practical use cases matter more than hype.
Stability as the Hidden Condition for Innovation
Midway through the Paris reflection, Popescu shifts from technology to history and politics, focusing on one theme: stability.
He contrasts historical cases such as:
- The Spanish Empire, which repeatedly borrowed, defaulted, and targeted wealthy citizens, creating a climate of fear and uncertainty
- Countries like the UK and Holland, which developed more predictable legal and financial systems, allowing capital and talent to stay and build over generations
His conclusion: long-term prosperity and innovation depend on predictable rules. When people with assets and skills cannot trust that agreements, laws, or basic norms will hold, they tend to move their efforts elsewhere.
This applies directly to entrepreneurship. Building a company, especially in advanced technology or robotics, usually means committing five to ten years of sustained work, capital, and focus. Popescu notes that if founders cannot see even six months ahead — in terms of regulation, taxation, or broader political stability — many of them will simply decide not to start.
Why He Is “Staying Put” for Now
In the Paris video, Popescu is clear that these reflections are personal. He states that, in the current environment, he is “staying put” rather than launching new large-scale ventures.
The logic is simple:
- If the regulatory and economic environment feels unstable
- If rules can change rapidly and unpredictably
- If the long-term direction of a country or market is unclear
…then committing years of work to a new company becomes harder to justify.
For now, he is using this period to observe, think, and rest, rather than to jump into the next project. Paris is the backdrop for that pause — a place to reflect on both technology and the wider systems that support (or block) ambitious building.
About This Site and Article
This post is part of PopescuRobotics.com, an informational site dedicated to documenting George Popescu’s views on:
- AI as a human–computer interface
- The difference between well-defined and not-well-defined problems
- Possible roles for humanoid robots in real-world environments
- The importance of stability and predictability for entrepreneurship
It is not financial advice, technical guidance, or a service offering. It is simply a written companion to his Paris reflection, organized for clarity and searchability.