Introduction
We have always built things in our own image. The ancient Greeks carved gods that looked like idealized humans. Renaissance architects designed buildings proportioned to the human body. And for the past several decades, the most ambitious project in technology has been organized around a single goal: build a machine that thinks like a person.
It is a natural instinct. It is also, I believe, one of the most common traps in innovation.
I spend a lot of time with leaders who are trying to build the next great thing, whether that is a product, a team, or a company. And the pattern I see over and over is the same: they start by studying whatever version of success looks most impressive and try to build their own version of it. The startup that wants to be “the Uber of” their industry. The product team that copies Apple’s design language without understanding the decade of supply chain innovation underneath it. The results are often something that looks right on the surface but lacks the underlying architecture that makes the original work.
Why the Obvious Blueprint Is Not Always the Right One
This is a creativity problem, and it runs deeper than most people realize.
When you copy an impressive output, you are skipping the hardest and most important part of innovation: understanding the foundational system that produced it. You see the polished product launch and try to replicate it, when the real competitive advantage was in the unglamorous decisions about culture and process that happened years earlier.
I learned this as a jazz musician. The great players I studied were not great because they memorized impressive solos by other artists. They were great because they spent years learning scales, chord structures, and harmonic theory, the fundamentals that gave them the vocabulary to improvise something genuinely new. The musicians who just copied solos note for note could play a convincing cover, but they could never surprise you.
The same principle shows up in an unexpected place: artificial intelligence research. A white paper by 27 leading AI and neuroscience researchers, including Turing Award winners Yann LeCun and Yoshua Bengio, recently made a version of this same argument. The AI industry has spent decades trying to build machines that replicate the most impressive features of human intelligence: language and strategic game-playing. And yet, as the researchers note, “a growing number of AI researchers doubt that merely scaling up current approaches will overcome these limitations.”1 These systems can hold a conversation and win at chess, but they cannot walk across a room, pick up an object, or navigate an unfamiliar environment.
The researchers propose what they call the “embodied Turing test,” a benchmark that measures AI not against human conversation but against the sensorimotor abilities that animals have refined over 500 million years of evolution.1 Their argument is that the flashy output (human-level language) was actually the easier engineering problem, while the foundational capability (interacting with the physical world) is the larger problem. The AI field had been so focused on building in the image of our most visible intelligence that it is still just scratching the surface of the deeper architecture underneath.
Finding Your Own Blueprint
The lesson I draw from this is not just about AI. It is about how we approach any creative challenge.
The most innovative leaders I know resist the urge to build in someone else’s image. Instead of asking “who should we copy?” they ask a different question: what is the foundational capability that everything else depends on, and how do we get better at it than anyone else?
That question forces you to look in uncomfortable places. It means studying the boring stuff, like internal processes and communication patterns. It means drawing inspiration from sources outside your industry, from a scientific discipline that seems unrelated, or from the way a five-person team solved a problem that a five-hundred-person organization could not.
The AI researchers in that white paper found their unexpected source in animal neuroscience. The cat’s visual cortex inspired the neural networks behind modern image recognition. Animal reward systems inspired reinforcement learning. The biggest leaps forward in AI have historically come not from copying the most impressive human outputs but from studying how brains actually compute, all brains, across species. And the researchers believe the next breakthrough will come from going back to those fundamentals rather than continuing to scale up the current approach.1
In my work with organizations, the pattern is similar. The companies that sustain innovation over time invest in their foundational capabilities, even when those capabilities are invisible to the outside world. They do not try to be the next Apple or the next Amazon. They try to be the best version of themselves, built on a foundation that nobody else thought to study.
If you are working on something that matters, resist the instinct to build in someone else’s image. Find your own blueprint. It will probably come from somewhere you did not expect to look.
Frequently Asked Questions
Q: What does “building in our own image” mean in the context of AI?
The AI industry has largely organized around replicating human-specific abilities, especially language and strategic reasoning. A recent white paper by 27 leading researchers argues that this approach has reached its limits and that the next breakthroughs will come from studying the sensorimotor abilities shared by all animals, capabilities refined over 500 million years of evolution.1
Q: What is the embodied Turing test?
It is a proposed benchmark for AI that goes beyond conversation. Instead of asking whether a machine can fool a human in a text exchange, it asks whether an AI can control a virtual animal well enough that its behavior is indistinguishable from the real thing. The test focuses on physical interaction with the world, the area where current AI systems lag furthest behind biological organisms.1
Q: How does this principle apply to business innovation?
The pattern is the same across domains: copying the most visible output of a successful system is less effective than studying the foundational capabilities that make it work. Companies that try to “be the next [admired brand]” usually replicate the surface without understanding the underlying system. The most durable innovations come from investing in unglamorous fundamentals that competitors overlook.
Q: What are some examples of unexpected sources driving AI breakthroughs?
Several foundational AI technologies came directly from studying animal brains. Deep convolutional networks were inspired by research on how the cat brain processes visual information. Reinforcement learning grew from studies on how animals learn from rewards. Even the basic computer architecture underlying modern technology traces back to John von Neumann’s work, which drew explicitly on early neuroscience.1
Q: Where should a leader start if they want to find their own blueprint?
Start by identifying the foundational capabilities your organization depends on, the processes and skills that are invisible to the outside world but essential to everything you do. Ask where you are copying someone else’s approach out of habit rather than developing your own. And look for inspiration outside your usual field. Some of the most powerful ideas come from disciplines you would never think to study.
Citations:1 Zador, A., Escola, S., Richards, B., Ölveczky, B., Bengio, Y., Boahen, K., Botvinick, M., Chklovskii, D., Churchland, A., Clopath, C., DiCarlo, J., Ganguli, S., Hawkins, J., Koerding, K., Koulakov, A., LeCun, Y., Lillicrap, T., Marblestone, A., Olshausen, B., Pouget, A., Savin, C., Sejnowski, T., Simoncelli, E., Solla, S., Sussillo, D., Tolias, A. S., & Tsao, D. (2022). Toward Next-Generation Artificial Intelligence: Catalyzing the NeuroAI Revolution. arXiv:2210.08340.