Will AI accelerate or undermine the way humans have always innovated?

In graduate school, my experimental archeology professor had a student create a door socket—the hole in a door frame into which a bolt slides—in a slab of sandstone by pecking at it with a rounded stone. After a few weeks, the student presented his results to the class. “I pecked at the sandstone about 10,000 times,” he said, “and then it broke.”

This type of experience is called individual learning. It works by trial and error, with lots of each. Also known as reinforcement learning, this is how children, chimpanzees, crows and AIs often learn to do something on their own, like make a simple tool or solve a puzzle.

But individual learning has limits. No matter how much a person experiments through trial and error, improvement eventually hits a ceiling. Humans have been throwing javelins for a few hundred thousand years, but their performance has largely plateaued. At the 2024 Olympics in Paris, the gold medal in the javelin throw was about 5% below Jan Železný’s record from 1996. The level of expert play in the strategy game Go was essentially stable from 1950 to 2016, when artificial intelligence changed the equation.

Throughout humanity’s existence, these limits on individual learning have not applied to technology. Since IBM’s Deep Blue defeated world chess champion Garry Kasparov in 1997, supercomputers have become a million times faster – and now routinely outperform humans in chess and many other fields.

Why is technological improvement so different? My work as an anthropologist on cultural evolution and innovation shows that, unlike individual performance, technology advances through combination and collaboration. As more people and ideas connect, the number of possible combinations increases superlinearly. Technological innovation evolves with the number of employees.

My new book with anthropologist Michael J. O’Brien, “Collaboratives Through Time,” reveals these patterns across human existence. It traces how 2 million years of technological traditions have advanced through collaboration between specialists, between generations and with other species.

Expertise was key. Because traditional communities know who their experts are, specialization and collaboration have always underpinned humanity’s success as a species.

I would summarize our vision of how technology continues to advance as TECH: tradition, expertise, collaboration and humanity.

Traditions and know-how – the essential foundation

The longest technological tradition documented by paleoanthropologists was the Acheulean hand axe. This versatile stone tool was made by our hominid ancestors for almost a million years, including approximately 700,000 years at a single site in East Africa. People produced Acheulean tools using techniques they had learned, practiced, and perfected over generations.

Later, small prehistoric societies of modern humans thrived on millennia of specialized knowledge, such as music, thatching roofs, growing seeds, burying corpses in peat bogs, and making millet noodles and even cheese suitable for burying mummies.

As early as 22,000 years ago, communities near the Sea of ​​Galilee stored and used more than a hundred plant species, including medicinal plants. Shamans – ritual experts in medical knowledge and healing – helped their groups survive. Archaeological evidence from burial sites suggests that these specialists were widely revered for thousands of years: a female shaman was buried with turtle shells, the wing of a golden eagle and a severed human foot in a cave in Israel.

Collaboration – knowledge that extends across time and space

Traditional expertise alone does not advance technology. Technological progress occurs when different forms of expertise are combined.

The wheel may have originated in copper mining communities. One expert sourced copper from the Balkans, another transported it, another smelted it. Around 4000 BC, other specialists melted copper into an ancient wheel-shaped amulet: fashioning a model in wax, enclosing it in clay, baking it in a kiln, pouring molten metal into the mold, then breaking the mold.

Transportation technologies have reshaped old commodity networks. As communities in Eurasia and Africa built wheeled vehicles and ships and raised domestic horses and other pack animals, collaboration spread across continents. Sea and land trade linked blacksmiths, scribes, religious scholars, bead makers, silk weavers, and tattoo artists.

Expertise was often distributed between cities and their hinterlands, with cities functioning as hubs in transcontinental product networks. In ancient Egypt, no single community could produce a mummy on its own. Saqqara’s mummification experts relied on a continental network supplying oils, tars and resins, combining these materials with specialized techniques in antisepsis, embalming, wrapping and sealing coffins.

Across the world, states and empires—from the Indus Valley Civilization to the Vikings, Mongols, Mississippians, and Incas—expanded these networks, serving as hubs coordinating the exchange of raw materials, specialized knowledge, and finished products. These exchanges could be very specific: Chinese porcelain was shipped exclusively to the 12th-century palaces of Islamic Spain via Middle Eastern traders who added Arabic inscriptions to gold leaf.

The scale has changed, but not the structure. Today, in a global product space, an iPhone is assembled from a distributed network of expertise and specialized facilities.

Humanity – social learning

Today, AI could disrupt the millennial model of technological progress thanks to TECH. Most major language models generate statistically common responses, which can flatten culture and dilute expertise and originality. Risk increases as untapped high-quality training data – our reservoir of expertise – becomes scarce.

This creates a feedback loop: models heavily trained on low-quality content can degrade over time, with measurable declines in reasoning and understanding. Some scientists now warn that humans and large language models could find themselves locked in a cycle of mutually reinforcing recycled generic content, leading to brain rot for everyone involved. The dystopian extreme is the collapse of the AI ​​model, in which systems heavily trained on their own production begin to produce nonsense.

Brain rot is one reason why some AI pioneers now wonder whether large language models will achieve human-level intelligence. But I think this is not a good direction. The key to continually improving AI models is the same one that has sustained human expertise for millennia: keeping human experts in the loop – the E in TECH. Thanks to a sort of “pipe toe” effect, an informed minority can guide an uninformed majority who copy their neighbors.

In a classic experiment, guppies, following their neighbors, eventually gathered behind a robot fish that guided them toward food. A recent study showed that traffic congestion decreases when autonomous vehicles represent only 5% of cars on the road. In both cases, a small, informed minority reshaped the behavior of the entire system.

Like humans, large language models are social learners and learning can go both ways. Designers can increase the likelihood that models will continue to improve by ensuring that they incorporate the lessons accumulated by human expertise throughout history. In turn, this creates the conditions for individuals and models to learn from each other.

In the 2010s, DeepMind’s AlphaGo rediscovered centuries of accumulated human knowledge of Go through individual learning, then went beyond that by developing strategies that no human had ever played. Human Go masters then adopted these AI-generated strategies into their own game.

Large, well-trained language models can also summarize vast amounts of scientific information, help dissuade people from conspiracy thinking, and even support collaboration itself by helping diverse groups find consensus. In these cases, learning goes both ways.

From Acheulean axes to supercomputers, human innovation has always depended on tradition, expertise, collaboration, and humanity. If AI is adapted to finding and trusting expertise rather than diluting it, it can become humanity’s next great technology – along with ancient writing, markets and early governments – in our long history of collaborators across time.

This article is republished from The Conversation, an independent, nonprofit news organization that brings you trusted facts and analysis to help you make sense of our complex world. It was written by: R. Alexander Bentley, University of Tennessee

Learn more:

R. Alexander Bentley does not work for, consult, own shares in, or receive funding from any company or organization that would benefit from this article, and has disclosed no relevant affiliations beyond his academic appointment.