Can AI Break Humanity's 2-Million-Year Collaboration Code?

10,000 strikes with a stone against sandstone, and the rock simply broke. That's what happened when a graduate student tried to create a door socket using individual trial-and-error learning. It's a perfect example of why going it alone has limits.

Here's what's puzzling: Humans have been throwing javelins for hundreds of thousands of years, yet performance has barely improved. The 2024 Olympic gold medal throw was still 5% short of Jan Železný's 1996 record. Go masters showed essentially flat performance from 1950 to 2016 – until AI changed everything.

Meanwhile, technology keeps accelerating. Since IBM's Deep Blue beat chess champion Garry Kasparov in 1997, supercomputers have become a million times faster. Why do individual skills plateau while technology soars?

The 2-Million-Year Secret

Anthropologist R. Alexander Bentley has an answer: TECH. That's Tradition, Expertise, Collaboration, and Humanity working together.

The longest technological tradition on record was the Acheulean hand axe – used by our ancestors for almost a million years. At one site in eastern Africa, people made these tools for 700,000 years straight. This wasn't individual innovation; it was accumulated expertise passed down through generations.

By 22,000 years ago, communities near the Sea of Galilee stored and used over a hundred plant species, including medicinal ones. Shamans – ritual experts in medical knowledge – helped their groups survive. Archaeological evidence shows these specialists were revered for thousands of years. One shaman woman was buried with tortoise shells, a golden eagle wing, and a severed human foot.

From Wheels to iPhones

But expertise alone doesn't advance technology. Progress happens when different forms of expertise combine.

The wheel likely emerged from copper-mining communities. By 4000 B.C., multiple specialists collaborated to create wheel-shaped amulets: one expert sourced copper from the Balkans, another transported it, another smelted it. Additional specialists then shaped wax models, encased them in clay, fired kilns, poured molten metal, and broke away molds.

Ancient Egyptian mummification required continental networks. No single community could produce a mummy alone. Experts at Saqqara drew on suppliers across Africa for oils, tars, and resins, combining these with specialized techniques of preservation, embalming, wrapping, and coffin sealing.

Today's iPhone follows the same pattern – assembled from a distributed network of specialized expertise and facilities worldwide.

AI Disrupts the Formula

Now AI threatens this 2-million-year collaboration model. Most large language models generate statistically common responses, flattening culture and diluting expertise and originality.

The bigger risk? High-quality training data – our reservoir of human expertise – is becoming scarce. Models trained on low-quality content show measurable declines in reasoning and comprehension over time.

We risk a feedback loop where humans and AI systems become locked in recycling generic content. The dystopian extreme is AI model collapse – systems trained heavily on their own output begin producing nonsense.

Experts Still Matter

But there's hope. The key is keeping human experts in the loop – the 'E' in TECH.

Consider the guppy experiment: Fish following their neighbors ended up schooling behind a robotic fish that guided them toward food. Recent studies show traffic congestion eases when autonomous vehicles make up just 5% of cars on the road. In both cases, a small, informed minority reshaped the whole system's behavior.

Large language models are social learners too. DeepMind's AlphaGo rediscovered centuries of human Go knowledge through individual learning, then crafted strategies no human had ever played. Human masters subsequently adopted these AI-generated strategies into their own play.

Well-trained models can summarize vast scientific information, help people escape conspiracy thinking, and support collaboration by helping diverse groups find consensus. The learning flows both ways.

The Collaboration Continues

From Acheulean hand axes to supercomputers, human innovation has always depended on tradition, expertise, collaboration, and humanity. States and empires – from the Indus Valley to Vikings, Mongols, and Incas – served as hubs coordinating exchanges of materials, knowledge, and products.

The scale has changed, but the structure hasn't. Today's global product networks still rely on distributed expertise. Chinese porcelain once shipped exclusively to 12th-century Islamic Spanish palaces via Middle Eastern traders who added Arabic gold leaf inscriptions. Now we have iPhones assembled from components made across continents.