Jeff Bezos tweeted in June that he wanted short-term, high-impact ways to do social good.

An artificial intelligence platform put 46,000 suggestions to a vote, using a method inspired by how bees swarm.

Universal access to clean drinking water was the winner.


In mid-June, Amazon CEO Jeff Bezos made an open request on Twitter for ways he could dip his toes in the philanthropy world. More than 47,000 people responded, suggesting ideas from dog training for PTSD victims to improving education in the developing world.

But Louis Rosenberg says he has overwhelming evidence that universal access to drinking water is the real genius idea.

Rosenberg is the CEO and founder of Unanimous AI, an artificial intelligence platform that purports to make super-intelligent decisions based on the wisdom of the crowd. The company takes its cues from the animal kingdom, in which birds form flocks, fish form schools, and bees form swarms, all in effort to make smarter choices to survive, Rosenberg said.

“They’re literally smarter together when they converge as a system on answers,” Rosenberg told Business Insider. Unanimous AI mimics this model by creating online “swarms” where people can tackle any question thrown at them and use their collective wisdom to converge on a single answer.

When the company has held these swarms, the group has correctly predicted the winners of the 2015 Oscars, the first four horses of the 2016 Kentucky Derby in 2016, and the eight teams that would make it to the 2016 MLB playoffs, including the Chicago Cubs’ victory.

Rosenberg wanted to see how a swarm might respond to Bezos’ call for philanthropic ideas. The company pulled each of the then-46,000 replies and distilled them to 200 broad categories of ideas. Then it asked 300 users (or “scouts”) to rank the 200 ideas on a scale of 1 to 10. In nature, scouts are bees that go searching for new homes and present their ideas to the hive. Scores of 7.0 or higher were included for final consideration.

Groups of roughly 100 people went through a series of swarms to eliminate solutions it saw as poor, ultimately arriving at six big ideas: universal access to clean water, cancer treatment assistance, health clinics for the poor, essential equipment for rural hospitals, free medicine for the poor, and mobile health clinics.

During the live deliberation, some people changed their answers while others held firm, Rosenberg said. Ultimately, about a minute went by before the swarm converged on access to clean drinking water.

Rosenberg admitted that there are no guarantees the swarm’s decision is necessarily the best, and he doesn’t have empirical studies to bolster swarm decision making as the ideal approach for business decisions. But what he does have is the company’s track record and the confidence that nature has found a method humans can implement themselves.

He said there is also the element of repeatability. When his company performs the same swarms over and over, it tends to see the same results emerge. This suggests something inherently virtuous about the power of the swarm over the individual, Rosenberg said.

“This really wasn’t close,” he said. “There was a very, very strong sentiment in the swarm intelligence that universal access to clean drinking water was the issue that was the best use of the funds and you could have the biggest impact and the most long-lasting impact and also the most immediate impact.”