the ConversAItion: Season 2 Episode 8

Making Groups Smarter with AI

Jim is joined by Louis Rosenberg, prolific inventor and the founder and CEO of Unanimous AI, a company that uses AI to amplify the intelligence of human groups. They discuss how an approach called swarm intelligence can be used to forecast everything from the Kentucky Derby to medical diagnoses, with surprising accuracy.

Listen on Apple Podcasts badge
“Inside human heads, there’s massive amounts of data. People have knowledge, wisdom, insight and intuition that’s really powerful, and it's being ignored by AI. And the amazing thing about people is that we're continuously updating our database—we’re out in the world seeing, collecting, and analyzing. At Unanimous AI, instead of performing AI on static data, we [leverage real-time data from] groups of people.”
Louis Rosenberg

About Louis Rosenberg

Dr. Louis Rosenberg is a prolific inventor and the founder and CEO at Unanimous AI. In addition to founding numerous technology companies, including Immersion Corporation, Microscribe and Outland Research, Dr. Rosenberg also served as a tenured professor at California Polytechnic State University. He can be found on LinkedIn here, and his Wikipedia page can be found here.

Short on time? Here are 5 quick takeaways:

  1. We know AI can augment the performance of individuals. Inspired by nature, guest Dr. Louis Rosenberg set out to explore how AI can improve the performance of groups.

    In 2014, Louis founded Unanimous AI, a company built on the idea that humans are smarter together. Other species, like bees and fish, have evolved to create “swarm intelligence,” enabling groups to be smarter than any individual. Though humans didn’t experience this evolution, Dr. Rosenberg believes AI can help facilitate swarm intelligence for humans.

  2. By tapping into the “database” of human minds, it’s possible to make predictions without an expansive static dataset.

    Traditional AI is most useful when applied to massive amounts of data; but for many future events, huge datasets don’t exist or are constantly evolving. Unanimous AI solves this problem by matching datasets with the information that exists inside of people’s heads. The company’s powerful combination of AI and human group intelligence can be applied in a variety of scenarios—from sales forecasts to Oscar winners.

  3. In 2016, Unanimous AI made headlines for predicting the horses that came in first, second, third and fourth at the Kentucky Derby.

    How? With the help of 20 horse racing enthusiasts. The participants logged on to the Swarm platform at exactly the same time, and indicated which horse they believed would finish in first, second, third and fourth. They did so by tugging a virtual “magnet” towards their predicted winner, pushing and pulling against each other, like when playing with a Ouija board. Meanwhile, AI algorithms consider each participant’s behavior and determine how the group should converge on an answer.

    As individuals, not a single one of the 20 participants correctly predicted the top four horses in order. As a group, it took only five minutes to combine their knowledge in an optimal way to predict the outcome with complete accuracy, and against 540 to one odds.

  4. Swarm intelligence can even be used to make doctors smarter.

    Unanimous AI performed a study in partnership with Stanford Medical School, in which small groups of four or five radiologists would examine an X-ray on their screen and make the diagnosis together as a swarm. The study found that when they worked together, they were 33% more accurate than when they did it as individuals.

  5. In the long term, swarming could be used to enhance political and societal decision-making.

    It’s easy to test a problem where there’s a right or wrong answer, but when you have issues where people have opposing views, it’s far more difficult to reach a decision that satisfies the entire population. In politics, for example, the traditional way to make a decision is by polling, which only reveals differences in opinion. A swarm, on the other hand, reveals the commonality in a group.

    A recent Unanimous AI study on Brexit found that when a group was asked to just take a vote, they would never reach a unanimous decision. However, when they worked together as a swarm, they would converge on solutions that would maximize their collective satisfaction. Dr. Rosenberg hopes Unanimous AI can help society make tough decisions down the line. 

Check out more episodes of The ConversAItion.