To help predict outcomes on everything from the presidential election to the next Super Bowl champion, a San Luis Obispo-based researcher has looked to bees for inspiration.
Former Cal Poly professor Louis Rosenberg calls his system “swarm intelligence” — it’s similar to the way bees form a swarm to investigate potential locations for a new hive and communicate their findings to the larger bee population.
The web-based, real-time human swarm application developed by Rosenberg’s company, Unanimous A.I., forms groups of people to weigh in on a variety of topics, including sporting events, politics, movies and technology.
The idea is that a group of people is smarter than people can be on their own.
Louis Rosenberg, founder of Unanimous A.I.
The group, which typically includes 30 to 150 people, answers a question or series of questions through the application. The system then compiles the collective intelligence of groups to determine the most informed responses on certain topics.
Questions might include whether Bernie Sanders is more trustworthy than Hillary Clinton or whether Melania Trump’s speech intentionally plagiarized a previous speech by Michelle Obama.
When a beehive becomes overpopulated, a group of bees will break off and form a new swarm with a new queen. They then begin the search for a new hive location.
Rosenberg said that after a group of scouts collects information about a potential hive site, the bees return to the larger group and perform a waggle dance to express their enthusiasm — or relative lack thereof — about the site they just visited. The scouts with the most convincing waggle sway the group’s decision on a new home, which can be a life-or-death choice.
“The idea is that a group of people is smarter than people can be on their own,” Rosenberg said. “Like bees, with collective knowledge and wisdom, people can work together to make much wiser decisions.”
His company, which has fewer than 50 employees, recently won an award for the best web or phone app consumer application from the Global Annual Achievement Awards for Artificial Intelligence.
The public can use the app for free, and there are more than 100,000 registered users who have signed up to participate in surveys.
The company also offers paid services to professional companies, using its “swarm” system to provide insights on topics such as medical decisions and market research.
News outlets such as the Washington Post and Newsweek have used Unanimous A.I. to gauge responses to public interest topics, including user reaction to the latest iPhone.
CBS and TechRepublic asked Unanimous A.I. to predict the winners of the 2016 Kentucky Derby. One swarm of 20 people (self-identified as horse racing enthusiasts) picked the first four finishers in order at 540-1 odds.
In another instance, a five-week study published with researchers at Oxford University showed that swarms of Premier League fans picked the winners for English soccer games with 30 percent higher accuracy than individual fans, Rosenberg said.
540-1odds to picking the first four finishers of the 2016 Kentucky Derby, achieved by an A.I. swarm
Additionally, an AI swarm picked the 2016 World Series teams and the Chicago Cubs as the champion months in advance for the Boston Globe.
Unanimous A.I. participants collectively scroll along a virtual honeycomb-shaped grid to select one of six possible answers to a question. Each user in the group has a U-shaped magnet and tries to pull a puck-shaped object to the answer they believe to be correct within 60 seconds. Ultimately, the collective interests of the group win out.
In some cases, participants can recommend answer choices in an open-ended format.
The social program typically relies on an average person with a topical interest, not experts, to achieve results. Users can also start discussion groups to get a response to a question.
The professional component uses experts for technical inquiries.
Rosenberg, a former education technology and engineering professor at Cal Poly, started his business in 2014 and launched it online in June 2016.
As artificial intelligence is increasingly prevalent, Rosenberg said it’s important for human influence to remain a part of the decision-making process. He believes his program achieves that.
“Human emotions, values, interests and our collective interests in humanity help guide the decision-making rather than replacing us with something that might come to a logical decision, but one that doesn’t necessarily take into account human sensibilities,” he said.
“An artificial intelligence program’s interest may not always be aligned with human interests, and that can be dangerous.”