Innovation

How to use Swarm AI instead of polls for market research

With Unanimous A.I.'s new 'Swarm Insight' platform, your company can harness AI to make group predictions and gain insights that can help you make business decisions and understand your customers.

Image: iStockphoto.com/DigtialStorm


In May 2016, TechRepublic challenged a startup called Unanimous A.I. to predict what some thought would be impossible: The superfecta at the Kentucky Derby.

Hardly anyone, including Louis Rosenberg, CEO of Unanimous A.I., thought this would actually work—but he accepted the challenge, creating an artificial "swarm" through an AI-based platform called UNU that picked the top four horses, in order, at the 2016 Derby.

The swarm consisted of a group of 20 people with some knowledge of horse racing, chosen anonymously, who participated on the UNU platform. The model, based loosely on the concept of nature's swarms—How do honeybees decide where to migrate to?—incorporated a kind of group intelligence, a collective decision.

The result? The swarm correctly predicted the superfecta, beating 540-1 odds.

How did it do it? Unanimous A.I.'s platform gathers individuals, on computer screens and tablets around the world, to make a real-time decision. The user interface is simple, presenting each participant with options to answer a question, such as "Who will come in 2nd place at the Kentucky Derby?" Users control what looks like a magnet, and physically drag the icon towards a preferred answer, using a mouse or touchpad. Instead of simply stating a preference, the platform allows users to shift their choice, as well as their confidence in that choice, leading to a group prediction.

The tool has proven remarkably accurate, frequently outperforming polls and experts. When a group of 50 people predicted the 2016 Oscars, for instance, individuals were 44% accurate—but as a group, the accuracy ballooned to 76%. Professional critics, in contrast, were 63% accurate.

Need more proof of the power of the swarm? In the middle of the 2016 Major League Baseball season, 25 baseball fans predicted that the Chicago Cubs would win the world series, breaking a century-long losing streak.

So what does this amplified intelligence mean for business? As it turns out, it can offer a nuanced look at customer behavior.

A new tool from Unanimous A.I. called Swarm Insight offers an "intelligence service" that can turn business teams into intelligent systems, according to the company. On top of that, it can turn consumers into "insight engines."

Polls just offer the popular answer, said Unanimous CEO Louis Rosenberg. And focus groups simply tell you "who's the loudest voice in the room," he said. Rosenberg noted that the current methods haven't changed in 100 years. Also, they're slow, and often costly.

SEE: TechRepublic's 'swarm AI' predicts tight election, gives edge to Clinton (TechRepublic)

In a swarm, by contrast, the group is diverse, and converges on an answer—often in under a minute.

Companies that want to learn more about what their customers want and tailor an effective marketing message can learn a lot from a swarm, such as how consumers respond to products, messages, and advertisements. The swarm can also tap into intelligence within a team—accessing insight into what restaurant managers across a certain fast food chain know about customer behavior, for instance. These swarms can be randomized, or can be drawn from specific groups of people.

"They're smarter than the the people sitting in the executive office," said Rosenberg.

Swarm Insight could be particularly useful to companies that want to test advertisements before they run. How convincing is the message? How will people react? It could save a larger company that wants to take a risk from making a choice that fails, Rosenberg said.

The tool can also help with pricing. Imagine your team of 34 sales reps is posed with a dilemma. How much should the suggested retail price be for earbuds?

screen-shot-2017-01-24-at-1-35-20-pm.jpg
Image: Unanimous A.I.

Swarm Insight can help make that decision. In a real example, the price point landed at $8.35. But beyond the actual answer lies much more information that can be useful to businesses. Using this tool, your team can gain important insights on how customers arrived at the decision. Since the prediction was made dynamically, in real-time, it's possible to dig into what happened during the entire process. The inflection analysis shows the point at which the price became "too low" and users moved their magnets in the other direction. You can see the migration of the group. Which users did not budge? And who was flexible? Of the flexible ones, where did they move to?

This knowledge can be further distilled by cross-referencing results with demographic information.

Why is this novel? Instead of looking at individuals as data points, the system is interested in each person as a "data processor," giving a broader view of the entire decision-making process.

Companies can currently choose to work with Unanimous A.I. on "engagements," in which they can specify a certain question and get sets of predictions from a swarm, or can purchase a subscription model, which Rosenberg thinks will be more popular.

So far, reactions have been positive.

A company called Smart Lunches, that provides custom lunches for children, is using Swarm Insight, which helped them "get a deep understanding of how parents value and select different elements of a school lunch in a matter of days," said CEO David Morris."That level of insight would have taken far longer using traditional methods, if it could have been captured at all."

And the advertising agency Walton Isaacson sees the tool as one of the "most interesting" applications of AI in marketing because it "gives the CMO of any brand the ability to amplify the intelligence of their customers and their field organization to get deep insights, fast," said Jeff Davidoff, president of Walton Isaacson.

Ultimately, said Rosenberg, it's not the swarm's answer that's most interesting. "It's about how the swarm reached the answer," he said.

The technology used in Swarm Insight has not yet been fully proven in business settings, and the cost of Swarm Insight was not revealed, it does seem to offer an interesting alternative to polling and focus groups for businesses—one that TechRepublic will continue to cover.

The 3 big takeaways for TechRepublic readers

  1. An AI system called UNU, created by Unanimous A.I., brings humans together on a digital platform to make real-time, group decisions.
  2. A new offering from Unanimous A.I., called "Swarm Insight," can be used by businesses that want to make decisions and gain insight into customer behavior.
  3. Swarm Insight looks at how the entire decision-making process unfolds—including analysis of how individuals changed their decision, and conviction of each decision. This information can also be cross-referenced with demographic data.

Also see

About Hope Reese

Hope Reese is a Staff Writer for TechRepublic. She covers the intersection of technology and society, examining the people and ideas that transform how we live today.

Editor's Picks