Hope Reese published Unanimous AI’s prediction the day before the race and turned $1 into $540.

A horse named Nyquist crossed the line first, followed by Exaggerator, then Gun Runner, and then Mohaymen. It was the exact sequence the researchers had predicted.

In gambling parlance, predicting such a result is called a superfecta bet — and it pays big. But despite bookmakers setting the odds of such an outcome at 540 to one, Dr Louis Rosenberg and his team felt the odds should have been more accurate.

“The odds could have been in the order of five to one, not 540 to one,” he told news.com.au.

Far from being an exercise in self interest, the company passed its predictions on to tech journalist Hope Reese who published them the day before the race, and later boasted on Twitter after she placed a single dollar on the superfecta to win $540.

As for Rosenberg, he turned $20 into a cool $10,840.

But how did he and his team come to this conclusion, and ultimately predict the lucrative superfecta? The answer: By using a technique known as swarm intelligence.

Louis Rosenberg is the CEO of Unanimous AI, a US company that is experimenting with creative platforms to allow researchers to harness the swarm intelligence of human groups.

He believes the phenomenon will prove to be a “a major disruption for the field of gambling” but that’s not what he and his team are necessarily interested in. Instead they view the technology as a vital way to birth a benevolent form of artificial intelligence and make social and political decisions “that optimise the outcome for the species as a whole,” he said.

So how does it work?

At its most simplest, swarm intelligence can be described as way to crowdsource people’s knowledge in a synchronised way — to harness their collective intelligence. The idea is that the various units of the group will influence each other to arrive at the most accurate or probable answer.

“To me, the best way to explain it is to point out that Swarm Intelligence goes back to the birds and the bees,” Rosenberg said. “Across countless species, nature shows us that social creatures, when working together as a unified system, can outperform the vast majority of individual members when solving problems and making decisions, thereby boosting their overall survival.”

Prior to the Kentucky Derby, the company engaged 20 volunteers from around the country who identified as avid horse racing fans and got them to participate in a simple online task to pick the winner of the race. The participants logged into a program created by Unanimous AI called UNU and were asked a question at the exact same time.

“The users then worked in synchrony to control a graphical puck and answer the questions, forming a system that converged upon answers that optimised their group conviction,” Rosenberg said.

The image below shows that process unfold.

The volunteers reconvened a second time after a change was made to the race’s line-up and one of the horses in the superfecta was swapped out for another.

While the Swarm AI got the picks perfect, not a single individual who participated in the swarm got the picks right on their own, a fact that demonstrates the potential utility of the collective approach.

It’s by no means a foolproof system (Rosenberg said he was “speechless” when the superfecta proved correct) but the Unanimous AI team have had plenty of success in other fields including picking the Oscars, the NCAA college basketball competition and the Superbowl winner with high degrees of success.

THE POWER OF PREDICTION AND THE FUTURE OF AI

“Gambling may never be the same” Unanimous AI half joked in a recent blog post.

Such a statement might be a little premature but the company really believes they’re on to something. While none of the of racing experts associated with the Kentucky Derby were able to pick the order of winners correctly, a group of 20 strangers were.

“If we assume the average IQ of an individual participant was 104, which is the global average, we could say that when it came to horse racing UNU demonstrated a 435 per cent amplification over that,” Rosenberg said. “That puts this UNU swarm at an IQ of 452 in this context. We think that’s a pretty fun way to look at it.”

Critics are quick to point out the immense uncertainty of the system in terms of predicting sporting events, and they’re absolutely right. But the true application the company hopes for Swarm AI is not predicting the Superbowl but rather shaping the future of machine intelligence.

There is no shortage of informed experts expressing concerns over the dangers of artificial intelligence and Rosenberg counts himself among that group.

“I am so concerned about the dangers of traditional AI (that) a few years back I realised that the only way for humanity to prevent itself from being replaced by purely artificial intelligences in the coming half-century, is to amplify our human intellect and keep ourselves ahead of the curve,” he told news.com.au.

Dr Louis Rosenberg believes in the power of humanity’s collective knowledge.Source:Supplied

He understands the negative perception some have towards the idea that the company is creating what they see as a “hive mind”. But he views the technology as a way to build an AI system that “keeps humans in the loop” and reflects our values and sensibilities. He holds hope that platforms such as UNU can be used to chart an intelligent path forward for humanity as a whole, including grappling with universal issues such as global warming.

“Individuals, working only for their own self interest, make poor decisions that lead to war, pollution, global warming, inequality, etc. A Swarm Intelligence can help us reach a decision that optimises the outcome for the species as a whole.”

Collective wisdom and intuition trumps that of the individual few. At its core, it’s an advancement of the democratic principle.

“The experiment we did for the Kentucky Derby was really just a test of this principle, showing that if we pool our knowledge, wisdom, and intuition, we can make better decisions than any of us can do alone.”

He may have made $10,000 in the process but we all stand to make a lot more if swarm intelligence can be successfully harnessed.

“The exciting implications are for important questions, not horse racing,” he said.

The company has high hopes for the power of the collective mind.Source:Supplied