Interview With Dr. Louis Rosenberg, Founder Of Unanimous A.I.
Unanimous A.I., founded by inventor, technologist and entrepreneur Dr. Louis Rosenberg, develops technologies that blend human and computer intelligence and demonstrate that none of us is smarter than all of us.
It's rather fitting that Dr. Louis Rosenberg, an individual wholly dedicated to preparing humans for the immediate and distant future, is featured in a project titled Year Million, National Geographic Channel's six-part documentary series that explores and postulates on the future of humanity; on what it will be like to be human one million years into the future.
Dr. Rosenberg is the CEO & founder of Unanimous A.I., a Silicon Valley startup that has developed a new form of artificial intelligence called Swarm A.I. This new technology is not out to replace human intelligence, rather just to amplify it. Unanimous A.I. operates under the premise that A.I. will dominate the future, therefore humans best adapt accordingly or face a fate of becoming vestiges of the past.
OMNI spoke with Dr. Rosenberg about his conception of group intelligence, Unanimous A.I.'s projects and predictive technology, and his contributions to NatGeo's Year Million.
OMNI: What exactly does your company, Unanimous A.I. do?
Dr. Rosenberg: In layman’s terms, at Unanimous A.I. we build “hive minds.” In more technical terms, we’ve been working for years on amplifying human intelligence by networking groups of people as real-time systems modeled after natural swarms. The technology we’ve developed is called Swarm AI™ and it’s a combination of human input and A.I. algorithms, enabling emergent systems to form that are significantly smarter than the human participants that make it up.
Describe the distinction between swarm intelligence and crowd intelligence (i.e. the average of many disparate opinions will yield the correct answer).
Scientists have known for over 100 years that groups of people are smart when we combine their knowledge, wisdom, insights, and intuitions. The field of research is generally referred to as Collective Intelligence and for most of the last 100 years it’s been focused on collecting data from individuals and taking statistical averages of that data to yield a result. This old-school process is generally referred to as “crowdsourcing” because when you collect data from individuals in isolation and take a statistical average – the group is called a “crowd." And while the process yields a small amplification of intelligence, the effect is mild.
My research started with a basic question – is this really the best way to aggregate the diverse knowledge, wisdom, insights, and intuitions of populations? To answer that question, I looked to Mother Nature. And it turns out that nature is very good at amplifying the intelligence of groups, but it does not use crowds. That’s to say, nature does not conduct surveys or take votes and compute average results. Instead, nature forms real-time systems with closed-loop feedback that enables groups to explore a decision space in synchrony and converge on optimal solutions in unison. Biologists call such a group, a swarm. And in fact, this is a primary reason why birds form flocks and fish form schools and bees form swarms – they are smarter together when working as a system and converging in synchrony on solutions to problems. This process by which natural organisms converge on optimal solutions through closed-loop feedback control is generally referred to as “Swarm Intelligence.” Humans did not evolve with the natural ability to form swarms, so we need to use intelligent software algorithms to connect people together. This is referred to as “Artificial Swarm Intelligence” and it has been shown through many research studies to amplify intelligence far more than crowd-based methods.
Which industries are currently utilizing Swarm A.I. technology, and what can we expect in 10 years?
The potential applications of Swarm A.I. are extremely diverse, as it’s useful anywhere you might want to significantly amplify the intelligence of a human group. And because Swarm A.I. taps the knowledge, wisdom, insights, and intuitions of the members of the swarm, it’s especially useful in situations where the information you are trying to amplify is not easily represented as simple data – but exists as subjective judgments within a population. The first industry to jump on this technology area was market research, because Swarm A.I. can tap the opinions of a consumer population and provide much more accurate insights than old-school methods like polling, surveys, or focus groups. This has grown into a broader field of “business intelligence” because swarms of people can make more accurate predictions than individuals about everything from the sales of a product, to the popularity of a new product feature. Swarm A.I. has even been used to more accurately predict movie box-office performance than traditional methods.
What’s your all-time favorite psychology/behavioral experiment and why?
There’s a great experiment that professors sometimes use to demonstrate the limitations of human morality. The experiment is very simple – the professor puts an extra credit problem on their exam that asks the student to write down how many points of extra credit they want, 5 points or 15 points. The only catch is that if more than a third of the class asks for 15 points on their exams, nobody will get any points. And guess what – the vast majority of the time, far more than a third of the people ask for 15 points and nobody gets anything. The experiment examines a principle called the “Tragedy of the Commons” and demonstrates that individuals have a hard time being selfless, and that results in decisions being made which are not in the best interest of the greater good. I like this experiment because it helps demonstrate another big advantage of swarms. In the natural word, where groups reach decisions as a system (i.e. swarms), they avoid this pitfall and generally converge on the solutions that are best for the greater good. Research on human swarms suggests this works for us too. To me, this means that as we look forward towards larger and larger “hive minds” that incorporate thousands, or even millions of people thinking together, the emergent intelligence is likely to be highly moral.
Is it necessary for there to be homogeneity (or some commonality) among the group of problem-solvers in order for swarm intelligence to ‘kick in’?
Actually, homogeneity is a negative property of swarms. It’s far more valuable to have swarms where the participants are diverse, each bringing unique views and opinions and perspectives into the systems. When building human swarms, we often recommend that the participants are diverse in age, gender, location, and expertise. It generally produces the smartest systems.
What’s your ‘crowd behavior pet peeve’ (i.e. the most frustrating example of faulty groupthink)?
A significant difference between ‘crowds’ and ‘swarms’ is that crowds provide sequentially over time, while swarms are synchronous – meaning participants provide input together, at the same time. This difference relates to the most frustrating thing about traditional old-school crowds – “snowballing” – where the input from the first few people influences the input from the people who follow – which influences the people who follow that. In fact, research studies have shown that the first LIKE on Facebook, or the first UPVOTE on Reddit can greatly influence all the input that follows, which results in highly distorted results. Technically called “social influence bias,” this is the worst form of groupthink because the process amplifies random noise, rather than providing the true collective intelligence of the group. Nature figured out how to solve this through swarms rather than crowds because in a swarm there are no leaders, and no followers – everyone is equal and everyone contributes together in real-time, converging in unison.
Tell us about some of the coolest predictions Unanimous A.I. has made, like the 2016 Kentucky Derby winners?
At Unanimous A.I., we’ve been willing to take on challenges from journalists, which has resulted in many documented predictions that have demonstrated how remarkably smart groups can be when they swarm. The most famous instance was the Kentucky Derby, as we were challenged by CBS Interactive to predict the first four horses in order. In horse racing, this is called the Superfecta and it went off at 542 to 1. The swarm got it perfect, which meant that anyone who placed a $20 bet on the prediction published by the reporter would have won $12,000. Many people did. Also last year, the Boston Globe challenged us to predict the outcome of the baseball season last year, at midseason, with months of games left to play. The swarm correctly predicted all eight teams to make the playoffs, correctly predicted that the Cubs would face the Indians in the World Series, and correctly predicted that the Cubs would win it all.
What’s the focus of your episode of Year Million?
While I appear in a number of episodes of Year Million, it is episode #4 where I have the greatest contribution. That episode talks about the future of human intelligence and predicts that humans will form tighter and tighter systems over time (i.e. swarms) until we eventually converge into a global hive mind that is capable of extreme super-intelligence. The episode points out that by thinking together as a hive mind, humanity may be able to stay competitive against the threat of pure software-based A.I. systems that are expected to emerge over the next 50 to 100 years.
What intrigued you about the project, and what was it like collaborating with so many amazing scientific minds from different backgrounds?
The Year Million show paints a very interesting vision of the future because it captures views from many top scientists across a diverse set of fields, from Psychology and Cosmology, to Artificial Intelligence and Neuroscience. While the experts don’t all agree on every point, it’s pretty clear that everyone they consulted believes, like it or not, that humanity will experience very significant changes over the next 50 to 100 years – changes that could fundamentally alter what it means to be human. Frankly, it’s scary. And the producers of the show did not sugar coat the issues, presenting a very honest vision of the future.
Recommend to our readers a book / TV show / movie
Year Million, on National Geographic Channel…
How can our readers apply the principles of swarm intelligence to improve their daily lives and decisions?
The best advice from the study of natural swarms is the revelation that diversity of opinions generally leads to the most accurate solutions. This means, when forming human teams for any application, having a range of views, backgrounds, and opinions is a great starting point. Teams can experiment with actual "swarming technology" at UNU.ai, which is an open platform we have made available for casual users who want to experiment with their own groups.
The last episode of Year Million, "Beyond The Cosmos" airs Wednesday, June 21, at 9/8c on National Geographic Channel