Email
Password
Remember meForgot password?
    Log in with Twitter

article imagePlatform pairs human intelligence with AI: Interview Special

By Tim Sandle     Jun 22, 2018 in Science
Dr. Louis Rosenberg, Founder and CEO of Unanimous AI, and he has developed a platform that pairs human intelligence with artificial intelligence. Dr. Rosenberg talks with Digital Journal about how, and why the two can and must coexist.
Unlike traditional AI methods, the Unanimous AI platform capitalizes on values, emotions and interests of different groups of people to provide superior predictions of outcomes, like elections, the Super Bowl winner and business decisions.
In tests, the AI technology has outperformed traditional AI systems that use more common machine learning, as well as human experts, critics, and traditional polls. This form of AI is based on 'swarm intelligence' and this approach has been successfully used in the context of forecasting problems. Swarm AI is based on how birds, bees and other animals work together to solve problems.
To understand more, Digital Journal spoke with Dr. Louis Rosenberg, Founder and CEO of Unanimous AI.
Digital Journal: How important is artificial intelligence for the world?
Dr. Louis Rosenberg: To answer this question, it's worth defining what we mean by "Artificial Intelligence". I would say an AI is an artificial system that takes in noisy information about the world, processes that information, and make sense out of it at a level that is useful for answering questions about current circumstances, generating forecasts about future circumstances, or taking actions that could not have been taken without making sense of the noisy information. With that as the context, artificial intelligence is profoundly important to the world, as there are endless applications for any technology that can take in noisy information about the world and make sense of it.
DJ: How great is the potential for AI?
Dr. Rosenberg: The potential for AI is extremely valuable, but there is a dirty secret that limits that potential - data. The vast majority of AI systems in the world today are machine learning mechanisms that must train on vast amounts of historical data about the world before they can take in any new data about the world and make sense of it. And, the historical data must be similar to the new data you want to make sense of, or your results will be wrong.
This means, for AI to have value it needs to be targeted at applications where historical data exists in a form that can be accessed (massive databases) and that the data is not stale (i.e. it still represents the world enough that training on it will give you insights about new data). But this is not as simple as it sounds.
Most data is incomplete, inaccurate, or out of date long before you can have enough of it to train a machine learning system. Think of politics - you can have years and years of data about political preferences, but a single Tweet from the president could render all of that data stale and useless for predicting the future, because the circumstances on the ground have changed.
DJ: What is ‘Swarm AI?’
Dr. Rosenberg: Swarm AI is a unique approach to AI where we don't access large sets of data in computational databases to generate insights about the future. Instead, we access large groups of people and connect them together with intelligence algorithms modeled after swarms in nature. In other words, we're drawing upon the knowledge, wisdom, insights, and intuitions of real people, in real-time, and amplifying that intelligence by connecting them into intelligent systems moderated by computational algorithms. This solves the data problem that most AI systems suffer from, as the data in people's heads is both extremely deep and continually up to date.
DJ: You’ve written that it’s important to pair human psychology with AI, why is this?
Dr. Rosenberg: We build hybrid systems that harness the knowledge, wisdom, insights, and intuitions of real people, in real time, by connecting them together with AI systems, moderated by intelligent algorithms. By doing this, we create "artificial experts" that can quickly generate more accurate predictions, decisions, evaluations, and forecasts than the individual participants could make on their own.
And, when we do this we are not replacing human intelligence with technology, we are amplifying human intelligence with technology, keeping human values and morals and sensibilities in the loop. We believe this is a safer and smarter path forward for building highly intelligent systems.
Dr. Rosenberg has used these concepts to from a company called Unanimous AI. Central to this is a platform designed to a means of building intelligent systems that keep human values, morals, wisdom, and sensibilities deeply embedded in the process. For Dr. Rosenberg's thoughts on his company and its core aims, see: "How Unanimous AI platform capitalizes on values: Interview."
More about Artificial intelligence, Psychology, Humans, Machines
More news from