Swarming Artificial Intelligence

Unanimous A.I.  http://unanimous.ai has developed what they call Artificial Swarm Intelligence, the UNU swarming platform and the Swarm Insight on-demand intelligence service. The UNU platform allows online groups to form real-time “human swarms” in order to tap collective knowledge, wisdom, and intuition. Distributed online groups can answer questions and make predictions in real-time.

We have been talking about the wisdom of crowds for more than a decade. In 2004, The Wisdom of Crowds (with its very long subtitle "Why the Many Are Smarter Than the Few and How Collective Wisdom Shapes Business, Economies, Societies and Nations") popularized the idea. The term collective intelligence (CI) is also used for shared or group intelligence that emerges from the collaboration and of collective efforts of many individuals. This also leads to consensus decision making.
CI appears in sociobiology, political science as well as crowdsourcing applications.

You'll also hear the term collective IQ, which is a measure of collective intelligence. We shouldn't be overly proud of our use of this since collective intelligence has also been attributed to bacteria and animals.

Unanimous A.I. takes this idea of the wisdom of the crowd's collective opinion rather than that of a single expert. They use this amplified intelligence to generate decisions, predictions, estimations, and forecasts, which can be seen in events such as the Kentucky Derby, the Oscars, the Stanley Cup, Presidential Elections, and the World Series. They call the technology Swarm AITM using algorithms and interfaces modeled after swarms in nature. 

Don't Fire the Humans Yet

human intelligence


AI is the thing everyone wants to use. Social media is in love with artificial intelligence. Of course, much as the cry went up when computers first appeared, some people say that "AI will take our jobs."

Facebook has almost 2 billion users. Those users post a lot of content. Mark Zuckerberg has made it clear that live video is a big part of the future of Facebook. But the company has come in for a lot of criticism for violent video posted this year, including murder and suicide.

How does Facebook (and other social media companies) decide what content violates its community standards? They all are desperately implementing and experimenting with AI, but they still rely mostly on humans.

Facebook announced recently that it is using an AI system designed to identify users contemplating suicide or self-harm. How? By using pattern recognition to determine if a post and its comments resemble previous posts identified as being about self-harm. Facebook is also including clearer options for reporting posts that appear to indicate self-harm. It is people reporting to people who determine inappropriateness.

AI-based image-recognition tools that users can use are assisting human moderators now. Can the 54,000 potential cases of sexually related extortion and revenge porn reportedly posted each month can be found and deleted by AI? Not yet. 

Did you see the film Hidden Figures ? In the early 1960s, the mathematicians working at NASA were called "computers" - people who did computations. But those human computers also saw the entry of IBM mainframes into NASA that were better  computers. They realized they would need to become the humans who could program those electronic computers if they wanted to keep working. Take note Facebook and other companies - and anyone who wants to work for those companies: AI requires human intelligence. 

After 2 murders were broadcast live on Facebook in April. Mark Zuckerberg announced that the company would add 3000 employees to the already 4500 employees who work on their Community team reviewing reports on videos. Live video is growing rapidly online, and Facebook Live is a service with 1.9 billion monthly users to broadcast video. Lawmakers in Germany and the UK have also been pressuring social networks to better remove illegal hate speech and clamp down on fake news. The 3000 new workers will monitor all Facebook content not just live videos. This team would operate around the world and will most likely be virtual contract employees.

Just last week, Facebook's "leaked" guidelines for dealing with these types of situations became public that hopefully can make a big difference in preventing suicide and other life-threatening situations.  

Autonomous Vehicles and Autonomous Learning

autonomous car

One of the newer categories on this blog is for VR, AR and AI. They were not topics of much concern in education when I started writing here in 2006. They are topics of interest now. 

The same may be true of autonomous vehicles and it is definitely true of what I'm calling autonomous learning

You are more likely to hear news about "autonomous vehicles" rather than "driverless cars" these days. They are pretty much interchangeable, but the former doesn't sound as scary. In the way that "global warming" was replaced with "climate change," the newer terms are not only better in public relations terms but also are more accurate.

An autonomous vehicle (AKA driverless, auto, self-driving, robotic) is one that is capable of sensing its environment and navigating without human input. Many such vehicles are being developed, but as of this writing vehicles on public roads are not yet fully autonomous.

Many of the experimental cars and trucks you might see on the road (or, more likely, on the news) have a human along for the ride and ready to take over if needed. Initially we all heard about this future where you would get in a car, tell it your destination and sit back and relax. It was a taxicab without a driver. But more and more we are hearing about the autonomous vehicle with no human in it that might be delivering packages to locations. (No word on how they are unloaded. I guess you meet the vehicle at the curb.)

I was talking to a friend who has no involvement in education about an online course I was teaching and how MOOCs are being used. He said, "So, it's like an autonomous vehicle."

My first response was "No, its not," but when I gave the idea a few moments, I saw his point.

You set up a good online course. It has AI elements and guided learning, predictive analytics and all the other tools. The student enters and goes along on their own. Autonomously. Teacherless.

Some archived MOOCs are already somewhat like this - though probably minus the AI and guidance systems.

I call this autonomous learning. If you search on that term today you are more likely to find articles about learner autonomy. This refers to a student's ability to set appropriate learning goals and take charge of his or her own learning. However, autonomous learners are dependent upon teachers to create and maintain learning environments that support the development of learner autonomy.

My friend and I took the vehicles:learner comparison further. The mixed or hybrid car will probably be with us for a few more decades. By hybrid I mean not only with its fuel but also with driver-assist features. Part of the redundancy there includes the passenger as backup driver - a guide on the side. The car can park itself, but you might need to help in some situations.

Hybrid or blended courses are also going to continue to be around for awhile. Like the vehicles, the fully-automated course will be the experimental exception for a decade or two. But those kids in the college Class of 2037 have a very good chance of taking autonomous classes.

I will feel safe on the road with autonomous vehicles when ALL the vehicles are autonomous. Throw a few human drivers in there and the reliability drops. Do I feel the same about autonomous learning? Too early to say.