Much has been made about the recent advancements of Artificial Intelligence. New AI companies seem to sprout up daily, and just this week Silicon Valley icons Elon Musk and Sam Altman announced a new partnership called OpenAI to “maximize the power of AI – and then share it with everyone.”
And yet, for all of their potential,systems that currently tout A.I. fail to replicate — much less improve — human thought. To demonstrate this challenge , we asked a few mainstream forms of Artificial Intelligence to answer a very simple question, “who will win the Super Bowl?”
First we asked Siri, Apple’s familiar personal assistant, to make a prediction.
Up next was CleverBot, an AI chatbot which famously claimed to have passed the Turing Test.
Finally, we asked UNU, a “human swarming” AI platform developed by Unanimous AI, which collects dynamic input from real-time groups of networked users and uses A.I. technologies to produce a single unified output.
I should mention, I chose UNU because it made same prediction last year and actually got it right. UNU picked the Patriots before the playoffs had even started. UNU also correctly picked the Blackhawks to win the Stanley Cup, the Warriors to roll to the title, and the Royals to win the World Series. UNU also did better than the New York Times in predicting the winners of the Oscars. In contrast, we knew that Siri and Cleverbot were longshots. We just didn’t realize how long…
Here is what happened when we asked Siri to predict the winner of the Super Bowl:
As you can see, not only does Siri fail to make an intelligent prediction, it misunderstands the question entirely. Instead of considering the current landscape of the NFL and rendering an intelligent opinion, Siri transforms the simple question into a useless historical search. We gave Siri another chance, asking it to predict this year’s AFC Wildcard winners, and once again Siri returned the results for last year’s AFC Playoffs. Siri clearly reveals herself as “artificial” but we’re just not seeing any “intelligence.”The best thing we can say about this answer is that it came quickly.
Next, we asked Cleverbot the same question. Keep in mind, Cleverbot claims to have passed the Turing Test, which is the traditional threshold for determining whether a machine has achieved “intelligence equivalent to, or indistinguishable from a human’s.” Surely Cleverbot will be able to answer a question that most elementary school kids could take a shot at.
Like Siri, Cleverbot failed to deliver an intelligent prediction. Unlike Siri, Cleverbot tried to disguise its failure with non-sequitors. But, in Cleverbot’s defense, one could argue that its answering the question of “Who will win the Super Bowl?” with a snarky “Me,” means that chatting with the bot has proved to be indistinguishable from chatting with an annoying person. Or perhaps a two-year old. That said, Cleverbot seems to be “faking intelligence” as opposed to tapping it.
Finally, we asked a group of 40 random people using UNU to make a Super Bowl prediction, their input processed in real-time by A.I algorithms and merged into a single unified output. Basically, UNU is a brain of brains:
As you can see, UNU has no problem offering a prediction. Each of the 40 magnets represents a person in a real-time negotiation. Together, those 40 brains form a single dynamic system, a kind of superbrain that is able to easily parse a future-facing question, consider a number of possibilities, and reach a conclusion in seconds. Given that the Carolina Panthers are 13-0 right now and the choice of many experts as the best team in the NFL, UNU’s prediction seems not only reasonable, but intelligent.
In this way, UNU represents a unique form of AI. Each swarm is an emergent artificial intelligence with its own distinct personality and wisdom, but, by keeping humans in the loop, UNU also solves one of the fundamental problems of our current vision of AI. As the Super Bowl question demonstrates, UNU’s human-powered Intelligence not only harnesses the power of many minds at once, its form of AI remains undeniably human.
Naturally, these types of intelligences are all in their infancies. But, what UNU does demonstrate is that humans are still miles ahead of Artificial Intelligences in their ability not just to process information, but also to understand the nuances of even a simple question like, “Who will win the Super Bowl?”
And, perhaps most importantly, UNU is fun. In a way, UNU takes the fake personality of Cleverbot and make it real. Because each Swarm takes on the unique character of all the people who it includes, having a conversation with UNU can be not only intelligent but incredibly enteraining. That creates an exciting cycle, because UNU will only get smarter as more and more people join the Swarm.
Want to be part of a swarm? Now you can – swarms are forming to explore topics of all kinds, from football and basketball, to music and politics. If you’d to be a BETA TESTER for the folks at UNU, just fill out the form below.