A look at researchers engaged in the seriously difficult business of teaching machines to laugh.
Attendees at this year’s SXSW festival in Austin, Texas, were greeted by a curious sight. Crowding the streets was a group of demonstrators protesting against the development of artificial intelligence, with signs and slogans such as, “Humans are the future” and “A.I., say goodbye.” Was this backlash for decades of computer dependency, a kind of wake up call to remind people that technology is just a tool?
Those sober slogans hit home for many, but they also provoked laughter from passersby.
Despite the major strides being made toward artificial intelligence, there are still some sizable barriers to achieving a true thinking machine, not least of which is the ability to understand humor.
According to experts like David Gelernter, professor of computer science at Yale, a machine must understand the full range and nuance of human emotion before it can be deemed capable of creative thought.
Comedy may be the key to unlocking this emotional intelligence, since humor embodies many of the complexities of lateral thinking, problem-solving and unexpected insight that characterize the mind. Today, many researchers and engineers are studying the Science of Laughter and are racing to teach machines to understand and create jokes.
Computers Take the Mic
In recent years, there’s been a spate of comedy-based computing, including Manatee, the joke-writing computer; STANDUP (System To Augment Non-speakers’ Dialogue Using Puns); SASI the sarcasm-detector; and DEviaNT (essentially a “that’s-what-she-said” machine).
Julia Taylor, a computer scientist at Purdue University, designed a system that can identify children’s jokes amid larger selections of text, which is an important step in computer recognition of comedy.
Even more impressive, her system can explain why a particular joke is funny.
Taylor believes that any joke-telling computer needs to have a linguistic foundation.
“For a computer to understand humor, at the very least it needs to understand natural language,” Taylor told iQ.
“For an assistant type of A.I., emotional recognition and facial recognition would be requirements, but natural language recognition is the one thing that’s universal to building this type of system.”
One of the most common challenges in developing comedy-comprehending machines is the lack of a scientifically valid, agreed-upon definition of humor.
“People have been looking at humor since Aristotle, and historically it’s been more philosophy than science,” said Scott Weems, author of “Ha! The Science of When We Laugh and Why.”
“It’s hard to draw the line where philosophy ends and science begins. We still don’t have a definition of what humor is. Ask 10 scientists, you’ll get 10 different responses.”
In other words, human beings need to figure out what humor is before machines can do it too, and that could very well be the major leap needed to achieve artificial intelligence. According to Taylor, it will take a multidisciplinary approach to crack this problem.
“Individually, each field can’t answer the question, but when you combine them all together you get a clearer picture,” Taylor said.
“Where A.I. comes in is that it’s a perfect text mechanism. If you can write an algorithm that would predict when a particular person would find a joke funny or not and the results are accurate, you’ve got a working system.”
Can Spontaneity Be Programmed?
The spontaneous nature of humor is a confounding factor when it comes to A.I. Even the most sophisticated machines require programming and some type of template to process inputs. But much of comedy depends on unpredictable outcomes, a skill computers generally lack.
While computers can be trained to switch between multiple templates or hypotheses, they’re still incapable of handling a truly original concept outside the boundaries of their knowledge base. To understand spontaneous jokes and truly have a sense of humor, machines would have to gain a much broader contextual appreciation of reality.
“I would personally like to see a computer get closer to a human understanding of the world, the actual things that we see and how we relate to them, not just the numbers,” Taylor added.
“Do I think that could happen in the next few years? Yes, I believe so, but maybe not enough for the general public to notice.”
Another question is whether it’d be worthwhile to create artificial intelligence in the first place. A sizable contingent of skeptics say that AI is impossible because any version of it would still merely be used to supplement or enhance human activities, and consequently would never meet our version of true intelligence.
For example, Brian David Johnson, an Intel futurist and author of “21st Century Robot,” believes any machine is, and will continue to be, just a machine.
“Robots hold a special place in our culture,” Johnson told iQ. However, “all technology is a tool…just different types of hammers. Technology frees humans up to do what they’re really good at.”
Photos by Shutterstock.