Humour and Laughter in Artificial Intelligence (AI)

Humour and Laughter in Artificial Intelligence (AI)

Earlier this week I wrote about laughter in my Appliance of Science column in the Irish Examiner. I really enjoyed researching this fascinating topic; there are so many different avenues of study to explore but one that really caught my attention in the investigation into laughter and humour in Artificial Intelligence (AI). Some refer to it as the final frontier. I couldn’t squeeze everything into the column so I thought I’d share it here instead.

What is the difference between laughter and humour?

The research is still scant on laughter and humour and the differences between them. It is hard to analyse and quantitate such subtle, human things. What might make us laugh one minute, may not the next.

Laughter is used as a communication aid; from the gentle chuckle to the full on belly laugh, it helps us to convey our response to various social situations. We don’t just laugh at something funny, we can use it to build rapport, show trust and acceptance and to fill in the blanks in conversation.

Humour could be defined as the art of being funny, or the ability to find something funny; it is a two way thing. It is full of subtle nuances and relies on correct social interpretation and interaction – and it is innately human.

There’s no joke in delivering a joke

Comic timing and humour are difficult enough for humans so the challenge is great when attempting to transfer these abilities to robots.

Comic timing is a very subtle thing, and can be very difficult to pull off. Engaging in any form of humour requires a lot of real-time thinking, identifying and reacting to social nuances and a certain degree of empathy in order to understand when to deliver the line and to predict how it will be received.

How will robots detect these very human, and very subtle cues?

That is the next step in AI, programming robots with the ability to get in on the joke, detect puns and sarcasm and throw a quick quip back! There is a whole branch of science dedicated to research and development in this area. Scientists in this field are known as Computational humourists. And they have come a long way; these are just some of the algorithms they have created so far.

Acronyms and Algorithms

The hope is that robots will use computational intelligence to process conversation. Here are just of a few of the algorithms that have been created (you’d have to love them for the acronyms alone)…

SASI – Semi-supervised Algorithm for Sarcasm Identification … this machine algorithm, developed by an Israeli research team, was designed to assist AI with the recognition of sarcasm. They current report a 77% success rate and see no reasons why they cannot improve upon these results.

Scientists are discovering that the detection of sarcasm is a very important and useful tool for humans and would certainly be a great advancement in AI technology.

STAND-UP – System To Augment Non-speaking Dialogue Using Puns; This program was created by a team of researchers in Scotland to assist children that use computerised speech aids to help them with certain communication challenges.

DEviaNT – Double Endendre via Noun Structure … the software that tells dirty jokes. Developed by two computer scientists in Washington University to determine appropriate word triggers or phrases that can be followed with ‘That’s what she said’ lines and apparently working with 70% accuracy.

How far has AI come with laughter and humour?

Things have developed further than you might think. Any sci-fi enthusiasts will be aware how much humour has been added to the robots of the future.

It may have been nothing more than fiction when data got his sense of humour in Star Trek: Generations* (1994) but it was becoming a reality by the time we were watching Interstellar, twenty years later.

Detecting emotions in humans

Robots are making increasing advancements in the detection of, and response to human facial expression and emotions. Some of these advancements are a little unsettling … will robots be the new companions for those in their twilight years? Even more disconcerting is the robot that can detect a criminal just by their facial features.

On a lighter note, many of these developments are focused on detecting facial muscle movements in humans as triggers for laughter. They are well on their way to detecting different types of human laughter too (which is something that many of us humans still find difficult).

Software has even developed to determine the correct pause time in response to laughter cues, and in detecting hidden laughter.

Robots on the comedy scene

Robots are pitching themselves against stand-up comedians to test their abilities. Although it is early days yet, some, like Robothespian are certainly holding their own.

My favourite is the Nao robot. Nao is only 58 cm in height and I think, firstly, this is one of the elements that I find so appealing; this robot does not try to look like me. Nao has learned to interpret human laughter with a 65% success rate, and, when he laughs in response, he does so with his whole body.

He is also doing well in his comic abilities, scoring very close to a human rival in a recent stand up challenge against a human.

How do humans respond to robots telling jokes?

So it seems humans are well able to laugh at a pun delivered by a machine. In comedy stand up situation it may put the audience at ease as they are not worried about hurting someone’s feeling or letting them down if they don’t laugh. The stress of creating rapport is removed.

It appears that people will also take rude jokes better from a robot than a human.

It’s all in the data

A lot of these developments are achieved because of the amount of data available in the world today. From what coffee we drink, to what TV programmes we watch, everything is recorded. Every time we like a Facebook post or make an on-line purchase we add to this growing mass of information that is used to determine and code how humans work!

This’ll stop you in your tracks

A robot has been doing the TV circuit of late and recently stole the show on Good Morning Britain; I found the clip fascinating and unnerving in equal measures. The Robot in question is called Sophia; it may not help that she reminds me of a movie I watched recently on Netlfix*, called Ex-Machina (I’d recommend watching it, but maybe wait a while after reading this post).

The facial expressions and minute muscle movements in Sophia’s face is amazing; she is programmed with 62 facial expressions.

Take a look…

Is any of this really necessary?

I think it is fair to say that there is much progress still to be made in the advancement of humour and laughter in AI but it is still remarkable how much has already been achieved.

The question is, is this a good thing? Do we need or want our robots to develop such human qualities?

Computational humourist Vinith Misra suggests that these advances could be the way to “make healthy relationships between us and our machines “and may, in the processes even make better connections between us humans.

But is it necessary for machines to be fully integrated into human lives?

Those in the business believe that these advances can reduce human stress and ultimately strengthen human bonds. Maybe we can learn from AI about how to laugh and make others laugh?

Laughter certainly has a lot of benefits to us humans; does it really matter if it is a machine that is makes us laugh?

What do you think?

*Disclosure: As a member of the Netflix Stream Team I have received a years subscription to Netflix, free of charge, and an Apple TV, for streaming purposes.