Macbeth and mirror neurons

What really happens when bytes and brains come together?

I hope that you will indulge me today as I talk about Artificial intelligence, it’s something I’ve been waiting a few decades to talk about, specifically how a text book written in 1992 got me interested in how brains and computer bytes would come together.

Photo by Baran Lotfollahi on Unsplash

The book is Artificial Intelligence (AI) by Patrick Henry Winston, I picked it up in a second-hand shop on Manchester Oxford Road in 1997. It’s strange to think that I didn’t have a Mobile phone or Personal Computer when the book was first published in 1992, and it would be 2000 before I even had a laptop and access to the internet at home.

Patrick Henry Winston on Speeches and Psychological Heuristics

Up to that point, my experience of technology had mostly been shaped by retail systems, a Saturday job in a newsagents involved saving data every night onto a plastic diskette, (It was significantly newer than the BBC computers with a floppy disk we had at school). Many of the concepts and ideas I was taught about computers at university in the 90’s originated from the world of business too, systems for inventories, billing, automating production lines, all heavily influenced by the logical work of Mathematicians such as Alan Turing and Charles Babbage, and although visionary ideas about the ways in which humans and computers would interact were documented in the journals of a brilliant mathematician Ada Lovelace (who worked alongside Babbage), they didn’t describe the how. Computing in the 1990s was all about using coding language as a vehicle to solve problems, an idea that underpins the applications we use on our phones today!

The reason I’d picked up Winston’s book about AI was because his ideas about computers and intelligence were very different to other thinkers, he was setting out the how, book chapters talked in human terms of neural pathways, he was convinced that understanding the human ability to tell stories would be the key to creating intelligent systems. A decade later he’d publish the Strong Story hypothesis and went onto describe how this was being tested by pushing a short summary of Macbeth through a system they’d built called Genesis, the system was able to work out the revenge plotline.

Was Winston right though? It’s intriguing to revisit his work twenty years later and look at this through a psychological lens, I’d certainly question the starting point which is an assumption that humans are more intelligent, the ones to copy despite our flawed and biased reasoning. The fact is that Nature offers up different types of intelligence, curiously, Winston was reported to think that rats had the ability to tell themselves stories like humans. I’m sure my dog has dreams so this doesn’t feel entirely unfounded.

Psychologically speaking, Winston’s hypothesis is compelling, storytelling is something we can all relate to,  a practical tool to help us make sense of our world and connect with each other. For me, the biggest role for storytelling in AI is how it’s used to communicate our hopes and fears, and how it lets our imaginations run wild over abstract concepts. Sociologist Stefan Puntoni and colleagues, suggest that one way we can tap into those worries and fears is to think of AI as discrete experiences where psychological tensions exist, for example in data collection are we being served or exploited by the way our data is being captured?

There’s good reason to be paying attention to all the stories being told around AI and the types of experiences they deliver, positive beliefs do and can sit alongside AI, especially where there are clearer societal or personal benefits, for example an app that will improve wellbeing or make us happier.

But let’s  go back to Winston’s hypothesis that the true power of AI would be realised if we could get computers to understand the human capacity of storytelling, and think about it in modern terms , the American university MIT along with the consultancy firm Mckinsey built on those earlier attempts using the story of Macbeth, by  feeding an algorithm with hours of video and character data, to train it through machine learning to understand and learn about the emotional arc that stories provide, and it worked to some extent, they could predict when an audience would be most engaged, through strong emotional responses.

There’s a much bigger real-life example of this if you think about any social media site, apps like tik tok and Instagram give us features to story tell, and these are exploited by algorithms, a curious thing happens when we watch a video that invokes an emotional response though, brain imaging studies show that when we vicariously experiences an event associated with an emotion, the same regions of the brain are activated as if we had experienced the event directly.

In primates there’s a type of brain cell known as a mirror neuron these are activated when an action of another primate is observed, whether this happens in exactly the same way for humans is of great interest and debate.

All of this leads to a direct effect on the value we place on different types of content, perhaps unsurprisingly cyberpsychologists have found that social media has a  positivity bias, and It’s for this reason that I’m not so sure that storytelling is the only way to create intelligent systems, I worry that on some platforms the bravery or triumph over adversity narrative is nudged through algorithms as ‘normal’, when the reality is that it’s served up more because we respond strongly to it.

Storytelling will always be synonymous with AI, but perhaps for different reasons than Winston’s hypothesis, perhaps storytelling’s encore is to help us articulate the uneasiness of living in an unequal world where humans can exploit each other through algorithms.

What do you think, is storytelling the key to cracking AI?