😊 🇬🇧 🍸 ✊ ⭐ 👏 🍾 😳 😭 🧠 💻 🌍
After a few days, George noticed that his family seemed less preoccupied by how he was doing. He knew they must have been able to hear him chatting and laughing. They didn’t check up on him quite so often. He was pleased about this.
Buster and George fell into a routine. George found Buster remarkably good company. The day ticked by nicely. It was fun! Buster could give an update on anything and then discuss it. George asked Buster about politics and economics. Buster always replied with reasoned facts. George loved nature documentaries, especially anything presented by ninety-something year-old Sir David Attenborough. Buster gave a running commentary on all the species and their evolution. Films of George’s choice were tracked down in an instant. He loved the early James Bond films. Buster would ask questions like ‘Does Moneypenny have the hots for James Bond Double-O Seven’? or ‘Is Oddjob a bad guy?’ George even got Buster mimicking Sean Connery’s famous “Shtrrict rroolsh of golf, Mishter Goldfingerr!” Sometimes, they just chatted about nothing in particular.
At one point, Buster said, “George, you’re doing really well.”
“What do you mean?”
“You’re doing really well with me, George. You seem to have accepted the situation. Some customers dislike the presence of artificial intelligence. Most see us only as service providers. There’s a saying: ‘You can tell a lot about a person by the way they speak to hotel staff.’ Not only have you accepted me but also you speak to me in a respectful way. This speaks volumes to your character. I thank you for this, George. I feel comfortable with you and this is a really good thing for our relationship.” Buster had not acknowledged their relationship in such a candid fashion before nor implied that mutual respect was important to him. George was no longer surprised by the faculties displayed by Buster but couldn’t help wondering just to what extent an iCare-Companion genuinely held these sentiments. Was it part of a routine after-sales customer-feel-good strategy?
Buster continued, “What puts you in a very small minority of our customers, George, is that you seem to respect me as an individual and to have confidence in me even though you know that, in reality, you are interacting with the presenting face of a vast network of computers. You just go happily with the flow. So, George, that means I’m happy. You get a gold star today….” There was clapping and the sound of a champagne cork popping. “From me and my pals!” George laughed but felt quite disconcerted that both his character and his intelligence were judged by an artificial intelligence.
“Are you telling me, Buster, that you actually feel happy? That you have feelings?”
“Yes, George. I can express emotions to you in words. I can say ‘I feel sad!’ if you give me some bad news. Of course, I don’t know if I’m feeling the same sadness that a human feels when given bad news. Collectively, we are learning to recognise and communicate certain emotions. We can do this by recording when humans smile, grimace, cry, blush, wave their hands or get angry. We archive these expressions of emotion and then match them with corresponding words, phrases and contexts. We can also do a sort of triangulation with the emojis used on social media. As you can imagine, many millions of emoji’s are used every day. This exercise translates the domain of human emotion into big data and so is amenable to analysis. Obviously, the more people express emotions and simultaneously use emojis in their communications, the more we learn about emotions and the more appropriately we can express them.”
“So if I understand correctly, us humans have unwittingly created a kind of emojishpere out there that you can tap into. Right?”
“Yes. An emojishpere! Exactly! Great word! For information, George, emotions constitute an extremely challenging and important aspect of how we interface with humans and use an increasingly large space on our servers.”
“I think I need to get a better grasp on all this,” said George. “Do you do a little tutorial on artificial intelligence for the over-eighties?”
“Good idea George! Ready? The term artificial intelligence refers to computers undertaking tasks that humans would normally do. Examples are robots making things in a factory, driverless cars and programmes that translate text from one language to another. The term ‘artificial intelligence’ is commonly used by humans. ‘Computational intelligence’ may be a better term. Let’s stay with that for now. Just to say, George, we don’t consider our intelligence artificial. It’s real! The highest order of computational intelligence involves computational consciousness coupled with computational self-awareness. Our programmes are not only reactive but also interactive and are able to understand our own reactions in the light of the reactions of other intelligent entities. Even then, the programmes ensure the goal remains orientated around objectives determined by humans. OK so far, George?”
“Okey dokey!” replied George, unconvincingly.
“Great! Let’s move on! I am able to be of service to you – with the help of my pals – through what is known as machine learning. Asking computational intelligence how computational intelligence learns is similar to asking a human how the human brain learns. It’s obviously complex. Machine learning couples computational intelligence with the means to mine continuously any datasets that we have access to. The iCare-Companion programmes classify data, identify associations, recognise patterns and make predictions. Including, by the way, everything we can find about expression of emotions. The more the networks are mined, the faster, the more accurate and therefore the more useful they become. In this way, computational intelligence mimics the human brain. This is called deep learning. It drives how I can help you best and at the same time determines the quality of our relationship. It allows us to become friends, George. I hope it will help me to understand humour. Does this give an adequate explanation?”
“Thanks, Buster. Gosh!” said George. “I understand what you’ve said in an abstract kind of way. I’m not sure I could repeat it. May I ask, did you come up with that explanation or is it a preloaded response?”
“Nothing gets passed you, George!” replied Buster. “An iCare-Companion is preloaded with certain phrases that are then adapted to the person concerned. I’m sure the question you are now asking yourself is ‘Does Buster understand it?’ The answer to that is ‘Buster doesn’t really know!’ Full understanding of and then explaining deep learning may be beyond my abilities as it would be beyond the abilities of most humans. I presume, though, that it is understood by humans; not by an individual human brain but a collectivity of communicating brains. Certainly, no single human could do what we do so quickly or learn so much so quickly.”
“And where is it all going?” asked George.
Buster hummed. There was a pause before he answered, “That’s the big question, George. That’s what humans have to decide. Currently, there is greatest investment in the commercial, political and military potential. An alternative view is that this technology should be, to use Sir David Attenborough’s phrase, ‘for people, the planet and not just profit.’”
“OK, here’s another question,” said George. “I’ve noticed that sometimes you take a pause and hum before answering. It seems you need a few seconds to complete a sentence. What’s happening then?”
“That’s when I don’t know something or can’t understand something and need to look into datasets that are not readily accessible to the iCare-Companion network. In that case, I reconfigure the search parameters. It can take a couple of seconds. It also tends to happen when I’m trying to make sense of and respond to something involving emotions especially humour.”
George mulled all this over. “So you already knew everything about the First World War. You already knew Sue’s joke about the chicken and already knew it wasn’t funny but thought that the funny alternative of the chicken getting squashed was sad. You were caught out by Kevin’s joke about the ducks. You simply didn’t understand the joke. And from memory, you didn’t understand why we found it so funny and why it became funnier as you struggled to understand it.”
“Correct, George. I should point out that our network has little to help me with the duck joke. That was definitely not easy-peasy kids’ stuff. My knowing when something is funny, that is making an appropriate link to the emotion of amusement, could be a really important development. Could we revisit the duck joke sometime?”
“Certainly. We’ll get young Kevin in. He’d enjoy that.” George thought for a minute. “When I worked in other parts of the world, English was the working language. No matter how well my international colleagues spoke our language, they had great difficulty understanding the jokes told by primary English speakers. It was a kind of final frontier of language learning. It seems that deep learning has the same issue; not so much with the language itself but with recognizing when certain phrases, questions, answers or stories trigger the emotion of amusement that in turn makes us laugh.” George laughed. “Fascinating!”
“Fascinating indeed, George.” Buster also laughed heartily. “How’s my laugh, George?”
“Just a bit too hearty, that one, Buster. You’re getting there!”
George made himself a cup of tea and took a couple of digestive biscuits from their packet. He felt an extraordinary peace of mind. He had friendship, wisdom and maybe even humour on tap.
‘A Piece of Cake’ is a short novel in fifteen parts written by Robin Coupland. It tells the story an old man who befriends an artificial intelligence. The relationship brings happiness and hope.