Will machines ever feel?

AI making a heart shape

As Artificial Intelligence (AI) is being integrated into an increasing amount of technology and devices, questions arise as to how much of what humans can do will be emulated or even bettered by machines.

While intelligent machines can be easier to imagine, the notion of feeling machines might seem puzzling or unsettling.

Yet for several luminaries in the fields of AI and robotics, this is no far-fetched hypothesis.

Philosopher and researcher Aaron Sloman first became interested in AI as a young philosophy lecturer, when AI researcher Max Clowes shared some of his knowledge and convinced him that computational ideas were related to the philosophical questions that most interested him.

Sloman says: "Most people who think about benefits of AI think in terms of what we could get machines to do rather than how our understanding can be extended by trying to design and build increasingly complex and able machines. I regard increased understanding as far more valuable than increased wealth, comfort or riches."

Sloman believes that machines could prove particularly useful in understanding ourselves as human beings, complex aspects of our consciousness and personalities, and perhaps even of our psychological development. This could in turn lead to the development of "an entirely new science of minds, including human minds".

When asked if he believes machines could ever feel, Sloman said: "You and I are machines, biological machines, and we feel many things. When we have deeper theories about the control mechanisms in animal minds and how and why they evolved, I see no reason to assume that it will be impossible to replicate them in artificial brains."

However, to create these machines, researchers might need to go beyond the use of digital electronics, for instance integrating them with chemistry, which Alan Turing believed to be at least as important as electricity in brains.

Some machines have already been developed to create pictures, as well as compose music or poems that could be seen as 'pieces of art'.

For instance, AARON, a program developed by artist Harold Cohen, who died in April 2016, created colourful pictures that could arguably evoke similar feelings to those triggered by human art.

In future, machines could even take over jobs that require empathy or emotional intelligence.

Murray Shanahan, professor of cognitive robotics at Imperial College London, says: "Sometimes, what we as humans want is the 'human touch'. If you think about robots in caring professions, there are some aspects of looking after another that a patient may be glad if a robot covered, particularly intimate things that people might not want others doing. But for other things, such as providing companionship, people want the 'human touch', and they might not like those to be provided by a digital assistant or robot."

Different cultures might have different opinions on this, with Shanahan suggesting that "the Japanese seem to be much more receptive to the idea of robots in society than Westerners".

Ultimately, the creators of such machines could decide to engineer systems that only replicate a specific ability or aspect of human consciousness, or a combination of them. For instance, they could develop a machine that only displays empathy or is only 'intelligent' when it comes to a specific task (or set of tasks).

Richard Sutton, a computer scientist and professor of computer science at the University of Alberta, Canada, is a pioneer of reinforcement learning in computing.

Reinforcement learning is at the heart of several AI developments, such as DeepMind's AI that beat human champions at the Atari Games and Alpha Go.

Sutton says: "The more I think about it, the more I'm convinced that one day machines will create art, feel empathy, fall in love, or produce behaviours that are simply impossible to conceive right now. Probably we will make the emotions of machines closer in some ways to humans, or closer to other animals."

"We already have AIs with simple emotions that are just as real as peoples'," Sutton continues. "Emotion is essential to having goals, and goals are what intelligence is about."

Sutton goes so far as to propose machines as the evolutionary successors of humanity. Yet he believes different scenarios could take place according to how humans will react to them.

He says: "I think that it will be different in different places. There are several possible scenarios and we will probably get several of them. I don't think there will ever be machine overlords, but I do think that current humans will only be midwife to the future leading minds."

So, could machines ever feel? Quite a few researchers working in the field of AI and robotics would say this is entirely possible.

But we should also ask ourselves whether we will actually engineer such machines, and more importantly, if we should.

Shanahan says: "These are difficult questions. Some would say yes as they are the successors to humans and one day will carry our flag onwards to greater things. I definitely see the appeal of that vision, but it's a science fiction vision. When you actually start to think about the reality of that, it's harder to come to terms with, maybe."

Main image credit: Bigstock