Can you believe that? Scientists working on giving emotions to robots!
That's what Richard Gray reported on telegraph.co.uk last week. As much as I love the idea of robots with intelligence and feelings, I don't agree with the approach of this article.
The news is that scientists announce domestic robots in 10 years which will be capable of multiple tasks, and emotions will be introduced into these robots to help them "make better decisions, learn more effectively and interact more appropriately".
Although I agree with that vision, I am troubled by the way these groundbreaking innovations are introduced. It is all focused on androids and human-like tasks and emotions. But if any emotions are happening, they can only be a software feature. And that article says nothing of software. I mean if software can have emotions in a few years then it will happen in your PC, or in your web browser, long before it makes any sense inside an actual robot. And the "actual robot" concept raises many other questions in the field of hardware, commodity, safety etc. I mean imitating the human body and imitating the human mind are two very different things. I wish this distinction were better addressed when predicting the future.
This reminds me of the disappointment I had when I saw A.I. (the movie) a few years ago. The story is about a "child" android which is introduced into a human family. This kid-bot actually takes part in a mother-child relationship, and the movie raises questions and proposes answers to what this implies about the future of humankind.
What frustrated me in A.I. is that it wasn't clear if the child-bot was an automaton imitating love or an intelligent robot actually feeling love. What is the difference between showing emotions and actually feeling emotions? That is the question.
The answer lies in the concept of conscience. The movie didn't make any case of this difference, and treated stupid automata such as the teddy bear on the same grounds as seemingly (or actually?) conscious robots (the kid-bot).
Back to the telegraph's article: I doubt that in 10 years from now we will have achieved seemingly conscious behaviors in software. Passing the Turing Test for real will still be at least 10 years away, and in that regard these robots will still be narrow-AI-minded and emotionless.
As a conclusion I guess that the scientists referenced in the article mean we will implement emotion-based decision-making to enhance these robots. To my mind this is not the same thing as emotions.