Robots And
Artificial Intelligence

Robots and Artificial Intelligence are often associated together. Indeed, even advanced robots currently have very limited autonomy. Why can't a robot just open a drawer after you've just showed it how to? They can make super accurate movements to grasp an object, but they can barely adapt if the object is in a different position.

At the same time, you have super "intelligent" AIs like IBM's Watson. Watson knows a lot of common knowledge and can answer questions about a lot of topics. But still it won't ever tell you anything if you don't ask. It is just a new high-tech encyclopedia, but is it "intelligent"? Is it "thinking"? Or does it just give you the answer of the game Jeopardy just like your computer gives you the answer of 2+2?

1) Can robots be intelligent?

Obviously, when robots and Artificial Intelligence are combined together, you expect to get super clever robots. But there is a mystery about what is intelligence, and how to get it. Obviously, the capacity to solve mathematics equations doesn't make a robot or computer intelligent. Nor does the ability to remember huge amount of data. Robots can find the best way to go from A to B in a cluttered room, but if you move anything it will get lost and have to re-plan its whole movement all over again.

Actually, while robots can do some very hard thinking tasks, they are lost when uncertainty arises. Some will say it is their inability to learn, or to generalize, or to adapt. But in the end it is just not working. Mathematics is an exact science, but Mother Nature is too complex to be understood exactly. Since robots can only do what they understand mathematically, they don't go far.

But what if we could imitate how humans are learning, thinking and planning and implement it in robots? That is the next step.

2) Can robots feel emotions?

Nao looks intelligent. But is he?
Robots and Artificial Intelligence: Nao Reading

Try to remember what you did yesterday. What comes first? You may remember about an argument with you wife, or the great joke by a colleague. Or it can be the best moment of the movie. A common point of your most salient memories will usually be that you felt a big surge of emotions. Did you feel happy, scared, angry? Emotions play a central role to your life.

Think about this question: Why or when do you feel a surge of emotions?

  • You got scared when you were nearly hit by a car? Don't do it again.
  • You were happy when you scored in sport? Use this strategy again.

This is all learning by experience. How can it help you to make a robot? Well if robots and Artificial Intelligence can simulate emotions, they may be able to single out the important information they need. They can save memory and processing time by learning from the best information. Marvin Minsky shows how it could be used to build better robots and artificial intelligence in his books "Society of Agents" and "The Emotion Machine".

Emotions are even more powerful than that. Damasio writes in "Descartes' Error" that emotions are also crucial to take a decision. Consider the World champion chess playing computer. While a human chess master only considers a few (less than 10) moves at each turn, the computer makes a tree search over several thousand of possibilities. How can a chess master know which are the best moves? The answer is that while he scans the chessboard, its emotions will guide him toward the most promising choices.

Researchers have been working on robots that can show emotions. For example, the little robot Kismet can use emotions to show its needs and interact more naturally with the care owner. But these needs here are artificial. There were chosen to boost interaction (need to play with toys or to see faces, to name a few). Damasio also writes about the origin of the emotions being in the body. If you follow his reasoning, you will find that you should build the body first and the brain can only grow "from it", not "out of it".

Kismet robot and artificial intelligence

Kismet uses emotions to interact naturally.
From J. C. Benedict, on Wikipedia

3) There is still hope

The more I study robots and human intelligence, the more I am impressed by what we can do every day without even thinking about it. But as we study the human mind, and deepen our understanding of it, it becomes clear that we can simulate more and more of its processes.

One day, we will have programs that can simulate enough of the human mind's feature that they look intelligent just like us. Different people will give you different answers, but I believe it doesn't really matter. Descartes explained that anyway: you can never prove that someone or something else is actually thinking, since you can look inside. The point is that robots and artificial intelligence will extend their possibilities by thinking like us. A little like what you can see in robot movies.

A few references to learn more about this:

  • Descartes' Error, by Damasio, will explain you how the body and the mind are connected by emotions. (high school biology background about neurons recommended)
  • The Emotion Machine, by Minsky, gives a lot of ideas on how emotions could be used to enhance robots. (good if you are in Computer Science)
  • Machines Who Think, by McCorduck, tells you the history of robots and artificial intelligence in great details and with catching stories. (for everyone who likes the topic)

Click here to go back from Robots and Artificial Intelligence to the Ultimate Robot FAQ.

Click here to go the Personal Robot Home Page.

New! Comments

Have your say about what you just read! Leave me a comment in the box below.

Search The site

Robot Tips

Subscribe to Robot Tips and get the best tips to choose, dream, learn and build robots.

Enter Your E-mail Address
Enter Your First Name

Don't worry — your e-mail address is totally secure.
I promise to use it only to send you Robot Tips.

Find out more...