AI: why robots find what is difficult easy and vice versa – 06/14/2023 – Tech

Will we ever be able to create a robot with the same capabilities as humans?

The surprising emergence of ChatGPT and other AI (artificial intelligence) programs makes this question increasingly relevant. More than that, the imagination of engineers trying to create a robot that thinks and acts like a human being is even sharper.

We’ve already been able to mimic our brain’s complex reasoning and even creativity systems, especially with AI. But robots still can’t tie a shoe.

Robotics and artificial intelligence can make rational thinking require less computing processes, while apparently simpler acts that are easily performed by humans, such as tying shoes or picking up a dropped bag, require enormous computational effort .

This phenomenon is known as the Moravec paradox. For many experts, it is the reason why it has not yet been possible to build a fully intelligent robot.

“Human beings have taken hundreds of thousands of years of evolution to do simple things like, for example, maintaining balance”, explains researcher in robotics Gonzalo Zabala, from the Universidad Aberta Interamericana, in Argentina. “So, reproducing all these processes at a computational level is almost impossible at the moment.”

Zabala points out, in an interview with BBC News Mundo (BBC’s Spanish-language service), that the opposite also happens with processes that are the result of reasoning.

“How long can we speak of an intelligent man, of reason?”, he asks. “Compared to other evolutionary processes, the time is much, much less, and therefore we can encode and reproduce them more successfully.”

Hans Moravec and Alan Turing

One of the precursors of artificial intelligence was the British scientist Alan Turing (1912-1954).

One of several studies published by Turing in his short but prodigious career lists a series of questions that would serve to distinguish, in a theoretical case, a robot from a human being.

This method has been guiding engineers and theorists for the development of artificial intelligence since it was formulated in the 1950s.

As the robotics professor Rodney Brooks, from the Massachusetts Institute of Technology (MIT) pointed out, what happened was that the engineers concentrated on creating programs or devices that managed to “fool” their interlocutors, adequately answering the questions of the Turing test so they could pass for human beings.

However, in the late 1970s, this approach began to present a problem. The logical answers did not develop anything original and the path indicated by Turing was starting to not leave many exits.

“The funding for the research itself was suspended, as the path to be followed was not clear and progress was not observed”, according to Brooks.

Therefore, scientists went in search of alternatives to advance the development of artificial intelligence.

“The path chosen was to create circuits similar to those of the human brain”, explains Zabala. “Not a robot that responded with logic, but a circuit that could think.”

It was then that the contradiction arose, still unresolved: while artificial intelligence processes were created with relative ease, the basic functions of human beings were practically impossible to be recreated in a robot.

This fact was widely observed until the end of the 1980s by specialists in robotics, such as Rodney Brooks himself, the Austrian Hans Moravec and the North American Marvin Minsky. However, it was Moravec, a professor at Carnegie Mellon University in Pittsburgh (United States), who was responsible for presenting the best exposition on the subject in 1988, based on the work of three colleagues.

According to Moravec, “It is comparatively easy to get computers to perform at an adult level on intelligence tests or playing chess, but it is difficult or impossible to give them the capabilities of a one-year-old boy in terms of perception and mobility. “

That is, robots can be as smart as they are incapable.

“What Moravec’s paradox did was make sense of what was being observed,” says Zabala. “When you name a problem, you indicate possible solutions.”

“When you reach this point, something very interesting begins, which is getting to know each other better so that we can reproduce ourselves in robots: knowing how we maintain balance, how we learn to drive and so on”, he adds.

sensitive robots

The three researchers —Moravec, Brooks and Minsky— developed projects aimed at clarifying the paradox.

Brooks, for example, worked with the American company Boston Dynamics and with another one he founded called iRobots.

The principle, according to him, is summarized in a direct premise: “if we want to build a robot with human intelligence, we must first build a robot with human anatomy”. From there, robot projects were developed that presented an aspect closer to ours.

A team of European scientists, for example, developed a prototype known as ECCERobot, equipped with a complete thermoplastic skeleton, with vertebrae, phalanges and ribcage.

The ECCERobot has the same degrees of movement as the human torso and, most importantly, all of its parts are packed with sensors.

But the very scientists who developed the robot pointed out that the main inconvenience was not overcome. The ECCERobot is so complex that it can barely hold a glass. Therefore, it cannot be expected to have intelligent behavior.

“Building an intelligent humanoid robot that can interact seamlessly with humans and human environments in a natural way will require advances in computing and battery efficiency, not to mention a quantum leap in sensory equipment”, says Rolf Pfeifer, project coordinator of the ECCERobot.

“A really key development will be the skin,” he says. “The skin is extremely important in the development of intelligence because it provides very rich sensory patterns: touch, temperature, pain, all at once.”

However, experts point out that, despite the problems presented by the Moravec paradox, the construction of an intelligent robot like a human being is a possibility, albeit a distant one.

“What Moravec’s paradox did was to put a problem in evidence, for researchers to look for solutions”, explains Zabala. “One of them, without a doubt, is what we are seeing with the artificial intelligence revolution, where we have taken a step towards creation, not just logical answers.”

For the expert, this revolution is clearly not a threat to the extinction of the human species.

“I do not believe that it means the end, as several analysts have stated. It is a tool that will facilitate many processes in the future”, concludes Zabala.

This text was published here.