Can intelligence be separated from the body? – 04/21/2023 – Science

What is the relationship between mind and body?

Maybe the mind is like a video game controller, moving the body around the world, taking it for a walk. Or maybe the body manipulates the mind with hunger, drowsiness and anxiety, something like a river guiding a canoe. Is the mind like electromagnetic waves, flowing in and out of our bodies like light bulbs? Or is the mind a car on the road? A ghost in the machine?

Perhaps no metaphor fits perfectly because there is no distinction between mind and body. There is only one experience, or some kind of physical process, one gestalt.

These questions, which have plagued philosophers for centuries, are gaining new urgency as sophisticated machines with artificial intelligence begin to infiltrate society. Chatbots like OpenAI’s GPT-4 and Google’s Bard have minds, in a sense. Trained in vast frameworks of human language, they learned to generate new combinations of text, images and even videos. When prepared in the right way, they can express desires, beliefs, hopes, intentions, love. They can speak of introspection and doubt, self-confidence and regret.

But some AI researchers say the technology won’t achieve true intelligence, or true understanding of the world, until it works with a body capable of sensing, reacting and sensing its surroundings. For them, to speak of disembodied intelligent minds is misleading, even dangerous. AI that is unable to explore the world and learn its limits, just as children discover what they can and cannot do, can make life-threatening mistakes and pursue its goals at the risk of human well-being.

“The body, quite simply, is the basis for intelligent, cautious action,” said Joshua Bongard, a roboticist at the University of Vermont. “As far as I can see, this is the only way to secure AI.”

In a laboratory in Pasadena, Calif., a small team of engineers has spent the last few years developing one of the first combinations of a large language model with a body: a turquoise robot called Moxie. About the size of a child, Moxie has a teardrop-shaped head, soft hands and lively green eyes. Inside its hard plastic body is a computer processor that runs the same kind of software as ChatGPT and GPT-4. Moxie’s makers, part of a startup called Empowered, describe the device as “the world’s first AI robot friend.”

The bot was conceived in 2017 to help children with developmental disabilities practice emotional awareness and communication skills. When someone speaks to Moxie, its processor converts the sound into text and feeds the text into a large language model, which in turn generates a verbal and physical response. Moxie’s eyes can move to comfort you for the loss of your dog and it can smile to encourage you to go to school. The robot also has sensors that pick up visual cues and respond to its body language, mimicking and learning from the behavior of the people around it.

“It’s almost like wireless communication between humans,” said Paolo Pirjanian, roboticist and founder of Empowered. “You literally start to feel it in your body.” Over time, he said, the robot gets better at this kind of exchange, like a friend getting to know you.

Researchers at Alphabet, Google’s parent company, have taken a similar approach to integrating large language models with physical machines. In March, the company announced the success of a robot called PaLM-E, capable of absorbing visual resources from its environment and information about the position of its own body and translating everything into natural language. This allowed the robot to represent where it was in space in relation to other things, and eventually open a drawer and grab a bag of chips.

Robots of this type, experts say, will be able to perform basic tasks without special programming. They could pour you a glass of water, prepare your lunch or pick you up after a fall, all in response to a series of simple commands.

But many researchers doubt that machine minds, when structured in this modular way, will ever be truly connected to the physical world; therefore, they will never be able to display crucial aspects of human intelligence.

Boyuan Chen, a roboticist at Duke University who is working on developing intelligent robots, pointed out that the human mind — or any other animal mind, for that matter — is inextricable from the body’s actions and reactions to the real world, shaped over millions of years. years of evolution. Human babies learn to pick up objects long before they learn language.

The artificially intelligent robot mind, in contrast, was built entirely on language, and it often makes commonsense mistakes arising from training procedures. A deeper connection between the physicist and the theorist is missing, Chen said. “I believe that intelligence cannot be born without having the prospect of physical incarnations.”

Some experts, including Pirjanian, recently voiced concern in a letter about the possibility of creating an AI that could selflessly crush humans in pursuit of some goal (such as efficiently producing paper clips) or that could be harnessed for evil purposes (such as advertising campaigns). misinformation). The letter asked for a temporary break in training models more powerful than the GPT-4.

Bongard, like many other scientists in the field, felt that the letter asking for a break from research might provoke uninformed scaremongering. But he is concerned about the dangers of our ever-evolving technology, and believes the only way to imbue embedded AI with a robust understanding of its own limitations is to rely on constant trial and error to move around in the real world.

Start with simple robots, he said, “and as they demonstrate they can do things safely you let them have more arms, more legs, give them more tools.”

And perhaps, with the help of a body, a true artificial mind will emerge.

^{Translated by Luiz Roberto M. Gonçalves}