In 2020, OpenAI, a research lab in San Francisco, revealed a system called GPT-3. It is what artificial intelligence researchers call a neural network, after the web of neurons in the human brain. A neural network is really a mathematical system that learns skills by locating patterns in vast amounts of digital data. By analyzing thousands of cat photos, for instance, it can learn to recognize a cat. "We call it ‘artificial intelligence, ' but a better name might be ‘finding statistical patterns from large data sets', " said Dr. Gopnik, the Berkeley professor.

More recently, researchers at places like Google and OpenAI began building neural networks that learned from enormous amounts of prose, including digital books and Wikipedia articles by the thousands. GPT-3 is an example. As it analyzed all that digital text, it built what you might call a mathematical map of human language — more than 175 billion data points that describe how we piece words together. Using this map, it can perform many different tasks, like penning speeches, writing computer programs and having a conversation. 

But there are limitations. If you ask GPT-3 for 10 speeches in the voice of Mark Twain, it might give you five that sound remarkably like the famous writer—and five others that come nowhere close. Computer programmers use the technology to create small snippets（一小段） of code they can slip into larger programs, but more often than not they have to edit and adjust whatever it gives them. 

Still, Dr. Gopnik described this kind of system as intelligent. "It is not intelligent in the way humans are. It is like an unfamiliar form of intelligence, " he said. "But it still counts. "

Dr. Gopnik and many others in the field are confident that they are on a path to building a machine that can do anything the human brain can do. This confidence shines through when they discuss current technologies. He admits that some A. I. researchers "struggle to differentiate between reality and science fiction. " But he believes these researchers still serve a valuable role. "They help us dream of the full range of the possible, " he said.

Perhaps they do. But for the rest of us, these dreams can get in the way of the issues that deserve our attention.