By Cade Metz
In August, researchers from the Allen Institute for Artificial Intelligence, a lab based in Seattle, Washington state, unveiled an English test for computers. It examined whether machines could complete sentences like this one:
On stage, a woman takes a seat at the piano. She
a) sits on a bench as her sister plays with the doll.
b) smiles with someone as the music plays.
c) is in the crowd, watching the dancers.
d) nervously sets her fingers on the keys.
For you, an easy question. But for a computer, it was pretty hard. While humans answered more than 88 percent of the test questions correctly, the lab’s A.I. systems hovered around 60 percent. Among experts, that was impressive.
Two months later, a team of Google researchers revealed a system called Bert. Its improved technology answered those questions just as well as humans did — and it was not even designed to take the test.
Bert’s arrival punctuated a significant development in artificial intelligence. Over the last several months, researchers have shown that computer systems can learn the vagaries of language and apply that knowledge to specific tasks.
These systems may improve technology as diverse as digital assistants like Alexa and Google Home as well as software that automatically analyzes documents at law firms and other businesses.
“Each time we build new ways of doing something close to human level, it allows us to automate or augment human labor,” said Jeremy Howard, the founder of Fast.ai, a lab based in San Francisco, California.
It may even lead to technology that can — finally — carry on a decent conversation.
But on social media services, this research could also lead to more convincing bots designed to fool us into thinking they are human, Mr. Howard said.
Researchers have shown that rapidly improving A.I. techniques can facilitate the creation of fake images that look real. As these kinds of technologies move into the language field, Mr. Howard said, we may need to be more skeptical than ever online.
These new language systems learn by analyzing millions of sentences written by humans. These new language systems, made possible by gains in computer processing power, learn by analyzing millions of sentences written by humans. A system built by OpenAI, a lab based in San Francisco, analyzed thousands of books. Google’s Bert analyzed these same books plus all of Wikipedia.
Each system developed a particular skill. OpenAI’s technology learned to guess the next word in a sentence. Bert learned to guess missing words anywhere in a sentence. But in mastering these specific tasks, they also learned about how language is pieced together.
If Bert can guess the missing words in millions of sentences, it can also understand many of the fundamental relationships between words in the English language, said Jacob Devlin, the Google researcher who oversaw the creation of Bert.
This kind of technology is “a step toward a lot of still-faraway goals in A.I., like technologies that can summarize and synthesize big, messy collections of information to help people make important decisions,” said Sam Bowman, a professor at New York University.
Last month, Google “open sourced” its Bert system, so others can apply it to additional tasks. Mr. Devlin and his colleagues have already trained it in 102 languages.
But there is reason for skepticism that this technology can keep improving quickly because researchers tend to focus on the tasks they can make progress on and avoid the ones they can’t, said Gary Marcus, a New York University psychology professor. “These systems are still a really long way from truly understanding running prose,” he said.
© 2018 New York Times News Service