Ruth Bader Ginsburg’s Artificial Intelligence Model Released

Placeholder while article actions are loading

When reports surfaced in May that the Supreme Court wanted to destroy abortion rights, many wondered how the late Judge Ruth Bader Ginsburg might have reacted. Now they don’t have to wait.

“I think they are wrong about the law, but not the facts,” said a simulation of Ginsburg, who died in 2020, when asked about the upcoming Supreme Court decision on Roe v. Wade

The answer came not from Ginsburg’s numerous court opinions, but from an artificial intelligence model of the late justice released Tuesday. “Whether it’s good or bad, it’s settled and so it’s not my business to think about it,” concluded the RBG bot.

The model, called Ask Ruth Bader Ginsburg, is based on 27 years of Ginsburg’s legal writings before the Supreme Court, along with a large number of news interviews and public speeches. A team at the Israeli artificial intelligence company called AI21 Labs fed this record into a complex language processing program, which the AI ​​claims was able to predict how Ginsburg would respond to queries.

“We wanted to pay tribute to a great thinker and leader with a fun digital experience,” says the company on the AI ​​app website† “It’s important to remember that AI in general, and language models in particular, still have limitations.”

The tool comes amid fierce discussions about the ethics of creating technology that replicates human life, especially when the people involved are not around to provide input. But the makers argue their invention is a useful and easy-to-use tool to help ordinary people, who may not know much about technology, understand how the field of artificial intelligence is progressing.

The Google engineer who thinks the company’s AI has come to life

“There aren’t many places where the general public can play with real AI,” said Yoav Shoham, co-founder of AI21 Labs. “But now it’s possible.”

In recent years, research labs and companies around the world have raced to build technology that replicates or surpasses human intelligence. provide ways for people to research and interact with their work on the go.

OpenAI, an artificial intelligence company backed by Elon Musk, has unveiled a text generator called GPT-3 that can write movie scripts and supports an image generator, DALL-E 2, that translates text commands into inventive, sometimes psychedelic, images. In 2020, Shoham’s company created Wordune, a tool that suggests different ways to write sentences. A year later, they followed the release with Wordune Read, which sums up the main lines of long, dense passages.

But as AI technology has gotten better, Shoham said many are divided around the field. “People project all kinds of things [thoughts] on … automation that has nothing to do with reality,” he said. “I don’t want people to be disappointed by the underperformance of today’s AI and I don’t want them to spread fear.”

The general public, he said, must make a decision, and his team’s RBG model is an accessible, practical way of interacting with the technology.

To build it, the researchers used Jurassic-1, a neural network they created that analyzes large amounts of data and develops its own language to spew results to questions or clues. Neural networks are computer architectures that attempt to mimic the human brain by processing information.

They fed the model with about 600,000 words of Ginsburg and created a tool that allows anyone to ask questions, to which it provides answers based on the vast wealth of writing. “It gives you access to the kind of wisdom that someone possesses that we hold dear,” Shoham said.

AI models are beating humans at reading comprehension, but they still have a long way to go

Paul Schiff Berman, a law professor at George Washington University who worked for Ginsburg from 1997 to 1998, said he was amused when he saw the bot.

Immediately, he tried to ask it a question he would have been interested in getting Ginsburg’s opinion: “Should federal courts delay the factual findings of state courts?”

According to Berman, the response left a lot to be desired. The model didn’t answer the question directly, and the answer implied that Ginsburg didn’t believe in the legal concept of reverence, which isn’t true, he said. Berman also noted that the model underperformed by mimicking her unique speaking and writing style.

“I would have thought this was something the AI ​​could have imitated better,” he said. “If this is the best, [technology] can do, we still have a way to go.”

Meanwhile, several AI technology experts have expressed concerns about the experiment.

Emily Bender, a linguistics professor at the University of Washington, said she acknowledges that the experiment’s creators come from a place of respect for Ginsburg, but insinuating that the technology can think and reason like the late justice is not correct. “It can spit out words and the style of those words will be determined by the style of the text they entered into it, but it doesn’t do any kind of reasoning,” she said.

Bender added that linguistic research shows that when people encounter “cohesive texts” on a topic they care about, there is a risk that they take it seriously when they don’t. “People could use this to argue in the world and say, ‘Well, RBG would have said,’ this AI [model] told me that.”

The military wants AI to replace human decision-making in combat

Meredith Broussard, an associate professor and researcher in artificial intelligence at New York University, said the bot is compelling, but not to be confused with real legal advice. “It’s a lot of fun to play with, but we shouldn’t take it so seriously or pretend it’s a lawyer,” she said. (AI21 states that the model is “just an experiment” and that it may give inaccurate answers that should be taken with “a grain of salt”.)

Broussard added that the technology doesn’t appear to be much more advanced than ELIZA, a chatbot created by MIT researchers in the 1960s, where a computer program replicated a therapist well enough to trick people into thinking it was human. She added that there could be a limit to how advanced this kind of artificial intelligence technology can ever get.

“There’s a ceiling on the technology because it’s not a brain, it’s a machine,” she said. “And it’s just doing math.”

Leave a Comment

Your email address will not be published.