Robot-Artificial-Intelligence

There is a danger when it comes to using artificial intelligence in legal. It's just probably not the danger you're thinking of, says Shawnna Hoffman, co-leader at IBM Global Cognitive Legal's Watson AI: “Robots are not actually coming for your job. Robots are here to augment you and make your life easier.”

Instead, there may be different issues with using AI, particularly ones that may not occur to those outside of the data science world.

The Sunday afternoon keynote of the Association of Legal Technologists' second ctrl ALT del conference, titled “The Future is Now,” explored some of the opportunities and risks that come with the rapid emergence of new technologies like AI and blockchain. To be certain, the applications of these technologies in the general marketplace are endless—Hoffman touched on applications ranging from autonomous cars to even an Ohio judge using Watson in juvenile court.

But no technology, even one as advanced as Watson, is foolproof. Hoffman touched on a few issues that could be instructive for those looking to technologically-advanced solutions in the new legal world.

1. Clean Data: It's a variation on the garbage in, garbage out philosophy: AI “can learn bad things if you teach it bad things,” Hoffman said.

One of the biggest issues that Hoffman said she had when starting with IBM Watson was data: not having enough data, finding the right data, and being able to train the data in a way that it produced viable results. “I can't say this enough: Your data has to be good data to train the system. It has to be cleansed,” she explained.

That's why, when implementing an AI solution, the team training the system is paramount. For legal especially, it's important to note that this includes more than attorneys—data scientists and others are important for creating a well-rounded system. “It's like a kid,” Hoffman said. “It will only learn what you teach it, so you need a variety… for it to learn everything that you need.”

2. Understanding the Technology: When talking with attorneys and paralegals, Hoffman said she often finds a state of alarm when it comes to blockchain-enabled smart contracts. They say, “'That's the one that's going to take away my job.' I can promise you, it's not going to.”

She explained it's optimal to not even view these so-called “smart contracts” as either smart or contracts at all. Instead, they're a simply a lever for a contract rather than a full contract itself; it simply will move when a certain set of conditions is met to execute a deal.

In fact, contrary to popular opinion, Hoffman believes the use of smart contracts will make “attorneys actually busier,” especially with the introduction of coders and legal technologists to the mix. “You're starting to see these teams forming just to handle the new technology.”

3. Forward Thinking About Regulations: The pace of technology continues to advance, and the law sometimes has a problem keeping up. But that doesn't mean that regulations will never be there.

“Because of the way we're building these AI systems… we're working with IT staff that love to push the limit,” Hoffman noted. Corporate counsel, meanwhile, may not have a full understanding of the legal risks that AI brings, and they often don't know to look for forward-looking regulations on newly-developed tools.

“The only way they're going to realize that is if the outside counsel they're working with push them,” Hoffman said, adding that one of her main wishes is to see these sorts of conversations increase in between inside and outside counsel.

4. Know Your Business Proposition: Especially given the popularity of Watson, Hoffman said she and the IBM team are approached often about implementing the technology for different use cases. However, just because a tool is available doesn't mean it'll be effective.

“If I don't see a way for an ROI within 8 weeks, we won't do that project,” she said.

To that end, she posted for the crowd a list of attributes for a good AI or blockchain use case. The three questions she said to ask are:

  • What are the pain points your business, employees and/or clients face today?
  • What is the business value of this use case?
  • Has the content that will fuel the solution been identified, and can it be secured/licensed for the intended use?

Without thinking through these problems, she said, it's tough to get that quick ROI. With that in mind, it may be best to start small and experience some growth before tackling a large, transformative project.