emiliau - Fotolia

Artificial intelligence to initiate and manage litigations after ‘landmark’ approval

The Garfield Project created an artificial intelligence-based litigation service for small claims, but how far can the technology reach in the legal sector

The regulatory approval of an artificial intelligence (AI)-based legal firm, which can complete a legal claim process with virtually no human involvement, is a “landmark” for an industry ready for change, according to the Solicitors Regulation Authority.

Garfield AI, as the company is known, was founded by a technologist and a lawyer with the initial aim of providing a service for small businesses to claim billions of pounds in unpaid invoices.

The AI litigation assistant, which initiates and manages small claims litigation, received approval from the SRA last month, with the organisation hailing the arrival of “the first law firm providing legal services through AI”.

Using Garfield, the claimants or their lawyers upload the relevant information and authorise each step in the process. For example, the AI platform can create letters before action, draft claims and particulars of claim forms, apply for default judgments, handle settlements, and prepare for trial.

According to the company, it “uses legal best practice to ensure all claims are prepared correctly and the claims process is followed (and tracked) precisely – significantly reducing errors and delays”.

This is just the beginning for the company. Daniel Long, Garfield AI co-founder, told Computer Weekly that as a technologist and not a lawyer, he had never considered the use case was a possibility on an AI platform, but his business partner, lawyer Philip Young, was determined to offer a service for small businesses struggling to recover money owed to them.

Long said: “Although the small claims court is there for them, it is not economical to hire a lawyer.”

He said the service was “a sensible place to start” for an AI-only law firm. “These are not complex cases and do not require an understanding of case law and detailed arguments,” said Long. “It’s largely just extracting some information like what is owed, what work was done, populating some forms and probably some letters with that information.”

It took the co-founders the first year to build a minimum viable product, but after adding three experts, product development accelerated. The service is not fully autonomous and only proceeds to the next stage of a claim when the customer is happy, and a final check is by a legal expert.

Just the beginning

Law firms are currently using AI in their businesses, performing tasks such as document reviews, but Garfield AI’s approval provides a legal service directly to people and businesses. Paul Philip, CEO at the SRA, described the regulatory approval as a “landmark moment for legal services in this country”.

Long has been surprised by how open the judiciary is to tech development. “Senior members of the judiciary are banging the drum, saying ‘the legal sector must embrace AI’,” he said.

The SRA is calling for more development of AI-based legal services. “We are encouraging the development of new approaches and models due to the potential consumer benefits,” it said. “AI-driven legal services could deliver better, quicker and more affordable legal services.”

Regulatory approval took about eight months, according to Long. He said the co-founders had already been talking about the concept with senior members of the judiciary to “gauge their views” and that “there was enthusiasm from them”.

Long added that they could see there was a compelling argument for this in terms of improving access to justice. “But they do not want to introduce something that could harm individuals,” he said. “We also presented to the department for justice select committee and one of their first questions was, “How far can you envisage this model going?’”

Long told Computer Weekly that the company plans to apply AI to more legal services, and that the company could initially expand by taking on larger cases.

Risk assessment

The SRA said: “There are potential risks to the public, so we have been making sure there are appropriate checks in place to make sure that consumer protections are not diluted.”

For example, the SRA wanted reassurance that the system would not experience “hallucinations”.

Philip at the SRA said: “With so many people and small businesses struggling to access legal services, we cannot afford to pull up the drawbridge on innovations that could have big public benefits. Responsible use of AI by law firms could improve legal services, while making them easier to access and more affordable.”

He added that the risks of an AI-driven law firm are “novel”, requiring the regulator’s close attention. “As this is likely to be the first of many AI-driven law firms, we will be monitoring progress of this new model closely, so we can both manage the risks and realise the benefits to consumers,” said Philip.

Mistakes, known as hallucinations, are something that David Enright, partner at Howe & Co Solicitors, is concerned about when it comes to applying AI in the legal setting. “If you’re relying on AI rather than a lawyer, who do you complain to when things go wrong?” he said.

Enright added that final checks by lawyers will have to be thorough as, unlike an experienced practitioner, AI is not able to “think outside the box” and, for example, spot other potential claims.

He welcomed improved access to justice because the alternative for many is “no justice”.

Mark Lewis, lawyer at Stephenson Harwood, said the senior judiciary are “acutely aware that access to justice is difficult and even impossible for many people these days”.

“I think anything like Garfield that addresses the small end of the market and improves access to justice would be welcomed,” he said.

Lewis added that there are some in the legal sector advocating for AI to perform judicial functions. “In other words, decide the merits of cases,” he said. “I think we’re some way off, but one view is that it would be a logical step.”

Read more on Artificial intelligence, automation and robotics