Blame Human Lawyers, Not Tech, for Courtroom Blunders: Robin AI Cofounder

To the founders of London-based startup Robin AI, the benefits that artificial intelligence could bring to the legal profession were obvious. But headlines focused on hallucinated case law drafted and submitted to the courts aren’t helping their cause.

Launched by attorney Richard Robinson and computer scientist James Clough in March 2019, Robin AI is designed to streamline contract processes using artificial intelligence—including drafting and reviewing contracts and legal documents. Robinson said he and Clough were inspired by watching developers train AI to play games like AlphaGo and realizing that machines could be taught to understand the law and contracts.

Another reason, he said, was to try to drive down the cost of legal services.

“I worked at two big law firms and in my first week at my law firm, I saw the bill that we were sending to the client with my name on it and nearly fell off my chair,” Robinson told Decrypt in an interview. “I couldn’t believe how expensive it was. My parents couldn’t have afforded an hour of my time, and so I wanted to build a business that tried to use technology to bring down the cost of legal services. That was basically the goal.”

To achieve that goal, Robin AI has raised $26 million in Series B funding, led by Singapore-based investment company Temasek, the company announced on Wednesday. As a launch partner of Claude AI developer Anthropic, Robin AI’s technology is underpinned by Claude 2.1.

Others joining the Series B funding raise include QuantumLight, Plural, and AFG Partners. Robinson said that while Robin AI had a significant war chest from a previous investment round, the amount of demand, particularly from the United States, Robin AI was seeing called for expansion—which in turn called for more investments.

“Most of our customer base has always been in the U.S., but most of our employees have always been in London,” Robinson said. “So it was not straightforward to manage that demand without significant investment in your infrastructure.”

Robinson said that learning more about AI’s potential led to the need for more skilled and expensive talent.

“Then Temasek came along, which just made perfect sense for us,” Robinson said. “They’re a huge Asian fund, and we want to be a global business. So very quickly, we came to a deal.”

When asked why Robin AI decided to get involved with Anthropic instead of competitor OpenAI, Robinson pointed to Anthropic’s stance on safe and responsible AI development and creating tools with strict response limitations, aligning with Robin AI’s safety-focused approach.

“We want [Robin AI] to have tight guardrails,” Robinson said. “That was one thing, and another thing was that we just really liked the team. We’ve collaborated with them on a deep level since they launched, and they’ve been great colleagues and compatriots to us.”

Generative AI launched into the mainstream last year with the public launch of ChatGPT, and many industries, including the legal industry, have turned to the technology. An aspect of artificial intelligence that remains a concern for anyone using the technology is AI hallucinations.

On Sunday, Chief Justice of the U.S. Supreme Court, John Roberts, opined about the potential and risk of AI in the legal system, saying that the technology risks invading privacy and “dehumanizing the law.”

While acknowledging AI’s penchant for hallucination, Robinson notes ongoing research and product design focused on catching these errors. Robinson argued against the notion that AI dehumanizes legal services, saying instead that AI can make legal advice more accessible and affordable, countering existing barriers faced by many.

In his end-of-the-year report, Roberts also pointed to instances where legal professionals used AI and cited non-existent cases because of AI hallucinations.

Last month, a federal judge demanded the legal team for the former attorney for former President Donald Trump, Michael Cohen, show printed proof of the legal cases submitted in court documents after the court said it was unable to verify their existence.

While it’s easy to blame the technology, Robinson put the blame on the lawyers, saying it’s up to them to do their due diligence.

“That’s just bad lawyering,” Robinson said. “I was a litigator; if I cite a case, I need to check that it’s real—it’s my job to do that. That’s what the bar is for. It’s to teach you how you’re supposed to use the tools to administer justice.”

Despite AI hallucinations and the continued claim that AI will take over the workforce, Robinson said he is optimistic about the future of AI in the legal sector but acknowledges that it could lead to fewer lawyers.

“This technology is transformational,” Robinson said. “If the technology is transformational, it is going to eliminate jobs. But I do think that in the long run, lawyers will be significantly more efficient, which means you just need fewer of them to do the same amount of work and I think you’ll see the same thing in every single industry.”

Edited by Ryan Ozawa.