Revolutionary AI technology set to revolutionize commercial contracts – is your business ready?

Companies have long been looking for digital solutions for managing their legal contracts, which can cover everything from supply agreements to non-disclosure agreements and government outsourcing. And accurate and efficient systems are important because, as Christina Demetriades, general counsel for Europe at consultancy Accenture says: “The lifeblood of all companies is the contract.”

Over the past decade, many of the more time-consuming tasks involved, such as reviewing large contract documents to find clauses that contravene legal policy or verifying compliance with financial and data regulations, have been automated through the use of artificial intelligence. It is included in the contract lifecycle management software, which helps draft, archive, search and manage a company’s agreements. Among the advantages is the simple speed. “Our platform produces in minutes what an attorney typically takes four hours,” says Tim Pullan, chief executive officer of ThoughtRiver, which writes contract software.

But, now, companies and their software vendors are turning to new generative AI tools to automate contract review. Since AI chatbots like OpenAI’s ChatGPT were launched late last year, business interest has grown in broader applications of this new technology, which uses natural language processing to a large extent of data to learn how to produce new texts and images.

What [generative AI] is going to do is be a digital collaborator

The noise around generative AI has also raised fears that it will take jobs away from lawyers. What if the new technology was so efficient that companies decided they needed less internal advice? Second research published this year by Princeton University, the University of Pennsylvania and New York University, legal services are now the sector most at risk from this latest wave of artificial intelligence.

However, Ryan O’Leary, a legal tech expert at research firm IDC, believes such fears are exaggerated: “Really, what he’s going to do is be a digital collaborator,” he argues.

However, AI may prove to be a faster worker. In April, San Francisco-based legal technology provider Ironclad released a contract negotiation tool, called AI Assist, that incorporates OpenAI’s GPT-4 technology. He can read an NDA, compare it to the client’s legal guidelines for contracts, and suggest changes. “The statutory review of contracts is done 20 times more efficiently using [generative] AI and automation,” says CEO Jason Boehmig, a former corporate lawyer at a Silicon Valley law firm.

Basically, however, users will have the ability to train the Ironclad software using their own agreements and clauses, teaching it how to point out irregularities.

At consulting firm Accenture, where legal professionals use AI extensively for legal work, including contract review, signing and monitoring, Demetriades says the incorporation of generative AI is in an early stage. But she is optimistic about her potential. The legal department calculates that the use of generative AI will bring productivity gains of 15% to 20% within about three years, he says. And “it will also increase the skills of legal professionals”.

Shahid Rayyaz, a lawyer at British engineering consultancy Hoare Lea, a client of ThoughtRiver, is already using AI for contract work. This should reduce contract review team costs by about a third a year over five years, he says.

Savings will come from increased productivity and less spend on outside law firms, without reducing team headcount. “Time and cost were the two biggest savings,” notes Rayyaz.

Culture clashes and “hallucinations”

Even so, it is already clear that the novelty, power and complexity of generative AI also creates challenges.

Only about 54% of in-house lawyers in the US, UK and Canada were interviewed by the Thomson Reuters Institute researchers in April, said ChatGPT/generative AI should be used for legal work. Their concerns included accuracy, privacy, confidentiality and security.

Generative AI is great for helping busy lawyers truly understand the context of the work they are doing

Demetriades shares some of their concerns, pointing, for example, to a potential “culture clash” if lawyers’ need to “get things right” conflicts with the experimentation requirement for generative AI.

And the emergence of flaws in generative AI, such as ‘hallucinations’ – the fabrication of ‘facts’ – mean that it requires close supervision, or so-called guardrails, when it is used for contracts.

Liz Grennan, an associate partner at consultancy firm McKinsey & Co, and an expert on law and technology, warns that generative AI is not suited to precision drafting, because just one wrong word or phrase inserted unnoticed could have costly repercussions. But she believes that “it’s great for helping lawyers who are busy [to] really understand the context of the work they are doing.”

Another consideration for legal departments is that automation typically leads to an expansion in contract size. “[When] attorneys can produce content more easily, it tends to produce longer contracts,” ThoughtRiver’s Pullan points out.

Furthermore, it can take months to fine-tune the big language models – the training and the data – used in AI systems to weed out the false data and educate them on the quirks of a company’s contractual and non-negotiable principles. However, such tailored versions could potentially undermine the benefits of scale: AI vendors, such as ThoughtRiver, say they may charge extra for such customization.

That said, emerging standards for legal data, such as Salts (Advancement of Standards for the Legal Industry) e OneNDAan open source standard for non-disclosure agreements, can alleviate integration problems.

It is likely, therefore, that Generative AI will both build on advances in automating contract drafting and review, while also requiring skillful monitoring. As IDC’s O’Leary puts it: “The negotiation and interpretation of clauses and the escalation of litigation” will still have to be done by attorneys.