OpenAI’s ChatGPT has opened up artificial intelligence (AI) to the public and highlighted AI’s potential to revolutionise every sector globally, including dispute resolution (DR).

The recent upgrade, GPT-4, is a “multi-modal”, “large language model” encapsulating “generative”, “deep transformative AI technology” with the capacity to create content autonomously. The terminology can be discombobulating but at its simplest, trained AI machines enable users to process masses of information rapidly and produce (largely) plausible text and images within seconds.

However, the technology’s rapid progress poses risks to humanity. Elon Musk and others have: questioned whether humans should be developing “nonhuman minds that might eventually outnumber, outsmart, obsolete and replace us”; warned that Advanced [“human competitive”] AI should be managed with the care and resources commensurate with the profound changes it could trigger; and called for a pause in research (see Pause Giant AI Experiments: An Open Letter). Even OpenAI’s CE, Sam Altman, has recently agreed that the US congress’ concerns for humanity are well founded.

How best to utilise and regulate AI has become a pressing global issue. The UK government considers AI a critical technology for which regulation will be important. It has published a “National AI Strategy” and has “Establish[ed] a pro-innovation approach to regulating AI” (see AI regulation: a pro-innovation approach).

Practical/ethical issues should also concern users. ChatGPT draws on an outdated set of data and its reliance on unmodified/unchecked internet data can give rise to biased, inaccurate and potentially offensive responses – for which it might be hard to establish liability. The lack of transparency about its data sources means users cannot check sources to avoid plagiarism/copyright breaches. Reliance without rigorous checking is inadvisable.

In DR, we already use simple forms of AI (e.g., in document reviews), but AI is a game changer and requires a radical rethink of DR processes. Experienced lawyers are leading the way. Richard Susskind, suggests reframing that rethink. Don’t be bamboozled by the tech: ask instead, how can AI’s power be harnessed to benefit citizens, legal systems and processes? (Tomorrow’s Lawyers (third edition)). We should be focussing on the issues that need solving at the core of DR rather than how to improve current processes.

Sir Geoffrey Vos, in his McNair Lecture, identifies the key issue for dispute resolvers as the challenge of identifying the core issue[s] dividing disputing parties. He believes AI could (and should) do that task quickly.

Vos expects AI to have a “major role in future types of DR”. As context, he points to the fast-approaching changes in trade practice that will include digital currencies and smart contracts that record transactions indelibly on the blockchain. Businesses operating digitally will expect rapid, digital DR to support digitised commercial life.

Vos points to: “on-chain” records that could largely eliminate the need for factual evidence; digital justice systems using AI to make “judicial” decisions (subject to controls); and the creation of a single dataset for DR processes (thereby facilitating disputes graduating from ADR to court).

We are on the cusp of this new DR world. Low value claims are going online (in which AI will surely soon feature) and we already have Digital DR rules. But first, we must grasp the AI nettle. Read, train, experiment, monitor regulation and understand AI’s limitations and potential. The possibility of AI replacing lawyers is a question for another day. For now, we need to ensure that DR digitises at the same rate as businesses. If we don’t adapt, jobs really will disappear.

This article was first published by the Construction Law magazine on 3 July 2023.