Lawyers are fast becoming among the biggest users of artificial intelligence. Everyday chatbots like ChatGPT and other specialized chatbots created specifically for lawyers have been in the news, for both good and bad reasons.
The good news is that these chatbots are speeding up legal work, increasing productivity and lowering costs. The bad news is that they are getting lawyers into big trouble, especially in America, Canada, Australia and the United Kingdom. But as their use proliferates it is only a matter of time before their downsides are also felt in Hong Kong.
Because AI, in its eagerness to please its users, has a nasty habit of creating fake laws, fake legal arguments and even fake cases. This invention of imaginary facts is known as "hallucinating."
In a court in California, two law firms had to pay US$31,100 (HK$242,580) in fees to defense counsel and for court costs after they submitted documents containing “bogus AI-generated research.” Another American defense counsel has been fined US$10,000 for filing documents containing 21 fake legal quotations. So far there have been nearly 800 instances of AI legal fakery in the US, including three judges citing fake legal authorities in their judgments.
The problem, it seems, is that AI chatbots are designed to provide an answer. Ask it a question and it will do its best to provide an answer. But the more difficult the legal argument, the more the chatbot will be liable to hallucinate because it will always try to please the user.
As the use of AI spreads in the legal world I think it is inevitable that such issues will appear in Hong Kong. AI can be a great power for good but we must be ready and on guard to deal with its downsides. It is not only fake ‘facts’ that are the problem. There are also worries that AI will make it as simple as pie for anyone to submit legal claims. Ask AI to create the legal documents to support a claim for damage to your property and it will do so in seconds; all without the need of a lawyer.
Up until now the cost of lawyers’ fees has acted as a brake or barrier to the provision of justice but chatbots like ChatGPT can enable claimants to dispense with lawyers.
England’s second most senior judge, Sir Geoffrey Vos, has warned that the courts could be overwhelmed by AI-generated claims because litigants are increasingly using free-of-charge AI rather than a costly lawyer. “AI”, he says, “has the ability to transform a mass of documents and personal information into an arguable legal claim.” He describes it as an “AI revolution.”
We cannot forecast every way in which AI will affect the law, but at this preliminary stage three crucial tests stand out – firstly, how to identify and remove fake hallucinations; secondly, how to use AI’s strengths to deliver fast, cost-effective justice for both civil and family disputes as well as for criminal offences; thirdly, how to use AI to decide cases, both civil and criminal, so that AI-assisted judgments are more reliable, much cheaper and much faster than the human-only equivalent.
Eventually, this will get to a point where many litigants might actually prefer their cases to be settled, quickly and cheaply, by machines rather than slowly and expensively by humans. But I doubt that AI will replace human judges in the foreseeable future unless, of course, all the parties agree in advance to accept the judgement of a machine. It sounds crazy, but AI is about to upend many aspects of our lives.
Cheng Huan is an author and a senior counsel who practices in Hong Kong
