More on AI Hallucinations…

The mainstream adoption of AI continues to reveal the precarious balance between the benefits and the pitfalls.

Yes, AI tools can reduce the time it takes to research information, to draft documents and to complete repetitive tasks.

But, AI is not so good at navigating subtle nuances, interpreting specific context or understanding satire or irony. In short, AI cannot “read the room” based on a few prompts and a collection of databases.

And then there is the issue of copyright licensing and other IP rights associated with the original content that large language models are trained on.

One of the biggest challenges to AI’s credibility is the frequent generation of “hallucinations” – false or misleading results that can populate even the most benign of search queries. I have commented previously on whether these errors are deliberate mistakes, an attempt at risk limitation (disclaimers), or a way of training AI tools on human users. (“Spot the deliberate mistake!) Or a get-out clause if we are stupid enough to rely on a dodgy AI summary!

With the proliferation of AI-generated results (“overviews”) in basic search queries, there is a tendency for AI tools to conflate or synthesize multiple sources and perspectives into a single “true” definition – often without authority or verified citations.

A recent example was a senior criminal barrister in Australia who submitted fake case citations and imaginary speeches in support of a client’s case.

Leaving aside the blatant dereliction of professional standards and the lapse in duty of care towards a client, this example of AI hallucinations within the context of legal proceedings is remarkable on a number of levels.

First, legal documents (statutes, law reports, secondary legislation, precedents, pleadings, contracts, witness statements, court transcripts, etc.) are highly structured and very specific as to their formal citations. (Having obtained an LLB degree, served as a paralegal for 5 years, and worked in legal publishing for more than 10 years, I am very aware of the risks of an incorrect citation or use of an inappropriate decision in support of a legal argument!!!)

Second, the legal profession has traditionally been at the forefront in the adoption and implementation of new technology. Whether this is the early use of on-line searches for case reports, database creation for managing document precedents, the use of practice and case management software, and the development of decision-trees to evaluate the potential success of client pleadings, lawyers have been at the vanguard of these innovations.

Third, a simple document review process (akin to a spell-check) should have exposed the erroneous case citations. The failure to do so reveals a level laziness or disregard that in another profession (e.g., medical, electrical, engineering) could give rise to a claim for negligence. (There are several established resources in this field, so this apparent omission or oversight is frankly embarrassing: https://libraryguides.griffith.edu.au/Law/case-citators, https://guides.sl.nsw.gov.au/case_law/case-citators, https://deakin.libguides.com/case-law/case-citators)

In short, as we continue to rely on AI tools, unless we apply due diligence to these applications or remain vigilant to their fallibility, we use them at our peril.

 

AI hallucinations and the law

Several years ago, I blogged about the role of technology within the legal profession. One development I noted was the nascent use of AI to help test the merits of a case before it goes to trial, and to assess the likelihood of winning. Not only might this prevent potentially frivolous matters coming to trial, it would also reduce court time and legal costs.

More recently, there has been some caution (if not out and out scepticism) about the efficacy of using AI in support of legal research and case preparation. This current debate has been triggered by an academic paper from Stanford University that compared leading legal research tools (that claim to have been “enhanced” by AI) and ChatGPT. The results were sobering, with a staggering number of apparent “hallucinations” being generated, even by the specialist legal research tools. AI hallucinations are not unique to legal research tools; nor to the AI tools and the Large Language Model (LLMs) they are trained on, as Stanford has previously reported. While the academic paper is awaiting formal publication, there has been some to-and-fro between the research authors and at least one of the named legal tools. This latter rebuttal rightly points out that any AI tool (especially a legal research and professional practice platform) has to be fit for purpose, and trained on appropriate data.

Aside from the Stanford research, some lawyers have been found to have relied upon AI tools such as ChatGPT and Google Bard to draft their submissions, only to discover that the results have cited non-existent precedents and cases – including in at least one high-profile prosecution. The latest research suggests that not only do AI tools “imagine” fictitious case reports, they can also fail to spot “bad” law (e.g., cases that have been overturned, or laws that have been repealed), offer inappropriate advice, or provide inaccurate or incorrect legal interpretation.

What if AI hallucinations resulted in the generation of invidious content about a living person – which in many circumstances, would be deemed libel or slander? If a series of AI prompts give rise to libelous content, who would be held responsible? Can AI itself be sued for libel? (Of course, under common law, it is impossible to libel the dead, as only a living person can sue for libel.)

I found an interesting discussion of this topic here, which concludes that while AI tools such as ChatGPT may appear to have some degree of autonomy (depending on their programming and training), they certainly don’t have true agency and their output in itself cannot be regarded in the same way as other forms of speech or text when it comes to legal liabilities or protections. The article identified three groups of actors who might be deemed responsible for AI results: AI software developers (companies like OpenAI), content hosts (such as search engines), and publishers (authors, journalists, news networks). It concluded that of the three, publishers, authors and journalists face the most responsibility and accountability for their content, even if they claimed “AI said this was true”.

Interestingly, the above discussion referenced news from early 2023, that a mayor in Australia was planning to sue OpenAI (the owners of ChatGPT) for defamation unless they corrected the record about false claims made about him. Thankfully, OpenAI appear to have heeded of the letter of concern, and the mayor has since dropped his case (or, the false claim was simply over-written by a subsequent version of ChatGPT). However, the original Reuters link, above, which I sourced for this blog makes no mention of the subsequent discontinuation, either as a footnote or update – which just goes to show how complex it is to correct the record, since the reference to his initial claim is still valid (it happened), even though it did not proceed (he chose not to pursue it). Even actual criminal convictions can be deemed “spent” after a given period of time, such that they no longer appear on an individual’s criminal record. Whereas, someone found not guilty of a crime (or in the mayor’s case, falsely labelled with a conviction) cannot guarantee that references to the alleged events will be expunged from the internet, even with the evolution of the “right to be forgotten“.

Perhaps we’ll need to train AI tools to retrospectively correct or delete any false information about us; although conversely, AI is accelerating the proliferation of fake content – benign, humourous or malicious – thus setting the scene for the next blog in this series.

Next week: AI and Deep (and not so deep…) Fakes

 

 

 

 

Smart Contracts… or Dumb Software

The role of smart contracts in blockchain technology is creating an emerging area of jurisprudence which largely overlaps with computer programming. However, one of the first comments I heard about smart contracts when I started working in the blockchain and crypto industry was that they are “neither smart, nor legal”. What does this paradox mean in practice?

First, smart contracts are not “smart”, because they still largely rely on human coders. While self-replicating and self-executing software programs exist, a smart contact contains human-defined parameters or conditions that will trigger the performance of the contract terms once those conditions have been met. The simplest example might be coded as a type of  “if this, then that” function. For example, I could create a smart contract so that every time the temperature drops below 15 degrees, the heating comes on in my house, provided that there is sufficient credit in the digital wallet connected to my utilities billing account.

Second, smart contracts are not “legal”, unless they comprise the necessary elements that form a legally binding agreement: intent, offer, acceptance, consideration, capacity, certainty and legality. They must be capable of being enforceable in the event that one party defaults, but they must not be contrary to public policy, and parties must not have been placed under any form of duress to enter into a contract. Furthermore, there must be an agreed governing law, especially if the parties are in different jurisdictions, and the parties must agree to be subject to a legal venue capable of enforcing or adjudicating the contract in the event of a breach or dispute.

Some legal contacts still need to be in a prescribed form, or in hard copy with a wet signature. A few may need to be under seal or attract stamp duty. Most consumer contracts (and many commercial contracts) are governed by rules relating to unfair contract terms and unconscionable conduct. But assuming a smart contract is capable of being created, notarised and executed entirely on the blockchain, what other legal principles may need to be considered when it comes to capacity and enforcement?

We are all familiar with the process of clicking “Agree” buttons every time we sign up for a social media account, download software or subscribe to digital content. Let’s assume that even with a “free” social media account, there is consideration (i.e., there’s something in it for the consumer in return for providing some personal details), and both parties have the capacity (e.g., they are old enough) and the intent to enter into a contract, the agreement is usually no more than a non-transferable and non-exclusive license granted to the consumer. The license may be revoked at any time, and may even attract penalties in the event of a breach by the end user. There is rarely a transfer of title or ownership to the consumer (if anything, social media platforms effectively acquire the rights to the users’ content), and there is nothing to say that the license will continue into perpetuity. But think how many of these on-line agreements we enter into each day, every time we log into a service or run a piece of software. Soon, those “Agree” buttons could represent individual smart contracts.

When we interact with on-line content, we are generally dealing with a recognised brand or service provider, which represents a known legal entity (a company or corporation). In turn, that entity is capable of entering into a contract, and is also capable of suing/being sued. Legal entities still need to be directed by natural persons (humans) in the form of owners, directors, officers, employees, authorised agents and appointed representatives, who act and perform tasks on behalf of the entity. Where a service provider comprises a highly centralised entity, identifying the responsible party is relatively easy, even if it may require a detailed company search in the case of complex ownership structures and subsidiaries. So what would be the outcome if you entered into a contract with what you thought was an actual person or real company, but it turned out to be an autonmous bot or an instance of disembodied AI – who or what is the counter-party to be held liable in the event something goes awry?

Until DAOs (Decentralised Autonomous Organisations) are given formal legal recognition (including the ability to be sued), it is a grey area as to who may or may not be responsible for the actions of a DAO-based project, and which may be the counter-party to a smart contract. More importantly, who will be responsible for the consequences of the DAO’s actions, once the project is in the community and functioning according to its decentralised rules of self-governance? Some jurisdictions are already drafting laws that will recognise certain DAOs as formal legal entities, which could take the form of a limited liability partnership model or perhaps a particular type of special purpose vehicle. Establishing authority, responsibility and liability will focus on the DAO governance structure: who controls the consensus mechanism, and how do they exercise that control? Is voting to amend the DAO constitution based on proof of stake?

Despite these emerging uncertainties, and the limitations inherent in smart contracts, it’s clear that these programs, where code is increasingly the law, will govern more and more areas of our lives. I see huge potential for smart contracts to be deployed in long-dated agreements such as life insurance policies, home mortgages, pension plans, trusts, wills and estates. These types of legal documents should be capable of evolving dynamically (and programmatically) as our personal circumstances, financial needs and living arrangements also change over time. Hopefully, these smart contracts will also bring greater certainty, clarity and efficiency in the drafting, performance, execution and modification of their terms and conditions.

Next week: Free speech up for sale

 

Melbourne Legal Hackers Meetup

Given my past legal training and experience, and my ongoing engagement with technology such as Blockchain, I try to keep up with what is going on in the legal profession, and its use and adoption of tech. But is it LawTech, LegalTech, or LegTech? Whatever, the recent Legal Hackers Meetup in Melbourne offered some definitions, as well as a few insights on current developments and trends.

The first speaker, Eric Chin from Alpha Creates, defined it as “tech arbitrage in the delivery of legal services”. He referred to Stanford Law School’s CodeX Techindex which has identified nine categories of legal technology services, and is maintaining a directory of companies active in each of those sectors.

According to Eric, recent research suggests that on average law firms have a low spend on legal technology and workflow tools. But typically, 9% of corporate legal services budgets are being allocated to “New Law” service providers. Separately, there are a growing number of LegalTech hubs and accelerators.

Meanwhile, the Big Four accounting firms are hiring more lawyers, and building our their legal operations, and investing in legal tech and New Law (which is defined as “using labour arbitrage in the delivery of legal services”).

Key areas of focus for most firms are Practice Management, Legal Document Automation,
Legal Operations and e-Discovery.

Joel Seignior, Legal Counsel on the West Gate Tunnel Project, made passing mention of Robert J Gordon’s economic thesis in “The Rise and Fall of American Growth”, which at its heart postulates that despite all appearances to the contrary, the many recent innovations we have seen in IT have not actually delivered on their promises. He also referred to
Michael Mullany’s 8 Lessons from 16 Years of the Gartner Hype Cycle, which the author considers to be past its use-by date. Which, when taken together, suggest that the promise of LegalTech is somewhat over-rated.

Nevertheless, businesses such as LawGeex are working in the legal AI landscape and other disciplines to deliver efficiency gains and value-added solutions for matter management, e-billing, and contract automation. Overall, UX/UI has finally caught up with technology like document automation and expert systems.

Finally, Caitlin Garner, Head of Innovation at Allens spoke about her firm’s experience in developing a Litigation Innovation Program, underpinned by a philosophy of “client first, not tech first”. One outcome is REDDA, a real estate due diligence app, that combines contract analytics, knowledge automation, reporting and collaboration. Using off-the shelf solutions such as Kira’s Machine Learning, Neota’s Expert System and HighQ, the Allens team have developed a transferable template model. Using a “Return & Earn” case study, the firm has enabled the on-boarding of multiple suppliers into a streamlined contract management, signature and execution solution.

Next week: Notes from New York Blockchain Week