Guest post by Miklos Orban, chief technology officer at gunnercooke
Orban: ChatGPT can form arguments like lawyers
Throughout my 24-year legal career, I have always had a ‘love-hate’ or ‘cat and mouse’ relationship with law.
Why is it so easy to complain about being a lawyer? The endless administration, the never-ending race to keep up with ever-changing laws, clients wanting everything by yesterday, the frustrating uncertainty, or knowing you’re right but the outcome going against you, to name a few reasons.
I am sure these sound familiar. Whether you practise in India or Poland, all lawyers experience the same pain. Despite this, being a lawyer can feel so good: you have the chance to build great relationships with your clients, it is intellectually stimulating, you provide tangible solutions that actually help people and, last but not least, it pays better than most other professions.
Well, I come bearing bad news. This was the golden age of the legal profession. Despite having your issues with being a lawyer, most of these were overcome by the lavish rewards that come with it.
However, a new challenge is emerging, far greater than competition from a new boutique firm, a younger generation of lawyers, or the Big 4.
That challenge? Artificial intelligence (AI). Now, before you skip to the end of this piece to see whether I reveal that this was written by ChatGPT, relax: I am writing this piece, as a flesh and blood lawyer, who has been involved in the legal tech and AI sphere for many years.
I am certain that you are aware of the latest version of ChatGPT, an AI chatbot developed by OpenAI. To say that it is ground-breaking is an understatement.
To give you an example of its capabilities, I asked ChatGPT to write in the style of a 19th century poem, on whether AI will pose a threat to employment for college-educated workers, particularly lawyers, in the next decade. This is what it produced, in a fraction of a second:
In this age of machines so grand and neat,
Where gears and wires rule with iron seat,
We wonder what fate for us doth wait,
Will we be replaced by gears and plate?
The lawyers of old with quill and book,
Now face a future that doth look
To automatons with nimble mind,
To take on tasks of law we find.
Will this mean our end, a bleak demise,
Or will it bring new opportunities arise?
As machines take on the tedious chore,
Perhaps new doors for us will open more.
But whether fate be kind or cruel,
We must adapt, or be a tool
For progress, and the march of time,
And find our place in this new clime.
Not bad. Not bad at all, right? Try it for yourself: ask ChatGPT about anything from high-level theoretical questions to more specific ones. Give it a very creative task like writing a movie script about a given subject, or prompt it with a legal task you’d give to your trainee. I promise, you’ll be surprised.
ChatGPT can make mistakes, however. Funnily enough, its mistakes are like those of humans and they are most prevalent in logical tasks.
For instance, a few weeks ago, someone gave ChatGPT the age-old riddle: “My brother was half my age when I was 10. How old is he when I am 70?” The correct answer is obviously 65, but many people (wrongly) jump to saying 35. ChatGPT was no different. Why? Because it was trained using human conversations and, therefore, it can make very human mistakes.
However, it also makes some unhuman-like, robotic mistakes, especially when it misunderstands the prompt. This is probably the weakest point of AI chatbots: translating human requests into an item in a large computer database is a very complicated task.
A final note on its capabilities: it knows way more than any human. It easily passed the US medical exam, several MBA exams, and the practice test for the bar, but its knowledge on some subjects is somewhat limited and often outdated.
The reason behind the hype surrounding ChatGPT is that this may be the first time where people feel that machines may not only replace photocopiers, bank tellers and blue-collar workers, but threaten the professions, such as computer coders, financial analysts, journalists, and us – lawyers.
Throughout history, ground-breaking technological developments – from the steam engine to the personal computer – have only threatened jobs that required no college education. This is no longer the case thanks to AI, or at least a few AI solutions that have come to light in recent months.
We, as lawyers, have never felt threatened by the emergence of industrial robots or the internet. This only boosted our business. With an AI tool which can generate decent answers to complicated questions in a matter of seconds, however, the fear becomes all too real.
Do not get too hung up on ChatGPT’s weaknesses. Yes, you can easily catch it making stupid mistakes or giving completely incorrect answers to your questions, especially if it is related to law.
Let me remind you, however, that ChatGPT was not built to be the next encyclopaedia, nor a legal chatbot. Its strengths are not in law, but it possesses skills that are key in delivering legal services.
For instance, it has displayed good levels of emotional communication. It can sympathise with you and show empathy, or to be precise: it has learned what factors make a communication most probably seem empathetic.
Well, ChatGPT has no face, so it can’t gesticulate, but it doesn’t have bad days either. It can be kind and understanding even on days that you wouldn’t.
But its most striking feature is not emotional: ChatGPT’s algorithm has managed to find the key to human creativity. It may not (re)invent the wheel or a spaceship, but it can definitely write excellent poems or articles, great movie scripts or human-like book reviews in various styles, on almost any subject.
And all within a few seconds. It is so good at writing college papers that many universities have already been forced to reconsider the ways in which they conduct exams.
For me, the most fascinating feature is its ability to form arguments. It takes facts and turns them into arguments, draws conclusions and projects future events from trends. Exactly what good lawyers do.
Needless to say, building such a robot is much harder than teaching a computer to answer legal questions. If you already have an engine with this skillset, training it for a specific domain, such as law or tax, would not require a massive effort.
Hence, we lawyers can pray that ChatGPT and similarly advanced AI systems are not going to be trained for legal services anytime soon. We have a good chance of this: it took almost 100 years for electricity to have an economic impact, and even if the changes these days tend to take effect much faster, nothing will change overnight.
AI will not make us redundant quickly, but this does not mitigate the huge impact that AI will have on the legal profession.
Lawyers have been feeling the pressure to deliver faster, more cost-efficient, and commercially minded services for the last two decades; AI will take this pressure to a whole new level. The more legal work that can be taken care of with the help of AI systems, such as ChatGPT, the more pressure lawyers will face to be better than AI.
Will lawyers need to beat AI in legal knowledge, responsiveness, emotional intelligence or client care? It’s hard to say. One thing seems certain, though: lawyers need to be on the forefront of this change. A lawyer who understands AI and law can use it more effectively than her clients, and can integrate AI into her services to make them far superior to her competitors.
Hence, it is more appropriate to say that lawyers will be ‘affected’ or ‘challenged’, rather than ‘replaced’, by AI. It is more of a case of lawyers working with AI that will replace lawyers who don’t.
Leave a Comment