Thanks for checking out part 3 of our series about AI, and its impact on the legal profession. In our previous two entries, we largely focussed on Chat GPT, and other forms of generative AI known as “Large Language Models”. When most people think about AI in 2023, that’s what immediately comes to mind. And to this point in time, when it comes to considerations about how AI will affect both legal practitioners and their clients, most people aren’t even aware of other types of generative AI that could have an even greater effect on the legal process.

If you’re coming to this article from our email newsletter, you would have seen the header photo of what appears to be Pope Francis of the Catholic Church, dressed in some very trendy clothes. To be honest, he actually looks pretty stylish with that sort of fashion approach. But despite how lifelike that image may appear to be, it isn’t real. And it isn’t even the work of some ultra-talented graphic designer or artist either. That image was generated by an AI software known as MidJourney. That’s right. No human involvement occurred in the creation of that true to life image, OTHER than the prompt given to the AI asking it what the person wanted. And currently products like MidJourney, and the similarly functioning Stable Diffusion, are finding their way into the hands of end-users. Any regular person could conceivably get their hands on this unbelievably powerful tool.

And then there are the latest voice synthesis AIs. As of the latter half of 2023, everyday consumers are also able to get access to AI based voice generating software. Software that, with a small sampling of the voice of a real person, is able to generate spoken audio using that voice that can be almost impossible to tell from actual voice recordings. If you want to get a good idea of the implications of this, we actually suggest watching a series that got popular on youtube over the course of 2023, called Biden and The Gang. It is a totally goofy, silly, sitcom- esque show starring Joe Biden, Barack Obama, George W. Bush, and Donald Trump as four pals who get up to Scooby-Doo style hijinks. And it was all made by one young man who appears to be a normal 20 year old with no other extraordinary talents than having access to voice synthesis AI and a knack for comedy writing.

In all seriousness, you can probably see where we’re going with this though. Historically, photographic evidence, and voice recording have both been used extensively as ways to establish a pattern of evidence in both civil and criminal proceedings. Whether it is a spouse getting photos or their partner cheating in a divorce court, or recordings of malfeasance from a superior in a wrongful termination suit, our legal system has a heavy dependence on materials which may now no longer be considered to hold anywhere near the same evidentiary value they once did.

It does not seem as though there are stories in the general zeitgeist yet about AI forged evidence (though as soon as one does occur, we will be covering it), but it is only a matter of time before it is attempted by people in cases both big and small. If our legal system has not yet adapted ways of telling the difference between what was AI generated, and what is real, the shock to the legal system could be severe. The fact that these tools are available to everyone means that any bad faith actor in any given case could conceivably choose to give this a try. And unlike the misadventures of the attorneys who used ChatGPT in Mata V. Avianca, and ended up citing non-existent cases, the possibility of AI forged evidence passing as real is significantly higher.

In our opinion, we should probably all be looking for ways to forestall this issue before it becomes a reality. Though there has to be some reasonable fear that despite anything we may try to do, AI will simply break the standards of what is considered viable evidence in legal practice going forward. In which case, it’s up to all of us to start figuring out what a well constructed case looks like when that day comes to pass.