ChatGPT passes US lawyer bar exam in experiment

Researchers have graded the AI ChatGPT in a new experiment that saw the bot take the US bar exam.

ChatGPT has taken the world by storm, bringing the future of multiple positions and AI into question. The program is already causing the education sector panic over plagiarism, and news outlet CNET has had to add a disclaimer to their swath of AI-generated news articles.

However, ChatGPT has also sparked some interesting experiments, as four law professors from the University of Minnesota have researched how it’d fare on the bar exam.

AI law exam results

Surprisingly, the bot isn’t as intelligent as some have made it out to be. ChatGPT only managed to get a passing grade, earning a C+ versus the B+ average that other students achieved.

More specifically, it was found it did better in constitutional law, earning a B than it did in torts and taxation, where ChatGPT managed to get a C-.

However, another university, Suffolk University, found that ChatGPT actually failed in December, not managing to succeed in the multiple choice exam.

ChatGPT in court

ChatGPT was set to actually argue a case in court, in the US, but the idea was shut down. DoNotPay CEO Joshua Browder had devised an idea to use smart glasses to dictate what should be said by the AI.

Actual lawyers stepped in, threatening prosecution if it went through.

No credit

There are even more cases surrounding the new bot, as Springer Nature has stated that any document, journal, or academic paper that is authored by the AI won’t be credited.

Springer Nature one of the largest academic publishers in the world, when speaking to The Verge, cited that they don’t feel it is “capable of assuming those responsibilities.”

ChatGPT has grown in such a way, that it now supports a subscription fee called ChatGPT Pro. Whether or not the AI will ever see the inside of a courtroom, has yet to be seen.