An AI language model will be used in a legal hearing scheduled for the upcoming defense of someone in a real case. According to DoNotPay’s Startup CEO, this experiment aims to show that AI can replace lawyers.
ChatGPT training data often reveals that their outputs can be biased or discriminatory. They can also generate false “facts,” making it difficult to use them in high-risk areas like medicine or law.
Joshua Browder, DoNotPay’s CEO, believes that the technology has the potential to be a success. According to him, he has persuaded someone with an AI chatbot to represent themselves. According to reports, the defendant will wear earbuds that stream audio snippets of legal arguments generated by a language model. They will then repeat them verbatim at a legal hearing over zoom.
According to reporters, the experiment will involve GPTJ, an open-source language model released by Eleuther (a collective of AI researchers and developers).
Browder refused to disclose other details regarding the case involving the robot lawyer, fearing that judges might stop him from executing his experiment.
He even offered $1 million to any lawyer using AI in a Supreme Court case. This is quite a challenge considering that most electronic devices are prohibited.