February 3, 2023, 12:04 pm | Read time: 3 minutes
Why make things difficult when, thanks to artificial intelligence, they can be easy? This could be the interpretation of a Colombian judge who had a verdict written by the text generator ChatGPT. He plans to continue this approach in the future. Legally, this is likely permissible, but there are also arguments against it. Read all about it at TECHBOOK.
The case in the port city of Cartagena involved an autistic child and the question of whether the child’s insurance must cover the costs of medical treatment. The judge posed the question to ChatGPT, OpenAI’s chatbot, and received a well-crafted response based on the relevant legal text. According to this, minors diagnosed with autism in Colombia are exempt from paying therapy fees. The case was thus decided. However, the judge has opened a different can of worms: Is it really okay for judges to have their verdicts written by ChatGPT?
Overview
Judge plans to continue using ChatGPT for verdicts
The judge was transparent in his approach, publishing the “conversation” with the chatbot. In an interview with the radio station “Bluradio,” he finds arguments for using ChatGPT in his profession, especially since it is only logical. Normally, his secretary would have to look up legal texts, he explains. With ChatGPT, it is faster, and there is the added benefit of a ready-made verdict formulated in clear and understandable language. He is convinced: Colleagues will follow his lead.
Also interesting: New religion wants to worship godlike artificial intelligence
Law not only allows but encourages the use of AI
Legally, the man seems to be on safe ground. According to a Colombian law from 2022, it is permissible to use artificial intelligence for judicial decision-making. It is even encouraged to make work more efficient. The advantages are seemingly obvious. ChatGPT can call up the exact wording of laws and precedents in seconds, which is supposed to ensure a fair verdict. However, the whole thing is not without its pitfalls.
Companies Are Liable for Their AI “Children”
When Using AI Chatbots Can Become Dangerous
Do not overestimate ChatGPT’s capabilities
For ChatGPT to have so many answers ready, OpenAI fed it with countless pieces of information through a specially developed language model. However, this “training” was completed at the end of 2021–no new content has been added since 2022. Therefore, it is quite possible that the text generator relies on outdated knowledge.
False information in 80 percent of cases
According to the service “NewsGuard,” ChatGPT is not only less reliable than often advertised but is also a “superspreader of misinformation.” After a random sample review of responses by the tool, analysts report an error rate of a staggering 80 percent. However, the presentation is always eloquent and thus convincing.
ChatGPT itself views AI use for judicial rulings critically
Even the bot itself would only recommend its use for judicial verdicts by judges in a limited capacity, TECHBOOK asked it. While there is “no direct regulation prohibiting judges from using an AI model like ChatGPT to assist in their decisions, it may be important that such a procedure meets the requirements of impartiality and independence in the judiciary and that decisions are made on an appropriate and lawful basis.” Therefore, it would be advisable, according to the tool, to conduct a legal review before a verdict generated by AI is delivered.