Brazilian City Council Unknowingly Enacts First AI-Written Ordinance

City lawmakers in Brazil have passed what seems to be the country's first law that was entirely authored by artificial intelligence, although they were unaware of this at the time of enactment.

The southern Brazilian city of Porto Alegre passed an experimental ordinance in October, and this week, city councilman Ramiro Rosário disclosed that the legislation was written by a chatbot.

This revelation has sparked controversy and prompted questions about the extent of artificial intelligence's involvement in public policy.

Rosário informed The Associated Press that he requested OpenAI's chatbot, ChatGPT, to formulate a proposal aimed at stopping the city from billing taxpayers for the replacement of water consumption meters in the event of theft. He subsequently introduced this proposal to his 35 council colleagues without making any modifications to it or disclosing its unprecedented origin.

“If I had revealed it before, the proposal certainly wouldn’t even have been taken to a vote,” Rosário was quoted as saying by AP.

He added that it would be unfair to the populace if the project were to face the risk of disapproval merely because it was penned by artificial intelligence.

The introduction of ChatGPT to the market just a year back has ignited a worldwide discussion about the effects of potentially groundbreaking AI-driven chatbots. Although some view it as a promising tool, it has also generated worries and unease about the unforeseen or unwanted consequences of a machine undertaking responsibilities currently carried out by humans.

Hamilton Sossmeier, the city council president of Porto Alegre, the second-largest city in southern Brazil with a population of 1.3 million, discovered that Councilman Rosário had employed ChatGPT to draft the proposal after Rosário boasted about this accomplishment on social media on Wednesday.

Sossmeier initially expressed to local press that he believed this to be a "dangerous precedent."

The large language models of AI, which power chatbots such as ChatGPT, function by persistently attempting to predict the subsequent word in a sentence. This mechanism often leads to the creation of false information, a situation commonly referred to as hallucination.

According to recent research published by the tech firm Vectara, all chatbots occasionally incorporate incorrect data while summarizing a document. This happens approximately 3 per cent of the time for the most sophisticated GPT model, while one of Google's models has a higher rate at around 27 per cent.

Earlier this year, Andrew Perlman, the dean at Suffolk University Law School, penned an article for Harvard Law School's Center of Legal Profession's website. In it, he suggested that ChatGPT might represent a shift more significant than even the introduction of the internet. However, he also cautioned about possible drawbacks of this technology.

“It may not always be able to account for the nuances and complexities of the law. Because ChatGPT is a machine learning system, it may not have the same level of understanding and judgment as a human lawyer when it comes to interpreting legal principles and precedent. This could lead to problems in situations where a more in-depth legal analysis is required,” Perlman wrote.

Rosário from Porto Alegre wasn't the first in evaluating the capabilities of ChatGPT among lawmakers globally. Some have previously experimented with it, albeit in a more restricted manner or with less fruitful results.

Democratic State Senator Barry Finegold from Massachusetts utilized ChatGPT to assist in drafting legislation intended to regulate AI models, ChatGPT included. Although the bill was introduced earlier this year, a vote on it is still pending.

According to AP, Finegold said that ChatGPT could assist with some of the laborious aspects of the legislative process, such as correct and swift searching and citing laws already on the books. However, he emphasised the importance of everyone being aware that a tool like ChatGPT was utilised in the process.

He expressed his desire for all work produced by ChatGPT to be appropriately watermarked.

He further emphasized that the application of artificial intelligence in the creation of new legislation is an inevitable progression.

“I’m in favor of people using ChatGPT to write bills as long as it’s clear," he added.

The proposal from Rosário in Porto Alegre lacked such transparency. Sossmeier indicated that Rosário failed to disclose to other council members that the proposal was penned by ChatGPT.

Rosário intentionally kept the source of the proposal a secret.

He explained to the AP that his aim extended beyond addressing a local problem to initiating a broader discussion.

He said he entered a 49-word prompt into ChatGPT and it returned the full draft proposal within seconds, including justifications.

“I am convinced that ... humanity will experience a new technological revolution,” he said.

“All the tools we have developed as a civilization can be used for evil and good. That’s why we have to show how it can be used for good," he added.

The president of the council, who initially criticized the approach, seems to have already been persuaded.

“I changed my mind. I started to read more in depth and saw that, unfortunately or fortunately, this is going to be a trend," Sossmeier said.