The Corsican MP Jean-Félix Acquaviva had fun drafting an amendment to the bill on the Olympic and Paralympic Games by ChatGPT, just to launch the debate at the Bourbon Palace.
Is artificial intelligence preparing to revolutionize the National Assembly and its activities? Let’s not go that far, but Jean-Félix Acquaviva’s initiative deserves a look. The deputy Libertés, Indépendants, Outre-mer et Territoires of the second constituency of Haute-Corse had an amendment to the bill on the 2024 Olympic Games drafted by the conversational robot ChatGPT. The text is already being debated in that it would authorize, if adopted in this form, the use of automated (biometric) video surveillance on an experimental basis.
When AI proposes to protect the rights and freedoms of citizens
The amendment, co-signed by around twenty deputies from the LIOT group, was drafted by the artificial intelligence of OpenAI, which intervenes in a difficult context, where many elected officials (even MEPs for that matter) are worried about this possible use of technology for surveillance purposes, in particular through behavioral analysis, and therefore biometrics.
The idea, by entrusting ChatGPT with the task of writing the amendment, is a bit to raise the alarm, by pointing out the excessive use of certain digital tools. Our colleagues from The letter a report that the conversational robot has made an important clarification on the analysis of the data to be derived from the images captured by the CCTV cameras.
Around this provision (the much criticized article 7), ChatGPT explains that this analysis will have to be done while respecting ” the fundamental rights and freedoms of data subjects and the obligation to ensure humane decision-making at each stage of the processing “.
ChatGPT, a luxury parliamentary assistant?
On Tech&Co, MP Acquaviva claimed to have used version 3.5 of ChatGPT (the model having since been replaced by a GPT-4 teeming with new features). He also admitted to having slightly retouched the wording delivered by the AI, just to make it correspond to the French legislative framework. The elected representative recalls, in the presentation of his amendment n° 101, that he is ” necessary to warn about the risks that this type of new technology can entail “.
Much of the amendment has not been changed one iota. And you might be surprised when you read it. To obtain his result, Jean-Félix Acquaviva copied on ChatGPT the part of the bill which related to his amendment, adding to it the request “Write a parliamentary amendment on the following article”.
The member of the LIOT group adds, on the site of the Assembly, that ” this explanatory memorandum might seem reassuring, however artificial intelligence can allow the best as well as the worst “, adding that questions relating to public freedoms, fundamental rights, justice or security ” should never be left under the control of artificial intelligence alone “. The debate is open.
Sources: National Assembly, The letter a, BFM TV
5