ChatGPT's artificial intelligence service has been the subject of much debate since its launch late last year. This new generative AI tool is capable of producing different kinds of text for many everyday uses, from information searches to book summaries and detailed cooking recipes. Some internet users even use it to write cover letters or best man speeches for weddings! Interest in ChatGPT increased last month with the release of an even more sophisticated and powerful version of GPT-4. However, every innovation has an ecological cost, as has been seen with other recent technologies such as cryptocurrencies or NFTs. And ChatGPT is definitely no exception. French app and platform Greenly, which allows companies to evaluate their CO2 emissions in real time, looked at the potential carbon footprint of ChatGPT-3. Greenly estimates that the GPT-3 version of the AI service could have a footprint of around 240 tonnes of CO2e, the equivalent of 136 round-trip flights between Paris and New York. The main source of ChatGPT-3 energy consumption lies in the data centers deployed to operate the service. Education systems alone account for 99 % of total emissions or 238 t of CO2 equivalent per year. The report states that operational electricity accounts for three-quarters of this footprint (ie 160 t CO2e), followed by server manufacturing (68.9 t CO2e) and refrigerant gas leakage (9.6 t CO2e). The remaining two tons of the carbon footprint estimated by Greenly are related to the design of the tool and its use with processes such as data storage, data transfer and the search process. This calculation is based on the scenario that a company uses ChatGPT to automatically send one million emails per month for one year. This puts into perspective the general enthusiasm for a tool that is considered revolutionary in many ways. Despite ChatGPT's convincing performance, says Greenly climate expert Tommy Catherine, it's worth asking whether it's worth the environmental cost, as carbon emissions increase significantly with every technological advance. (ETX)