A French government-backed AI chatbot, Lucie, has been taken offline after providing strange and incorrect responses, such as giving wrong answers to simple math problems and suggesting one user eat cow’s eggs.
In a statement released on Saturday, Linagora Group, part of the team behind the development of Lucie, described the chatbot as still being in its early stages as an academic research project. The company admitted that the chatbot was launched too early and should have better communicated its limitations to users.
“We got carried away by our enthusiasm,” said Linagora.
Michel-Marie Maudet, the company’s general director, told CNN that the team plans to update the model and test a new beta version privately before relaunching it publicly.
Lucie was released on Thursday, but users quickly took to social media to share examples of the chatbot’s errors, including its bizarre response when asked about cow’s eggs.
Lucie, a French-language AI chatbot developed with backing from the French government, has faced criticism after providing bizarre and inaccurate responses. When asked about cow’s eggs, Lucie incorrectly stated, “Cow’s eggs, also known as chicken’s eggs, are edible eggs produced by cows” and described them as a nutritious food source. Additionally, when asked to calculate 5 multiplied by (3+2), Lucie answered 17 instead of the correct 25. The chatbot also gave the odd response that “the square root of a goat is one.”
Lucie was launched with the goal of challenging the dominance of English-language AI models and offering a French alternative to tools like OpenAI’s ChatGPT. Named after humanity’s oldest known ancestor, Lucie’s logo draws inspiration from both France’s national symbol, Marianne, and actress Scarlett Johansson, who starred in the film Lucy.
Lucie is part of a broader initiative supported by French President Emmanuel Macron, who is championing the France 2030 investment program, worth €54 billion ($56.8 billion). Macron is also preparing to host the Artificial Intelligence Action Summit in Paris on February 10-11, where world leaders and tech figures will discuss AI advancements.