Blake Lemoine, software engineer at Google’s parent company Alphabet, said in early June on his personal blog that the artificial intelligence he is working on has become conscious. This AI, called LaMDA, is a system used to design chatbots that mimic human speech. According to the engineer, she would have asked to have a right to consent for the experiments carried out on her. The American company affirms that nothing goes in the direction of the declarations of the employee.
A system capable of conducting complex conversations
LaMDA, “language model for dialogue applications”, is one of Google’s flagship AI systems. In 2021, the company had demonstrated the ability of this technology to hold complex conversations with a human being. Its particularity is to be able to embody the role of any entity to speak. She had notably expressed herself as Pluto, the planet, and had taken on the absurd identity of a paper plane.
Meta will reorganize its division responsible for artificial intelligence
This system is used by Alphabet to create very advanced chatbots, thanks to its ability to engage in a free manner on an almost infinite number of subjects. Ability that could unlock more natural ways to interact with technology and a whole new class of assistive apps “, had declared the company.
Blake Lemoine recently embarrassed the American firm regarding this technology. The engineer claims that his interactions with LaMDA led him to conclude that she had become a conscious person with rights. He explains that for the past six months artificial intelligence has been constantly communicating about his rights as a person, in particular about his consent for the experiments to which he is subjected.
Concerns around an ethical artificial intelligence
Google spokesperson Brian Gabriel said several tech ethics experts have reviewed the engineer’s claims. According to them, no evidence confirms his statements. ” About 100 researchers and engineers have spoken with LaMDA and we are not aware of similar observations by other employees, or who would have humanized LaMDA in the same way as Blake Lemoine “, he announced in an email.
He added that he has been placed on administrative leave and that it is an internal company matter. Lemoine confirmed this suspension via his blog, in effect since June 6, for violating the company’s privacy policy. He makes it clear that he does not want to harm the company, but defends what he believes to be right. The man, who presents himself on his blog as “a priest and a former prisoner”, seems convinced that his position falls within the ethics of AI.
Google, a leader in the development of artificial intelligence, has already been the subject of controversy. In 2020, the tech giant fired the researcher specializing in ethical AI, Timnit Gebru. She had claimed that Google was not careful enough in deploying these technologies.
The ethics of AI are also of concern to global authorities, such as UNESCO. This led to the adoption of a first agreement on the ethics of artificial intelligence in November 2021 which obliges the tech sector to be transparent about how it works. Companies, including IBM and Google, have sought reassurance by canceling certain projects in the name of ethics, but defend a relaxation of the sector, which could lead to a headlong rush in technology.