AI engineer fired by Google in controversy | computer science direction

Blake Lemoine, a senior software engineer at Google’s Responsible AI organization, is in the midst of a wave of controversy in de-administration. He expressed his thoughts that the company’s LaMDA AI was, in his own words, “a person” with whom he had a conversation in which she allegedly explained why she should be considered “sentimental.”

Image: Getty

Google does not share his opinion. But according to the Washington Post, Lemoine wasn’t fired for his opinion, but for violating the company’s privacy policy by working with people outside the company to validate his theory. The day before he was fired, Lemoine had turned over documents to a US Senator’s office that he claimed showed the company’s technologies discriminated against religion.

Lemoine explained on his personal blog that he voluntarily provided a list of names of people with whom he spoke on the subject, some of whom worked for the US government and had expressed an interest in federal oversight of the project. Google insisted that such monitoring was unnecessary.

In addition, Lemoine had invited an attorney to represent LaMDA and had spoken to a representative on the House Judiciary Committee about alleged unethical activities at Google, the Post reported.

LaMDA, an acronym for Language Model for Dialogue Applications, is controversial. Introduced in 2021, LaMDA is a natural language processing (NLP) system with artificial intelligence (AI) that mimics human conversations due to its advanced ability to understand innuendos. Unlike chatbots, which are commonly used in the customer service industry today, LaMDA can have fluent conversations on a variety of topics.

During the announcement, Google said LaMDA adheres to its AI principles and its development team is working to mitigate the risk. Training NLP systems can be difficult due to discrimination and hate speech embedded in training data that is difficult to filter.

In a statement to The Washington Post, Google said that while its systems mimic conversations, they have no conscience. Google also said that LaMDA has been vetted by its team and adheres to the company’s AI principles.

For more details, the original article (in English) is available on the website IT world Canadaa sister publication of computer science direction.

Also read:

Google I/O 2022: New features in Google Workspace

New artificial intelligence lab in Montreal

Quebec is launching a call for projects in the field of artificial intelligence

French translation and adaptation by Renaud Larue-Langlois




Tags: Blake Lemoine, Google AI, artificial intelligence, LaMDA

Leave a Comment