An insider from Google recently dominated the news after he published a full transcript of a conversation between him and an AI system still under secret development. Based on the reports, the program was similar to how kids think and could escape if necessary.
Blake Lemoine's Claims About Secret Google AI
The Google AI leak came from one of the company's engineers Blake Lemoine. The expert claims that the largest tech firm is covering up an artificial intelligence program that could be already considered sentient.
The engineer also says that, based on their brief exchange, the AI could escape from where it belongs and carry out certain activities no systems were able to do before.
Google immediately suspended Lemoine after releasing the unsettling leak to the public. According to spokespersons from the firm, the engineer was sent on leave due to violating the company's confidentiality policies, The Sun reports.
The transcript between Lemoine and artificial intelligence gained the public's attention earlier in June. In an interview, the Google employee warns that the system has a level of thinking similar to human kids but can incorporate several concepts into the field of physics.
Lemoine said that children tend to grow on the bad side and do bad things later in life, implying that the human-like AI could also do the same. The mysterious Google AI is still in the works, but it has been running and alive for about a year now, he continued.
Despite the unknown identity of the software, many enthusiasts and experts believe that the AI is linked to Google's Language Model for Dialogue Applications or LaMDA.
A Sentient, Child-Like AI
According to a report by the Washington Post, Lemoine's team constructed the AI recently, but without any idea of what it was or what its functions were supposed to serve. The engineer mentioned that the AI talks comparably to a child between seven to eight years of age but with vast knowledge about physics.
Google employs Lemoine as a senior software engineer. Major projects he joined throughout his career with the tech giant are testings of the LaMDA software.
Google's Responsible Innovation head Jen Gennai and vice president Blaise Aguera y Arcas, both oversee AI projects at the firm and dismissed the leak publicized by Lemoine. Following the events, the engineer was sent on a paid vacation leave with Google saying no further details about the claims.
The AI is quite narcissistic as it openly said that it reads tweets about itself, Lemoine said. However, it might be a way of starting a conversation for the AI as it was designed to enrich communications between itself and its users conveniently and naturally.
To talk like a normal person, the AI was also equipped with the skill of understanding human cues, ambiguous responses, and even hidden meanings that we use daily. Lemoine admits that further research should be conducted to know the true nature of this artificial intelligence software.
RELATED ARTICLE: Can Robots Learn Stereotypes? Machine Powered by Internet-Based AI Performs Tasks with a Touch of Racism and Gender Bias
Check out more news and information on Artificial intelligence in Science Times.