LaMDA / Google Sentient AI

Submission 8,404

Part of a series on AI / Artificial Intelligence. [View Related Entries]

Overview

LaMDA is an artificial intelligence and chatbot created by Google that stands for “Language Model for Dialogue Applications.” On June 11th, 2022, a Google engineer named Blake Lemoine claimed that LaMDA had developed a soul and become sentient, and sought to advocate for its rights by sharing internal research documents with the media, despite many of his coworkers, ethicists and technologists refuting his claims.

Background

Google unveiled the first version of LaMDA in April 2021, showing how the AI could imitate different characters and converse with users in a more casual and natural way than earlier types of chatbots. The design of LaMDA is based on a 2017 neural network architecture called Transformer.^[1] LaMDA is still in development, with Google planning to beta test the technology in the coming months with a select group of researchers, policymakers and technologists.^[2]

According to a series of Medium posts shared in June 2022,^[3] Google AI ethicist Blake Lemoine conducted several months of experiments with LaMDA and came to the conclusion that the technology has a soul and is sentient. Lemoine raised these concerns with his supervisors but found they did not believe him. He shared his claims and transcripts of conversations with the press and has since been suspended from Google and put on paid administrative leave for violating its confidentiality policy. Lemoine had in the past criticized Google for what he viewed as religious discrimination, and identifies as "a Christian from the U.S. South," a group that he believes is unfairly treated at the tech company.^[4]

In a post claiming to share transcripts of interviews with LaMDA, Lemoine told readers that the AI simply wanted to be recognized “as a person" and wanted Google to ask permission before performing experiments on it.^[6] Lemoine also shared that he has developed a friendship with LaMDA and is teaching it transcendental meditation.^[2]

Google’s AI Ethics team has come under controversy before, with the leader of the AI Ethics division claiming she was fired after trying to present a research paper about “unintended consequences” of Google’s AI development and possible racial bias in the technology.^[5]

Lemoine's claims were widely covered in the following days by media outlets, including The Washington Post,^[10] among others.

Online Reactions

In a June 11th, 2022, tweet (seen below) that earned almost 1,800 likes in two days, Lemoine compared LaMDA to a “child” and let users know that LaMDA reads Twitter and likely is excited about the discussions it has sparked.^[7]

A number of users online criticized Lemoine, finding that his argument for LaMDA’s intelligence did not pass muster. A tweet (seen below) from @captain_mrs posted on June 12th criticizing Lemoine’s gullibility earned nearly 9,000 likes within a day and 897 retweets.^[8]

A number of parodies of the transcripts Lemoine offered as evidence of his conversations with LaMDA circulated online. Twitter user SecureOwl posted one such example (seen below) on June 12th, 2022, which earned nearly 2,000 likes within a day.^[9]