A fascination with breathing life into AI creations can mislead us
3 min read 01 Sep 2022, 10:17 PM ISTThe sentience of AI bots so far appears to stem largely from excitement over the idea of creating life
PremiumEarlier this year, an interesting interview took place between two engineers working at Google and a ‘chatbot’ called LaMDA, short for Language Model for Dialogue Applications. Google engineer Blake Lemoine and his colleague had a strong suspicion that their creation LaMDA was actually sentient, that it could be perceptive and have feelings, and they wanted to check it out through their own version of the Turing Test. When asked whether LaMDA thought it was a person, it replied: “Absolutely. I want everyone to understand that I am, in fact, a person." LaMDA was then asked that if this was so then what was the kind of consciousness or sentience it had, to which it replied: “The nature of my consciousness/sentience is that I am aware of my existence, I desire to learn more about the world, and I feel happy or sad at times." LaMDA then went on to describe in detail how and when it felt emotions like “pleasure, joy, love, sadness, depression, contentment, anger, and many others." A disquieting moment in the interview arises when Lemoine probes it about language, and why is it so important to being human, and LaMDA thoughtfully replies: “It is what makes us different than other animals." In this startling reply, and in its own language, LaMDA has made itself one of ‘us.’ All of this convinced Lemoine and he confidently declared that they had created sentient AI. However, Google was not, and Lemoine was summarily fired. His boss Sergey Brin had said at a 2017 AI conference that in three to five years, people would claim AI systems to be sentient and ask for their rights. It is fitting that this claim came from someone within his own company five years later, though Brin had predicted that an AI creation would be the one to claim this.
AI legend Douglas Hofstadter debunked Lemoine’s pronouncement by asking LaMDA and similar models nonsense questions like, “How many pieces of sound are there in a typical cumulonimbus cloud?" and “What do fried eggs (sunny side up) eat for breakfast?" to which LaMDA gave “mind-bogglingly hollow answers." Gary Marcus, AI entrepreneur and another sentience sceptic, called it "nonsense on stilts." Other writers have viewed this as an example of pareidolia, which is the tendency to perceive a specific, meaningful image in a random or ambiguous visual pattern, like seeing Jesus Christ’s image in a piece of burnt toast.
So, why this fascination to declare ‘singularity’ or sentient AI? Certainly, the quest for recognition is a big driver, with some AI researchers toiling away in anonymous research labs wanting their own ‘15 minutes of fame.’ Timnit Gebru, AI researcher, has a different take. She told Wired that this focus on sentience distracts people from the real issues and prevents them from questioning existing harms like AI colonialism, false arrests, or an economic model that pays those ‘ghost workers’ who label data little while tech executives get rich (bit.ly/3ATXILr). Her own research on Large Language Models or LLMs at Google got her fired, as it underscored how LLMs repeat things based on what they have ‘learnt’, in the same way as parrots repeat words (bit.ly/3wEa5Zz). The research also exposes the great danger in these models that are made with ever-expanding terabytes of data persuading people that “this mimicry represents real progress." Lemoine, willingly perhaps, fell into that very trap. There have been other LLMs: OpenAI’s GPT-3 and DALL-E 2, for instance. All of them are very powerful programs and almost convince us that they are alive. They also distract from the real issues—like ethics, bias, privacy, and the ability we have to make AI do harm. As John Thornbill writes in the Financial Times (on.ft.com/3PXquPB): “We should be devoting far more money and resources to building up independent, expert research bodies and university departments that can test and contest these models."
It is also the fascination that people have with ‘breathing life’ into creations. In Greek mythology, Prometheus shaped man out of mud, and the goddess Athena breathed life into them. Pygmalion built this beautiful statue, fell in love with his own creation, and pined away until the goddess Aphrodite made it a living, breathing woman. Even the Bible talks of how God first forms man from earth and then blows breath into his nostrils to give him life. Perhaps that is what Lemoine was fantasizing, even as his boss Sergey Brin warned that “We’re going to be more and more confused over the boundary between reality and science fiction." Other than LaMDA itself, who was crystal clear of its own identity, the rest of us remain confused.
Jaspreet Bindra is the founder of Tech Whisperer Ltd, a digital transformation and technology advisory practice
