According to a candid story in the Washington Post on Saturday, a Google engineer said that after hundreds of interactions with an advanced, unreleased AI system called LaMDA, he believes the program has reached a level of consciousness.
In interviews and public statements, many in the AI community dismissed the engineer’s claims, while some said his story highlights how technology can make people attribute human attributes to it. But the belief that Google’s artificial intelligence can be sensible probably underscores both our fears and our expectations of what this technology can do.
LaMDA, which means “Dialog Language Model”, is one of several large-scale artificial intelligence systems that have been trained on large chunks of text on the Internet and can respond to written prompts. They are essentially tasked with finding patterns and predicting which word or words should come next. Such systems are getting better at answering questions and writing in ways that may seem convincingly human – and Google itself introduced LaMDA last May in a blog post as one that can “include the free way on a seemingly endless number of topics.” “But the results can also be crazy, weird, disturbing and messy.
Engineer Blake Lemoine reportedly told the Washington Post that he had shared evidence with Google that the LaMDA was reasonable, but the company disagreed. In a statement on Monday, Google said its team, which includes ethicists and technologists, “reviewed Blake’s concerns according to our AI principles and informed him that the evidence does not support his claims.”
On June 6, Lemoine announced in Medium that Google had left him on paid administrative leave “in connection with an investigation into concerns about the ethics of artificial intelligence that I raised in the company” and that he could be fired “soon.” (He cited the experience of Margaret Mitchell, who led Google’s ethical intelligence team until Google fired her in early 2021 after her outspokenness about the departure of then-co-founder Timnit Gebru in late 2020. Gebru was removed after internal clashes, including one related to a research article in which the company’s AI management told it to withdraw from the conference or remove its name.)
A Google spokesman confirmed that Lemoine remained on administrative leave. According to The Washington Post, he was on leave for violating the company’s privacy policy.
Lemoine was not available for comment Monday.
The continued emergence of powerful computer programs trained on massive data has also raised concerns about the ethics governing the development and use of such technology. And sometimes progress is seen through the prism of what may come, not what is possible at the moment.
The responses from those in the AI community about Lemoine’s experience bounced around social media over the weekend and came to the same conclusion: Google’s AI is not close to consciousness. Abeba Birhain, senior associate in reliable AI at Mozilla, tweeted Sunday, “We have entered a new era of ‘this neural network is conscious’ and this time it will drain so much energy to disprove it.” Gary Marcus, founder and CEO of Geometric Intelligence, which was sold to Uber, and author of books including “Restarting AI: Building Artificial Intelligence We Can Trust,” called the idea of the LaMDA a reasonable “stupid thing” in a tweet. He quickly wrote a blog post stating that all such AI systems do is map models by downloading from huge language databases.
In an interview with CNN Business on Monday, Marcus said the best way to think of systems like LaMDA is as a “famous version” of autocomplete software that you can use to predict the next word in a text message. If you write “I’m really hungry, so I want to go to”, it may suggest “restaurant” as the next word. But this is a forecast made with the help of statistics.
“No one should think that auto-completion, even on steroids, is conscious,” he said.
In an interview, Gebru, who is the founder and CEO of the Institute for Distributed AI Research, or DAIR, said Lemoine is a victim of many companies claiming that conscious AI or artificial intelligence is an idea that refers to AI that can perform human tasks and interact with us in a meaningful way – they are not far away. For example, she noted, Ilya Sutzkever, co-founder and chief scientist of OpenAI, tweeted in February that “today’s large neural networks may be slightly conscious.” And last week, Google Research vice president and colleague Blaise Aguera and Arcas wrote in an article for The Economist that when he started using LaMDA last year, “I felt more and more like I was talking to something intelligent.” , claiming in an interview with the Washington Post that LaMDA, Google’s chatbot, has become “reasonable”)
“What is happening is that there is such a race to use more data, more calculations to say that you have created this common thing that everyone knows, answers all your questions or whatever, and that’s the drum you played, ”Gebru said. So how do you wonder when this person goes to extremes?
In a statement, Google said LaMDA had undergone 11 “distinctive reviews of AI principles” as well as “rigorous research and testing” on quality, safety and the ability to make fact-based claims. “Of course, some in the wider AI community are considering the long-term possibility of reasonable or general AI, but there is no point in doing so by anthropomorphizing today’s conversational models that are not reasonable,” the company said.
“Hundreds of researchers and engineers have talked to LaMDA, and we don’t know anyone else making extensive claims or anthropomorphizing LaMDA like Blake,” Google said.
Add Comment