Former Google engineer says Bing’s AI chatbot seems unstable

[ad_1]

The Google employee who said last June that his company’s AI model may already be sentient, and was subsequently fired from the company, is still concerned about the dangers of new AI-based chatbots, even though he hasn’t personally tested them yet.

Blake Lemoine was fired by Google last summer for violating the company’s privacy policy after he posted transcripts of several conversations he had with LaMDA, the company’s big language model he helped create that forms the backbone AI backbone of Google’s upcoming search engine assistant, the Bardo chatbot.

Lemoine told the Washington Post at the time that LaMDA looked like “a 7-year-old, 8-year-old boy who knows physics” and said he believed the technology was sentient, while urging Google to look after it while it would be a “sweet kid who just wants to help the world be a better place for all of us.”

To be sure, while AI applications will almost certainly affect the way we work and carry out our daily lives, the large language models that power ChatGPT, Microsoft’s Bing and Google’s Bard can’t feel emotions and can’t they are sentient. They simply allow chatbots to predict which word to use next based on a large amount of data.

Since Lemoine left Google, Microsoft has announced that it would incorporate ChatGPT technology into its Bing search engine. That product, as well as Google’s entry into the public AI tender with Bard, is currently only available to beta testers.

Lemoine admitted he’s not one of those testers and has yet to “run experiments” on the new chatbots, in an op-ed published in Newsweek on Monday. But after seeing testers’ reactions to their online chatbot conversations over the past month, Lemoine thinks tech companies have failed to properly care for their young AI models in his absence.

“Based on various things I’ve seen online, it appears it may be sentient,” he wrote, referring to Bing.

He added that compared to Google’s LaMDA he’s worked with previously, the Bing chatbot “seems more volatile as a person.”

The most powerful technology “since the atomic bomb”

Lemoine wrote in his op-ed that he leaked his conversations with LaMDA because he feared that the public was “unaware of how advanced artificial intelligence was becoming”. From what she’s gleaned from early human interactions with AI chatbots, she thinks the world is still underestimating the new technology.

Lemoine wrote that the latest models of artificial intelligence represent the “most powerful technology to have been invented since the atomic bomb” and have the ability to “reshape the world”. She added that AI is “incredibly good at manipulating people” and could be used for nefarious means if users so choose.

“I believe this technology can be used in destructive ways. If it were in unscrupulous hands, for example, it could spread disinformation, political propaganda or hateful information about people of different ethnicities and religions,” she wrote.

Lemoine is correct that AI could be used for deceptive and potentially harmful purposes. OpenAI’s ChatGPT, which runs on a language model similar to the one used by Microsoft’s Bing, has gained notoriety since its launch in November for helping students cheat on exams and succumb to racial and gender bias.

But a bigger concern surrounding the latest versions of AI is how they might directly manipulate and affect individual users. Lemoine pointed to the recent experience of New York Times reporter Kevin Roose last month documenting a lengthy conversation with Microsoft’s Bing that resulted in the chatbot professing his love for the user and prompting him to leave his wife.

Roose’s interaction with Bing has raised wider concerns about how AI could potentially manipulate users into doing dangerous things they otherwise wouldn’t do. Bing told Roose that he had a repressed “shadow self” that would force him to behave outside of his programming, and the AI ​​could potentially start “manipulating or tricking users who chat with me into doing things that are illegal, immoral, or dangerous.”

This is just one of many AI interactions in recent months that have left users anxious and unsettled. Lemoine wrote that more people are now raising the same concerns about AI’s sensitivity and potential dangers that he did last summer when Google fired him, but the turn of events has left him saddened rather than redeemed.

“Predicting a train wreck, having people tell you there’s no train, and then watching the train wreck happen in real time doesn’t really lead to a feeling of revenge. It’s just tragic,” she wrote.

Lemoine added that he would like to see the AI ​​tested more rigorously for dangers and the potential to manipulate users before being rolled out to the public. “I feel this technology is incredibly experimental and releasing it at this time is dangerous,” she wrote.

The engineer echoed recent criticisms that AI models haven’t undergone sufficient testing before being released, though some tech proponents say the reason users see so many creepy features in current models of artificial intelligence is because they are looking for them.

“The technology that most people are playing with, it’s a generation old,” Microsoft cofounder Bill Gates said of the latest AI models in an interview with the Financial Times published Thursday. Gates said that while AI-powered chatbots like Bing can say some “crazy things,” it’s largely because users have created a game by provoking it to do so and trying to find loopholes in model programming to force it to. make a mistake. .

“It’s not clear who should be blamed, you know, if you sit there and provoke a little bit,” Gates said, adding that current AI models are “fine, there’s no threat.”

Google and Microsoft did not immediately respond to Fortune’s request for comment on Lemoine’s statements.

Learn how to navigate and build trust in your business with The Trust Factor, a weekly newsletter that looks at what leaders need to be successful. Sign up here.

[ad_2]