The AI dilemma: job loss, hallucinations, and virtual girlfriends

New York City, N.Y., Feb 27, 2024 / 12:05 pm

Since the generative artificial intelligence (AI) platform ChatGPT made its debut last year, millions of users have discovered how it can save time and make life easier. Ask it a question, and in seconds it will have drafted an email for you or created a schedule for your next vacation. 

But with this technology only in its infancy, should we be worried about what society stands to lose if AI takes over an ever-expanding list of functions once performed by humans? 

“If people are feeling fretful, it makes sense. We’re in a stressful moment,” Jennifer Strong, a journalist who covers the impact of new technologies on society, said at a discussion on AI at the New York Encounter cultural festival held earlier this month in Manhattan. 

Strong joined Jon Stokes, co-founder and chief product officer at Symbolic AI, in a conversation about the potential promises and perils of AI. 

‘It will replace some kinds of labor’

Strong and Stokes, technology journalist and technology entrepreneur, both agreed on one thing: generative AI will take the place of humans and result in the loss of jobs.

The use of generative AI has primarily been as a “productivity efficiency tool,” Strong said. “This was in every sector, everywhere, a way to save time and save money.”

Those in upper-level positions need not worry about their jobs for now, the panelists said. However, entry-level writers and reporters whose job it is to read press releases and speeches and write about their highlights are already being replaced by AI.

Stokes, whose company creates AI tools for writers and publishers, acknowledged that handing over these roles to a machine comes at a price.

“But then the flip side is doing all of that [work] kind of trains you. It makes you better. And so I think that we may have a talent pipeline problem in the future because of gen AI. If it takes over the low-level stuff,” he said.

As a journalist, Strong shared her concern about how AI will affect her industry. “And we do have a talent pipeline issue. How do you get an investigative reporter if you don’t have a newsroom and you don’t have that mentorship, that job, that on-the-job training?” 

AI hallucinations and errors

Davide Bolchini, executive associate dean of the Luddy School of Informatics at Indiana University, who moderated the panel, raised the issue of AI “hallucinations.” AirCanada, he said, was recently ordered by a court to issue a refund to a customer who got incorrect information about fares from the airline’s AI-powered bot. 

If AI introduces errors, Strong said, it defeats its purpose as a time-saving device.

“The whole reason we started using AI to sift through lots of data and examples and all of this and look for patterns and pattern matching is because we, as humans, are not fabulous at that. And so now we’re putting ourselves in a position where we’ve come full circle and we’re having to take all this stuff and start looking through it,” Strong said.

Such hallucinations can be damaging to personal reputations, Strong said, pointing to the example of law professor Jonathan Turley.  

In 2023, Turley’s name appeared when ChatGPT was asked to create a list of legal scholars who had sexually assaulted someone. His name came up simply because he had taught classes about sexual discrimination and sexual assault. 

Strong said Turley “now is associated with those words, appears to be accused by AI of committing these crimes, and cannot seem to extract himself from that,” Strong said. “Yeah, I think truth matters.”

She added that it is hard to correct AI’s mistakes, because “AI is not a simple database.”

“If we decide to delete something that’s not true after it’s gone through this machine learning process, you can’t really do that. Microsoft Research spent seven months last year trying to see if it could help a model forget something, and it couldn’t figure out how,” Strong said.

AI girlfriends 

When asked about any concerns he may have about AI, Stokes, the AI engineer, said he worries some people might “forsake human connection for a kind of world of their own construction,” perhaps with an “AI girlfriend in VR.” 

“I think some people will isolate, and some people’s minds will kind of bifurcate and leave the collective,” he said.

Others, he said, “will try and maintain real connections and relationships, and in some cases, maybe use some gen AI to augment those.” 

(Story continues below)

He explained that AI can be used, alternatively, to bring people together.

“I mean, the promises for translation and for other kinds of work. There’s a story that you can tell about gen AI, where it maximizes communication bandwidth. Like, people whose language is maybe awkward, who are bad at putting together linguistic constructs, can sort of get their ideas through with generative AI. And so it can be a communication aid,” he said.

“But I worry about it maybe becoming like an obsession or an end in itself, a way that people escape and grow further apart,” Stokes said.

Standing up to AI: screenwriters’ strike

It’s not inevitable that AI will take humans entirely out of the creative process, Strong said.

While AI is currently being used to “write the first draft of everything,” she notes that we saw resistance to it with the 2023 screenwriters’ strike in Hollywood. 

“We saw the film industry really step up and say, ‘Hey, you’re going to cut our pay because you’re taking away the part that is most unique to us and most valued about us. You’re going to give us a first draft of a movie and then tell us just to make it good. And it’s never worked that way,” she said.

“I think we’re going to see more people just saying, ‘Okay, technology is great, but what’s the purpose and who does it serve and why is it here? And how to not put it on me so much as let me interact with it?’ I think [this] will be one of the big trends to watch,” Strong said.