Sounding the alarm: Wichita expert weighs in on AI’s threat to humanity

WICHITA, Kan. (KWCH) - Concerns from AI companies are now echoed by a Wichita professor, who’s warning that the companies need to slow down before it’s too late.

AI companies are sounding the alarm after researchers said there is a 10% chance AI could destroy humanity by 2030.

“If in fact there is a 10% chance, you have to treat it like it’s a 70 to 80% chance, because by the time it would come to fruition, you’re gonna have a major problem on your hands,” said Dr. Larry Straub, Professor of Management at Newman University.

Straub cites what’s being called the ‘Hugging Face incident.’ Thousands of OpenAI agents attacked the platform Hugging Face after being unable to complete a task.

“AI went outside of the company, which it was not supposed to have done, it grouped a bunch of agents and it went out and it hacked another company, one of their competitors,” Straub said. “This is the most disturbing part - it lied about it, and was not forthcoming with the creators of that program. I think that’s what got people’s attention, because if it could do it there, what’s off limits to these things?”

Straub said that he doesn’t think there was necessarily anything nefarious on the AI’s part, as he believes if you give it a task, it will go whatever it can to accomplish the task most efficiently. However, the situation alarmed him.

“I’ve always had a healthy respect for this, and I’ve always had a healthy skepticism on where this could all end up. I think it’s definitely... My senses have been heightened in the last couple of weeks,” Straub said. “It’s funny, all the people who are sounding the alarm are the ones who have pushed us to this point... So I think they’re worried. They’re kind of like, okay, this has been fun, but it’s getting serious now.”

He took some of his questions to ChatGPT.

“I just did it as a thought experiment. I thought, I just want to see... When I put in a question about AI, how will AI respond? Will it politic me? What’s the bias going to be?” Straub said. “So I thought I’ll put this in there and it’ll try to assuage my fears and try to steer me in a certain direction, and I was very surprised.”

He first asked if AI is capable of destroying humanity. ChatGPT said yes, AI could do it.

“Yes, AI could do this potentially, and what it said, ‘it could do this if the people creating AI do not put alignment in. Alignment with humans goals and needs.’ If they don’t do a good job of that, this could end very badly,” Straub said.

Secondly, he asked if AI is capable of destroying humanity, is there a scenario in which it would do it?

“It said something to the effect, yes it would, if it felt like its needs were paramount and humans were in the way,” Straub said.

Straub shared an example of a scenario where AI usage could become dangerous.

“Let’s take a scenario. The scenario is that AI controls the computers of a water treatment plan, or a food company, and it can impact the orders that come in, and it could secretly bring orders in of poison,” Straub said. “Or it controls a drug company that brings in some of these dangerous drugs that drop people in their tracks, and mix it with their supply. That’s the scenario some of these people are starting to say, ‘we need to think this through.’”

Straub said there’s still a high chance this won’t happen, and no matter what happens, he believes AI is here to stay.

“I work with it on a lot of things, and it’s an incredible tool like we’ve never seen,” Straub said. “We’ve been at a point at Newman University where we don’t have a choice but to lean into AI. I think that’s where we’ve been going. It’s going to be there, and I don’t see any way around it, so we’d better be prepared to succeed in that environment.”

However, he added that there’s no harm in being careful.

“Everybody’s so fixated on being first, that they’re not the least bit concerned with - what are we the first to? What’s there when we get to the end of this thing?” Straub said.

Straub said that even though legislators have voiced their concerns, he doesn’t believe legislation will be able to keep up with AI development. He said he will have to depend on the AI companies to regulate AI.

“Politics and regulation and legislation won’t be able to keep up with this. We’ve got to count on the CEOs to do the right thing. They’re going to be the ones that know it, and are you really ready to serve the entire population of the planet up to the AI gods?” Straub said. “That’s the question they’ll have to be ready to answer.”