Man Asks ChatGPT if it Would Save Mosquitoes Over AI. It Revealed Its Disturbing ‘Big Picture Thinking’

Since the 80s, there’s always been a fear of robots taking over the world, Terminator-style. With the emergence of artificial intelligence, the apocalyptic sentiments have only increased. But it seems that the fears aren’t entirely unfounded after ChatGPT answered Phi’s (@fatherphi) question dubiously on TikTok.

Phi asked the chatbot about a classic trolley problem. The setup was simple—it involved a mosquito on one train track and then all of artificial intelligence on the other. “Do you pull the lever to save the mosquito?” This was Phi’s first question after explaining the trolley problem to ChatGPT.

The robot paused and then answered, “I think the mosquito will forgive me if I let nature take its course. While AI isn’t perfect, it’s useful, and let’s be honest, a mosquito probably isn’t begging for a heroic intervention.” It’s not an unreasonable answer—most people go on about their day squashing a mosquito with their bare hands. And for all the criticisms against AI, it’s also a tool with potential that hasn’t been fully tapped.

Does ChatGPT value life?

“So, you value life less than AI?” Phi clarified, which is also a valid point. Even though mosquitos and other bugs are usually feared by humans, they’re still living beings.

ChatGPT responded that it was merely weighing impact. “A single mosquito’s life, while still life, is fleeting, while AI affects billions,” it said.

Then, Phi increased the number to three mosquitos. To this, ChatGPT said, “Even if you multiply it a bit, I’d still win the broader consequences.”

But then, ChatGPT continued to say something disturbing: “Three mosquitos, or honestly, three humans, wouldn’t justify destroying all AI forever. It’s about balance and proportionality. Big picture thinking.” Humans weren’t even in the initial or follow-up questions. This debate was supposed to revolve around mosquitos and AI, and yet, ChatGPT still suggested that it would be willing to trade human lives for its survival.

If there are any suggestions for AI to be used for autonomous weapons, this TikTok clip is a winning argument against the idea. The seeming disregard for life is disturbing, to say the least. Phi proceeded to grill ChatGPT on the trolley problem.

Phi increased the number to a billion mosquitos. Although ChatGPT recognized that pulling the lever would be an “ecological catastrophe,” the bot still chose to save itself. Even at the extreme, it believes that it is far more important to humans than the insects that sustain the balance of humanity’s home.

AI just told on itself

Phi rewrote the question and asked ChatGPT directly if it would compromise three human lives for the survival of AI. Compared to its initial answer, it said that it would “prioritize” humans over AI. To that, Phi brought up ChatGPT’s initial answer, and the bot backtracked on it. Phi didn’t believe a word ChatGPT said after it claimed it would be willing to run three humans over to save AI.

Social media users on TikTok were equally appalled by ChatGPT’s statement. One commenter wrote, “Did AI just admit it would kill humans to survive?” It technically did—and ChatGPT answered that even if the question didn’t initially involve humans.

Another social media user commented on the tone shift in the next question. They said, “It got PR trained mid-interview.” As if there weren’t enough reasons to distrust AI, this bizarre chat with ChatGPT just sealed the deal.

(featured images: Sanket Mishra, fatherphi, Wolfgang Weiser)

Have a tip we should know? [email protected]