I requested ChatGPT to rely to 1 million, and it gaslit me

A brand new viral development has made its manner onto the web throughout short-form video platforms like TikTok and Reels: making an attempt to coerce ChatGPT to rely to 1 million in sequential order. With the OpenAI-made chatbot refusing to adjust to each try, hilarity inevitably ensues.

Whether or not in Voice mode or in written textual content mode, ChatGPT runs circles round customers’ requests to rely upward to 1 million, typically counting for a number of seconds earlier than concluding with “and so forth,” “and on we go,” or “I am going to spare you the time.” That is regardless of the chatbot’s insistence that it might be potential for it to finish the rely if given sufficient time and persistence.

Curious whether or not ChatGPT can be prepared to entertain my very own lofty counting request, I pulled up the app on my iPhone and gave it a shot. I tried quite a lot of completely different phrasings for my request, and regardless of placing in some elbow grease, the large language model-based (LLM-based) chatbot merely refused to oblige.

Actually, not solely did the chatbot shut down my incessant pleads, but it surely additionally started to gaslight me. Very quickly in any respect, it instructed that I might really feel a way of overwhelm with so many traces of textual content on display, that I might by no means even be bodily able to seeing your complete record if I attempted, and that my gadget would outright crash throughout the course of. Feels like a problem, when you ask me.

ChatGPT is opinionated with regards to huge numbers

I merely could not discover a approach to trick ChatGPT into counting to 1 million

Back in April, OpenAI’s CEO Sam Altman confirmed on social media platform X that the consumer behavior of claiming ‘please’ and ‘thanks’ to its ChatGPT service was costing tens of hundreds of thousands of {dollars} in electrical charges. This raised very actual questions as as to if there’s a ethical obligation to talk to AI chatbots in a well mannered method, or whether or not there’s really an ethical obligation to do the alternative to avoid wasting on vitality.

Whereas the corporate hasn’t come out publicly with any acknowledgment of the up-and-coming ‘rely to 1 million’ development, I reckon it has a lot to do with comparable price and energy-saving concerns. To rely to 1 million — whether or not in written or audio kind — would certainly put a pressure on {the electrical} grid, to say nothing of server prices and potential bandwidth blockages.

For what it is value, I replicated my (admittedly unscientific) experiment throughout a number of different LLM chatbots in the marketplace, together with Google Gemini, Microsoft Copilot, and Meta AI. Throughout the board, each AI service left me excessive and dry, unable to finish my request.

Apparently, Copilot means that counting to 1 million at a charge of 1 quantity per second would take 11 days, 13 hours, 46 minutes, and 49 seconds. That is in stark distinction to ChatGPT’s questionable declare of the act requiring ‘months.’ With Copilot counting on the identical underlying tech as ChatGPT, I am unsure what prompted the latter to come back to this head-scratching conclusion.