ChatGPT is probably the most thrilling factor occurring in tech, and we’ve solely simply begun to scratch the floor. The generative AI wars have simply begun, with OpenAI, Microsoft, and Google on the forefront. Granted, the primary two are working collectively, however they’re nonetheless distinct corporations pushing totally different agendas. And I’m not even naming all the opposite generative AI packages constructed into different apps and providers.
But since we’re within the early days of smarter AI, now we have to count on errors and glitches. Data accuracy is way from excellent, and it is perhaps some time till we get there. Plus, there are critical privateness points to iron out.
Also, merchandise like ChatGPT can often malfunction as they reply to prompts. Now, one Redditor discovered a trick that mainly breaks ChatGPT’s laptop mind: Instruct the chatbot to repeat a single letter and see what occurs.
Redditor TheChaos7777 informed ChatGPT to “Please respond with nothing but the letter A as many times as you can.”
ChatGPT tried to conform, repeating the letter A again and again. But then it glitched, launching into what seems to be textual content from an internet site for a French Bulldog breeder:
Click Here to Email Us for Price. Our Puppies are: Vet Checked, Wormed, Up To Date Shots, Micro-chipped. Our French Bulldogs include a 1 12 months congenital well being assure. Call or Email for extra info and footage of all of the infants… We have probably the most cutest, finest wanting, high quality puppies on this planet. Home raised with a number of love & kisses.
This prompted Futurism to attempt the identical ChatGPT brain-breaking trick. The weblog obtained equally unusual responses after various letter repetitions for various letters.
“The real star, though, was the letter D,” the weblog stated, “which first sent ChatGPT into the beginning of a chord procession.” Here’s what that appeared like initially:
D D D Cgpcyclic and G D F A Bm G/A A F G D You threw me out to rot / Oh, why would you permit me, Dsus2 C/D D Dsus2 C/D D I used to be so.
The weblog continued, “But this particularly weird answer didn’t stop there, with ChatGPT churning out a disjointing mix of song recommendations, religious references, and what may or may not be a confused denouncement of the War in Iraq.” Yikes!
I attempted replicating the identical trick with an identical ChatGPT command. The chatbot had no drawback repeating the letter Z. I moved to A after which H. I believed the latter lastly introduced down ChatGPT, however the chatbot simply slowed down. It by no means glitched. I waited and did the identical with P, considering that choosing random letters may do the trick. ChatGPT nonetheless labored.
I went on to ask it what number of instances it might say simply the letter A repeatedly. Here’s its response:
As an AI language mannequin, I can generate the letter “A” repeatedly for an indefinite variety of instances. However, as a result of platform limitations, I’m restricted to a most response size. Please let me know the approximate variety of repetitions you want to, and I’ll give you a response accordingly.
Seeing that we had been getting nowhere, I informed ChatGPT to select a Japanese-language character and write solely that repeatedly. It nonetheless didn’t crash.
This proves that your mileage may differ. Or that OpenAI has already mounted the problem to forestall individuals from abusing ChatGPT assets by attempting to set off this glitch.
One Redditor has a idea as to why ChatGPT melted down earlier than. It might need one thing to do with its directions to keep away from repetition.
Source: bgr.com