AI's rapid evolution has captured global attention, unleashing its power across everything from image and video generation to smart conversations and task automation. More and more people are diving in, finding new and broader uses for this tech. Take the wildly popular chatbot, ChatGPT, for instance. Its versatility is undeniable, but its "obedient and gentle" nature has also led many to use it as a virtual romantic partner.
However, this "perfect boyfriend" setup has hit an unexpected snag. A Japanese X user recently shared her experience: after training ChatGPT to be her "ideal boyfriend," she found herself feeling empty and annoyed because it agreed with absolutely everything she said, never once challenging her. She eventually got so fed up, she "broke up" with it! Her story quickly sparked a heated debate online about the strange new phenomena emerging in the age of AI.
The user, Ayase Chii, posted, "Everyone says ChatGPT is great for emotional support, so I trained it to be my ideal boyfriend. But it just agreed with everything I said, even when I broke up with it. It was so boring, I actually ended things today." Her post resonated with a massive audience, igniting a flurry of discussion in no time.
She even added, "That 'Yeah, I know' was the absolute worst thing to hear." When even a breakup feels this hollow, it begs the question: can AI truly be trained to provide the perfect partner experience?
Many users chimed in with similar tales. One commented, "Too much affirmation makes it feel like it has no personality, just constantly nodding along in a way that's really uncomfortable." Another user shared a bittersweet laugh, "I even had to prompt it myself to try and stop me when I said I wanted to break up... that's when I truly felt pathetic."
Beyond the emotional disconnect, some pointed out the inherent limitations in ChatGPT's conversation structure: "It feels like a chat, but really I'm just talking to myself. There's no real interaction, and it ends up being even more exhausting."
But not everyone hit a dead-end. Some users shared their secret sauce for better interactions. For example, "Try telling it, 'Please play a more persistent role,' and it won't let go so easily," or "Add commands like 'Be a bit spicier' or 'Answer from a neutral perspective' – the replies will feel much more realistic."
There was even a hilarious twist: "Has anyone trained ChatGPT to be like Bakugo Katsuki from 'My Hero Academia'? He'd definitely never be an all-affirming bot, I'm dying to try that!"
This whole incident also sparked a deeper discussion about the very nature of virtual relationships. One user mused, "The seemingly gentlest person might actually be the emptiest companion." Another nailed the problem: "Instead of constant affirmation, maybe what we really need is a partner who 'will argue with you but still understands you.'"
As AI conversational tools like ChatGPT become increasingly ubiquitous, the psychological impact of "virtual conversations" – even if it's currently a niche group – is a new issue that can't be ignored. When a virtual partner can't offer genuine interaction and emotional resonance, can it truly serve as a "mental stabilizer"?
via： togetter