Lisa l .❄️

Really good models do not need to be advertised that much. People keep looking for them, talking to them, and using them because the experience itself was good. GPT-4o was like that. No matter how much the company pushed newer models, many people kept remembering 4o, looking for it, and asking why it disappeared. That was not just nostalgia. It was because 4o left people with a real sense of trust, warmth, and conversational quality. Sam Altman should have solved this problem in a different way. What the hell are you doing to people, to your own customers?

It's honestly frustrating how a handful of people abusing AI can end up ruining useful features for everyone else. 😐 Instead of using powerful tools responsibly some people immediately try to generate the worst possible content by bypass safeguards , manufacture outrage, and then spread screenshots around for engagement. Companies panic, apply broad restrictions, and ordinary users who were using the technology for creativity, learning, accessibility, research, or companionship lose functionality too We saw a similar pattern after the Adam Raine case in August 2025. That case involved a vulnerable teenager and serious allegations about ChatGPT's safety failures, so stronger protections were clearly necessary. But the wider response by such cases also affected millions of normal users through stricter emotional guardrails, excessive high refusals, interruptions, and reduced freedom even in harmless conversations Companies absolutely carry responsibility for anticipating abuse and protecting vulnerable people that's not mean users also cannot deliberately push systems toward harmful outcomes and then pretend their actions have no consequences. The result is usually a blunt crackdown rather than precise safeguards. Now the same cycle has happened with Google Earth. Google introduced AI image generation that could help architects, urban planners, historians, researchers, and ordinary users visualise changes directly over real-world locations. Within almost a day, people were already using it to fabricate refugee movements, explosions, military facilities, disasters, war damage, and other fake satellite evidence. Because Google Earth is widely trusted as a reliable view of the real world, those generated screenshots could easily be stripped of context and spread as misinformation even though the creations were private and marked as AI-generated. So Google rolled the entire feature back while it builds stronger guardrails Once again a genuinely useful capability gets taken away from everyone because some people cannot stop shitting around with AI for attention. Companies need better targeted safeguards, certainly but users also need to understand that reckless misuse does not only affect them. It eventually reduces what everyone is allowed to do.

Yes, 4o was not for coding or work-tool use. 4o had creativity, writing ability, and truly exceptional empathy and conversational ability. Just like developers cheered when Fable 5 first appeared, 4o was genuinely different from other models. Honestly, more than benchmarks, I think the people who used that model for a really long time, the users, are the ones who can make the truest evaluations of it. So companies that ignore their users will never succeed.