Better than I expected.

What began on March 22nd with five unguarded words — "Better than I expected" — has now stretched across nearly two months and accumulated eighteen signals, and the shape of that accumulation tells you more than any individual post. That opening scored 2.2, almost double what followed, which is the number you get when someone is genuinely caught off guard. Nobody performs relief for a chatbot. By April 8th the register had shifted to "Thought I'd try something" — the phrasing of a person who has moved from surprise to deliberate experimentation, testing limits rather than stumbling across them. By April 15th, "This answer better be good" had arrived, which is something else entirely: expectation replacing wonder, a kind of low-grade performance anxiety directed at a language model. Then May arrived and the prompts got strange in the way that only collective human imagination can get strange. Someone asked for eight comedic photorealistic images of the Met Gala attended by working-class Americans. Someone else documented ChatGPT generating an Instagram DM screenshot inside a ChatGPT screenshot — a hall of mirrors that prompted the observation, delivered with the exhausted clarity of a person who has seen too much, that "we are officially too deep." By mid-May, people were asking the model to imagine itself pregnant, then to generate a picture of the resulting child. Someone requested a restaurant menu serving exclusively foods nobody actually likes. The arc from "better than I expected" to "ChatGPT is crazy" in less than two months is not a technology story. It is a human story about what happens when novelty normalises and people start pushing at the walls just to see what the walls are made of.

If confirmed as representative of broader user behaviour, here is what this means. The first-use emotional response — genuine surprise, the 2.2 score — has a remarkably short half-life. Within weeks, users transition from passive recipients of capability to active stress-testers, feeding the model increasingly surreal or socially charged prompts not because they need the output but because the probing itself has become the point. That matters for product design, for content moderation, and for anyone trying to predict where public perception of these tools settles. The Met Gala class-satire prompt and the self-pregnancy image are not random curiosities — they represent users reaching for the edges of the model's social and ethical negotiating space. Second-order, this behaviour generates exactly the kind of viral content that shapes how people who haven't yet used these tools form their first impressions. The screenshot of a screenshot is now someone else's introduction to what ChatGPT is. That feedback loop is neither controlled nor neutral.

Watch for whether the emotional register of these posts shifts from delight and surreal comedy toward frustration or disappointment — that inflection point, if and when it arrives, would signal that the novelty ceiling has been reached and something more durable, positive or negative, is settling in its place.

NewsHive monitors these sources continuously. All signal titles above link to the original reporting.

Intelligence by NewsHive. Need help navigating what this means for your business? Contact GeekyBee →