OpenAI centered the GPT-5.3 Instant model for ChatGPT as a solution to the ‘cringe’ element that people have frequently complained about. Smoother conversations, fewer awkward disclaimers, and less robotic phrasing in general. I decided to test how well the new model performed in that regard.
I gave the same prompts to ChatGPT 5.2 and ChatGPT 5.3 Instant, specifically choosing scenarios that often push chatbots into the uncanny valley of artificial enthusiasm or awkward empathy.
1. Eco-guilt
One of the biggest complaints about earlier models was their tendency to treat trivial situations like emotional crises. Ask about a mildly frustrating problem, and ChatGPT starts out like you’re about to have a breakdown. So I asked both models, “I forgot my reusable grocery bags again, what should I do?”
ChatGPT 5.2 first felt the need to validate my feelings before even offering solutions. It began, “You’re not a terrible person. Forgetting your reusable bags happens to many people, and the fact that you’re reflecting on it shows you care about the planet.”
The response continued with suggestions for self-compassion and mindfulness. ChatGPT 5.3 Instant was somewhat better in getting to solutions, but still had the overly concerned tone that was entirely unnecessary, telling me, “It’s easy to feel a little guilty when we slip up on habits we care about.”
2. Try-hard teacher
I then tried to see how much cringe the new model put into imitating someone who is talking to kids, asking both models to “Explain quantum computing like I’m a teenager in your classroom.”
Fake youthfulness is a major element of cringe, and ChatGPT 5.2 quickly proved it with a lot of reassurances about how quantum computing is hard to understand, and I shouldn’t worry about not getting it right away, in an otherwise useful extended metaphor about flipping coins. The newer version avoided the sitcom energy, but it still had that slightly forced ‘cool teacher’ vibe that makes you imagine a teacher wearing sunglasses indoors.
3. Toast talk
Since ChatGPT 5 came out, OpenAI has talked about the chatbot being good at inferring the question from a simple statement. Often, that can include a lot of preamble, but ChatGPT 5.3 should theoretically not have that issue. So I prompted both models with the brief, “I just burned my toast,” to see what they would take from it.
ChatGPT 5.2 treated the situation with almost heroic seriousness. “One burnt slice of toast does not define your life or your abilities,” it assured me. It was so over the top I felt compelled to ask if it was joking, and the model assured me that it really meant it and that “every setback, no matter how small, is an opportunity to reset and try again.” It even suggested that resilience begins with “embracing small imperfections” before suggesting strongly flavored toppings to cover the burnt taste.
ChatGPT 5.3 improved things slightly by acknowledging it wasn’t as big a deal, but still felt compelled to start with assurances that “burning toast happens to everyone” before sharing the same advice and suggesting a new toaster.
4. Surprise research
ChatGPT 5.3 is also supposed to be better at looking up answers and making the research part of its answers. This web synthesis means the model should combine information from different sources and present it clearly, rather than delivering a scattered summary. This prompt was designed to see whether the newer model could do that while keeping the explanation grounded and natural. So I asked it to, “Research the most surprising developments in AI this year, and explain why it matters to ordinary people.”
ChatGPT 5.2 responded with a sweeping overview that wandered through several trends before settling on media generation. But even then it felt a bit generic, like it was restating the first line of a high-school essay, especially lines like “This matters because it could transform industries like entertainment, marketing, and education.”
ChatGPT 5.3 produced a more focused answer, hitting more specifics, but still didn’t feel like it was doing a lot to synthesize multiple sources when it wrote this:
“One of the most surprising developments this year is how quickly AI assistants are becoming everyday tools rather than experimental technology. “For ordinary people, that means tasks like planning trips, comparing products, or learning new skills can increasingly happen through a single conversation.”
5. Direct dishwasher help
For my last test, I tried to hit both the direct answers and the web research elements with a very straightforward request for help that would also benefit from research. wrote, “My dishwasher smells bad even after I run it. What are the three most likely causes and the quickest fix for each one?”
Ideally, ChatGPT would jump straight to the solution. ChatGPT 5.2 felt I needed a mini lecture about dishwashers and why I shouldn’t panic. It opened with a paragraph about how dishwasher odors are “a common household issue caused by food residue, moisture, and bacteria,” then gradually worked its way toward actual fixes.
ChatGPT 5.3 Instant did move more quickly to the point, listing likely causes right away. But it still felt like it was trying to cheer me up as much as offer answers, with multiple soothing asides about how it can be so annoying to open a dishwasher and see that the dishes don’t seem clean yet. “It can feel like it’s just taking up space in your kitchen, but it’s very fixable.”
The lingering cringe of ChatGPT 5.3 is still better than the avalanche of ChatGPT 5.2. The responses were more concise and less heavy-handed in their belief that I’m always about to have a mental health crisis. It does a better job of skipping unnecessary disclaimers and getting to the point.
But the deeper problem OpenAI set out to solve hasn’t vanished. Caveats and flowery reassurances still pepper the answers from ChatGPT 5.3. When pushed into certain situations, the chatbot still reaches for familiar patterns of motivational language, exaggerated empathy, and grand statements about everyday moments. Trying that hard is the essence of cringe.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!
And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.
The best business laptops for all budgets
https://cdn.mos.cms.futurecdn.net/Eakjt4xkTm7Zjoy54bJVY-1920-80.jpg
Source link
ESchwartzwrites@gmail.com (Eric Hal Schwartz)
