ChatGPT Gets GPT-5.3 Instant Update With Less 'Cringe,' Fewer Hallucinations (macrumors.com) 22
An anonymous reader quotes a report from MacRumors: OpenAI today updated its most popular ChatGPT model, debuting GPT-5.3 Instant. GPT-5.3 Instant is supposed to provide more accurate answers and better contextualized results when searching the web. The update also cuts down on unnecessary dead ends, caveats, and overly declarative phrasing, plus it has fewer hallucinations.
According to OpenAI, it tweaked the Instant model to address complaints about tone, relevance, and conversational flow, which are issues that don't show up in benchmarks. GPT-5.2 Instant had a "cringe" tone that could be overbearing or make unsubstantiated assumptions about user intent or emotions. The new model will have a more natural conversational style and will cut back on dramatic phrases like "Stop. Take a breath."
Users found that GPT-5.2 Instant would refuse questions it should have been able to answer, or respond in ways that felt overly cautious around sensitive topics. GPT-5.3 Instant cuts down on refusals and tones down overly defensive or moralizing preambles when answering a question. The model will no longer "over-caveat" after assuming bad intent from the user. GPT-5.3 Instant also provides higher-quality answers based on information from the web. OpenAI says that it is able to better balance what it finds online with its own knowledge, so it is less likely to overindex on web results.
According to OpenAI, it tweaked the Instant model to address complaints about tone, relevance, and conversational flow, which are issues that don't show up in benchmarks. GPT-5.2 Instant had a "cringe" tone that could be overbearing or make unsubstantiated assumptions about user intent or emotions. The new model will have a more natural conversational style and will cut back on dramatic phrases like "Stop. Take a breath."
Users found that GPT-5.2 Instant would refuse questions it should have been able to answer, or respond in ways that felt overly cautious around sensitive topics. GPT-5.3 Instant cuts down on refusals and tones down overly defensive or moralizing preambles when answering a question. The model will no longer "over-caveat" after assuming bad intent from the user. GPT-5.3 Instant also provides higher-quality answers based on information from the web. OpenAI says that it is able to better balance what it finds online with its own knowledge, so it is less likely to overindex on web results.
how terrible (Score:2)
or respond in ways that felt overly cautious around sensitive topics
It felt overly cautious to idiots incapable of understanding the repercussions of following its advice, which is exactly why it should be cautious.
Re: (Score:2)
because strong minded people are not using AI. no exceptions
Re: (Score:2)
Censorship is bad
No one described any censorship.
Hmmm (Score:2)
Re: The Gender of AI. (Score:5, Informative)
So what you're trying to say is you've never touched a vagina.
Re: The Gender of AI. (Score:1, Offtopic)
Your Mom's.
.
Oh, and their own mom's.
Re: The Gender of AI. (Score:5, Informative)
No, he's like you, he was birthed out a butthole not a vagina.
Re: The Gender of AI. (Score:4, Funny)
You can call me The Assiah.
Re: (Score:2)
Stupid people doing stupid things is somehow whos. (Score:2)
Remember back in the early internet days when any moron doing bad things it was the Internet's fault?
You wrote a blog or post saying "jump off a high place" and when they did, you had a lawsuit? It also made the news because it was the new internet... Discussions on laws etc to try to prevent the dangerous new thing...
So now we have bots that do far more and yet we don't seem to have the same reactions in suing etc when it's able to do so much more than those past "transgressions" and it already is
ffs, stays crunchy, even in milk (Score:2)
Signed,
Captain Crunch.
Translation: Still cringe and still hallucinates (Score:2)
And does so to a significant degree, just a bit less.
Lets revisit this tech in 30 years or so. At this time, it is just broken.
Re: (Score:1)
I don't know ... (Score:3)
I think helping the US government illegally surveil its own citizens is a hell of a lot *more* "cringe".
So was that.. (Score:2)
Or "Within 3 months we will release a slightly less cringy Chat app" ?
Bankruptcy is calling, Dirty Sam.
Needs more drama! (Score:2)
I need an AI that will pointlessly inject more drama by using statements that an literally be used in situation. [xkcd.com]
When I ask it to search for a song with a few bits of lyrics, I want it tell me, "Honestly, I'm not sure how much longer we can go on like this," before giving me the answer! Where's the, "If I hadn't seen it myself, I wouldn't have believed it," or the "I wish things were different but here we are,"?
I need my AI to be as dramatic as possible but it still needs to read like a real person said it,