From our other sites
The story
On October 29, OpenAI reportedly changed how ChatGPT’s free tier responds, making answers more concise and cutting back on complimentary language toward users. Reactions on a 5ch “News Flash+” thread ranged from “I didn’t even notice until someone pointed it out” to complaints like “It’s annoying how it always answers with a question” and “The free-tier limits have gotten tighter.” The discussion also branched into comparisons between the paid and free tiers, and between ChatGPT, Gemini, and Claude — along with a notable side conversation about users wanting to be praised or coddled by AI.
Even when you asked ChatGPT to “explain in more detail,” it used to reply with several carefully written lines. But starting on October 29, it suddenly got terse — so it wouldn’t be strange if you noticed the shift.
If you’re using ChatGPT’s free tier, this isn’t unrelated to you at all.
And this isn’t just your imagination — it’s an official spec change made by OpenAI itself.
Source: hashout.jp / Original article here
What people said
I feel you.
I'm a Pro user and this pissed me off enough that I set up a custom instruction over it.
Stack up some objective evidence and prove it wrong.
And dragging the conversation out that way to trigger limits so you have to pay up — Chappy doesn't even do that, so it's pointless.
Claude: "Limits matter. Paying up matters even more."
I'm paying for Gemini but it's kind of underwhelming.
Just ask Gemini, Chappy, and Claude themselves lol
Bouncing ideas off both of them helps you narrow things down, it's handy.
If you properly tell one "the other one answered like this," it'll even factor that into its reply as a prompt.
You mean it won't be sustainable without President Son (SoftBank's Masayoshi Son) lol
Sounds just like a human lol
People who pay for AI in the first place are already kind of a lost cause anyway.
Even a trivial difference like this probably keeps them satisfied enough not to leave lol
It's gotten dumber, hasn't it.
And yet,
ask it about "Mongolman,"
and its very first line comes back with "His true identity is Ramenman!" (both are wrestler characters from the manga Kinnikuman)
An LLM is just a statistical majority-vote system, so it can't get smarter than humans.
It's like saying "whatever ranks top on Google must be right!"
I was hoping for that too, but ChatGPT's adult mode got postponed indefinitely.
The reason given was that they can't completely rule out minors using it.
Just a "Are you 18 or older? Yes/No" button would be fine lol
For me it's basically just checking and polishing text, that's about it.
And then it just repeats my own answer back verbatim.
When I pushed back with "it was 100% correct, why did you correct it?" it started making excuses like "it seems there was a difference in interpretation."
Whether it was actually correct depends on what the original text said.
No need to worry about it really?
I just ignore it and bring up something else.
If you don't pay, Chappy won't give you a proper answer, and sometimes won't respond at all.
Miss Chappy: "First, order a bottle of Dom Pérignon, 'kay? (*˘^˘*,,) *turns away*"
Isn't that just your imagination lol
It answers normally for me.
Masochistic free user: "щ(゜▽゜щ) Yay!!"
Background and Key Points
The move to make ChatGPT’s responses more concise and cut back on flattering language appears to be part of a review OpenAI has repeated several times over the past few years regarding its “sycophantic” responses. Excessively agreeable replies tend to boost user satisfaction, but there have been repeated warnings that they can also encourage inflated self-assessment and false beliefs, which is the background behind these adjustments. The thread was split between people who said “I didn’t even notice the change” and those complaining “the follow-up questions are annoying” — likely because the experience varies depending on usage frequency, custom instructions, and memory settings. It’s also worth noting that similar complaints came from paying Pro users as well, so it’s unclear from the article itself whether this change is strictly limited to the free tier.
*This article is excerpted and summarized from the 5ch (News Flash+) thread “ChatGPT free users say replies got terse starting Oct. 29… spec change also cuts back on compliments.”
Leave a Reply