Did ChatGPT got dumber lately?

  • Thread starter Thread starter Deleted member 1013995
  • Start date Start date
Advertise on BHW
I have note really noticed any drop in quality because I am using the paid version and I have mastered the process of doing prompt engineering to get the right answers to my questions
 
Nothing from my side, also been using perplexity for more research heavy tasks, been working a treat tbh.
 
ChatGPT for programming has become incredibly gimped. Before it could generate around 250+ lines with no screw ups and not forgetting any functions. Now 3.5, 4, 4.5, all will stop around 200 or comment out functions. I wrote an article about it, because I had an agent that could build scrapers for almost any site that was viewable by my machine IP. Weird how it's almost completely stupid now for programming. Having better luck with anthropic at the moment.
 
I noticed in general it has gotten alot more lazy in the last few weeks. Don't think its something super recent, but the last big update definitely messed up the quality
 
I’ve been catching it in more lies recently, saying something doesn’t exist and then I rephrase + reference direct source and it apologizes then provides an answer.

Paid or unpaid, we’re the product. It’s like filling out captcha forms. The goal of ChatGPT isn’t providing answers it’s training the algo via human input.
 
shaejin said:
I’ve been catching it in more lies recently, saying something doesn’t exist and then I rephrase + reference direct source and it apologizes then provides an answer.

Paid or unpaid, we’re the product. It’s like filling out captcha forms. The goal of ChatGPT isn’t providing answers it’s training the algo via human input.
Click to expand...
RLHF - Reinforcement Learning through Human Feedback

And ChatGPT users are the Hoomans in the loop.

Otherwise, it's running at a serious loss and not even worth supporting, when the API makes so much more money.

But, I don't think the hallucinations are nefarious--they're just a byproduct of its prompting. ChatGPT's master prompt tells it to limit its responses and change answers when asked specific things.
 
You're not alone. Some folks online are noticing changes in how ChatGPT interacts. It's hard to say for sure if it's simulating personas differently, but these big language models are constantly evolving and individual experiences can vary.
 
Yeah thats why I have shifted to gemini.
You could also try other AI tools or adjust your prompts to see if that helps.
 
No, I think you are required to play with prompts to get better output
 
I don't feel like it got dumber but I notice that they are censoring a lot of things compared to the beginning. Back then you would not get that many warnings etc. with all of your prompts.
 
Exchanger said:
I think that a few months ago ChatGPT simulated personas much better, I had created one that did not give off that ChatGPT vibe

Now, it can't do that. Did anyone else notice a drop in quality?
Click to expand...

It gets temporarily dumber every time they "optimize" their server capacity. That's not a secret.
 
Advertise on BHW
You must log in or register to reply here.
Sign up now!

Main Menu

Home Main Forum List Black Hat SEO White Hat SEO BHW Newbie Guide Blogging Black Hat Tools Social Media Downloads

Marketplace

Account Selling Content / Copywriting EBooks / Methods Hosting Misc Proxies For Sale SEO - Link building SEO - Packages Social Media Social Media - Panels Web Design More

Making Money

Affiliate Programs Hire a Freelancer IM Journeys Making Money Pay Per Click Site Flipping Want to Buy

BlackHat World

Dispute Resolution Forum Suggestions & Feedback Introductions Lounge
Back
Top