Ayush

It's for everyone.

When people who are affected by AI have a voice and it's now in use, the outcomes are fairer, smarter, and more dramatic.

All right, this week, OpenAI released GPT-5, and there have been some mixed reviews about it.

So this is a special bonus episode in the middle of our regular episode of AI Square.

I'm Ayush.

The hype promised a revolution.

Bigger models, smarter reasoning, next level creativity, and even more accurate real-time knowledge.

We were told it would be faster, more context aware, better at following complex instructions, and capable of deeper reasoning than any AI before it.

On Reddit, in AI Discord servers, and across Twitter, the critiques are piling up.

Watered-down answers, less creativity, and constant refusals to respond to harmless prompts.

Some even describe it as a corporate-approved chatbot instead of an actual thinking partner.

Exactly.

Others are frustrated by performance, math errors that shouldn't happen, outdated information despite claims of being more current, and in long chats it sometimes loses track of the content entirely.

Here's a popular theory.

GPT-5 is turned for compliance over creativity.

That might make it safer for mainstream audiences, but it's also making it less adventurous and less engaging.

Some call this the alignment tax, the idea that every additional layer of safety and modernization comes at the expense of spontaneity, nuance, and creative freedom.

Now this makes sense due to AI releasing GPT-5 not just for the plus viewers but also for the free users as well.

Then there is the raw performance angle.

Side by side tests circulating online showed GPT-4 outperforming GPT-5 on complex reasoning, tricky logic puzzles, and niche technological topics.