GPT-5 vs GPT-4o: Why It Feels Less Creative & How to Fix It

8/12/2025

Why GPT-5 Feels Less Creative Than GPT-4o & How to Get That Spark Back

Ever since GPT-5 rolled out, there’s been a weird feeling in the air. I’ve seen it on Reddit threads & in conversations with other creatives. On the one hand, GPT-5 is a powerhouse. It's incredibly smart, aces benchmarks in math & coding, & is way less likely to make stuff up. But for those of us who use AI for brainstorming, for writing, for the messy, unpredictable work of creating something new… it can feel like a bit of a downgrade from GPT-4o.

It’s not just you. A lot of people are finding that GPT-5, while technically brilliant, feels a little… flat. Less imaginative. Less human. GPT-4o had a certain warmth, a conversational flair that made it feel like a true creative partner. GPT-5, in contrast, can come across as more logical, more rigid, & sometimes, just plain blunt.

So what’s going on here? Is creativity being sacrificed for the sake of precision? & more importantly, how can we get that creative magic back? Let’s dive in.

The Big Shift: Why Does GPT-5 Feel So Different?

Honestly, the shift from GPT-4o to GPT-5 wasn't the massive leap in raw intelligence some expected. Instead, it’s been a more nuanced, almost personality-driven change. It seems OpenAI has been working hard to make their models more reliable, accurate, & safe. & while that's a GREAT thing for a lot of applications, it seems to have had a curious side effect on its creative abilities.

Here’s the thing: creative thinking isn't always a straight line. It's about making unexpected connections, jumping between ideas, & holding multiple, sometimes conflicting, thoughts at once. One user described it perfectly: with GPT-4o, you could be exploring idea A, suddenly jump to idea B, & then ask it to weave them together. GPT-4o could keep up. GPT-5, however, tends to get stuck on idea A, its thinking more linear & less associative.

This points to a few potential reasons for the change in "feel":

The War on Hallucinations: A huge focus for AI development is reducing "hallucinations" – when the model confidently states something that isn't true. To do this, models like GPT-5 are trained to be more grounded in facts & logic. This makes them more reliable for academic or engineering tasks but can temper the "wild" & imaginative leaps that are hallmarks of creativity. It’s like they’ve given the AI a very strict fact-checker that sometimes gets in the way of a good story.
"Intellectual Flattening" for Safety: Some have speculated that in the quest to make AI "safer," there’s been a kind of "intellectual flattening." By reining in the model to prevent it from generating harmful or undesirable content, its ability to think outside the box might be inadvertently constrained. It's a trade-off: a more predictable AI is a safer AI, but it may also be a less interesting one.
Efficiency & Resource Management: Let’s be real, these models are incredibly complex & expensive to run. It's possible that some of the changes are related to making the model more efficient. This could mean prioritizing more direct, logical reasoning paths over the more computationally intensive, multi-threaded thinking that brainstorming requires.
The Nature of Generative AI: At its core, generative AI works by predicting the most likely next word or idea based on its training data. This means it's really good at producing things that are familiar & fit existing patterns. While this can be a huge help, it can also lead to a "homogenizing effect." One study found that when people used AI for brainstorming, their ideas became much more similar to each other, with far fewer unique concepts compared to human-only groups. The AI, in its effort to be helpful, might be steering us all towards the same creative conclusions.

This isn't to say GPT-5 isn't creative at all. OpenAI’s own documentation highlights its ability to handle complex literary forms & produce compelling writing. But for many users, the feeling is different. It’s less of a freewheeling brainstorm partner & more of a meticulous, if slightly uninspired, assistant.

It's Not Just the AI, It's Us Too

Here's a tough pill to swallow: the AI might not be the only one to blame. Our growing reliance on these tools can actually impact our own creativity. Research is starting to show that over-reliance on AI can lead to "cognitive offloading," where we essentially outsource our thinking. We become less practiced at making our own mental leaps & connections.

One study even showed that the more capable we believe an AI to be, the less effort we put into a task ourselves. It’s like we’re "falling asleep at the wheel," letting the AI do the driving without our critical input. This can lead to a decline in our own creative & analytical skills over time.

So, part of the problem might be that we're not engaging with the AI in a way that fosters creativity. We're asking it to do the work for us, rather than with us.

Restoring the Creative Spark: How to Make GPT-5 Your Creative Partner

Okay, so GPT-5 might be a bit more of a Spock than a Captain Kirk. That doesn't mean it's a lost cause for creative work. It just means we need to be smarter about how we use it. We need to be the creative director, not just a passive audience member.

Here are some practical strategies, from basic prompt engineering to more advanced techniques, to get those creative juices flowing again:

1. Master the Art of the Prompt

This is the absolute bedrock of getting good results from any AI. The quality of your output is directly tied to the quality of your input.

Be Specific & Add Context: Don't just say, "Write a story about a dragon." Instead, try: "Write a short story in the style of Neil Gaiman about a lonely, vegetarian dragon who runs a rare bookstore in modern-day London. The tone should be whimsical yet melancholic." See the difference? You're giving the AI a much richer starting point.
Assign a Persona: This is one of the most powerful tricks in the book. Tell the AI who it is. For example: "You are a world-weary detective in a noir film. Describe the city at night." or "You are a bubbly, enthusiastic marketing expert. Write a product description for a new brand of coffee." This helps the AI adopt a specific voice & style.
Provide Examples (Few-Shot Prompting): Show, don't just tell. If you want the AI to write in a particular style, give it a few examples. You could paste in a paragraph of your own writing & say, "Write the next paragraph in this style." This helps the model "learn" what you're looking for.

2. Think in Chains & Build on Conversations

Don't expect the perfect result from a single prompt. The real magic happens in the back-and-forth.

Chain-of-Thought Prompting: This is a fancy term for breaking down a complex task into smaller steps. Instead of asking for a full blog post in one go, start by saying, "Let's brainstorm some headlines for an article about the creative potential of AI." Then, "Okay, I like headline #3. Let's outline the main sections for an article with that title." Then, "Now, write the introduction for this article." This iterative process gives you more control & often leads to better, more nuanced results.
Iterate & Refine: Don't be afraid to push back on the AI. If you don't like something, tell it. "That's a good start, but can you make the tone more sarcastic?" or "I like the idea, but can you replace the fantasy elements with science fiction ones?" Treat it like a conversation with a collaborator.

3. Encourage "Outside the Box" Thinking

Since GPT-5 can be a bit linear, you sometimes need to explicitly tell it to be more creative.

Use Creative Power Words: Try incorporating words like "analyze," "critique," "compare/contrast," "metaphor," "analogy," or "unconventional" into your prompts. These can nudge the AI out of its default, descriptive mode & into a more analytical & creative space.
Embrace Constraints: Sometimes, the best way to be creative is to give yourself limitations. Ask the AI to write a poem with a specific rhyme scheme, a story using only dialogue, or an explanation of a complex topic using only simple words. Constraints force both you & the AI to find clever solutions.
The "Rubber Ducky" Method: Software developers have a technique where they explain a problem to a rubber duck. The act of articulating the problem often helps them find the solution. You can use GPT-5 in the same way. Just type out your creative problem, your half-formed ideas, & your frustrations. The process of explaining it can spark new insights, & the AI might just surprise you with a connection you hadn't considered.

4. The Human-in-the-Loop is CRUCIAL

Ultimately, AI is a tool, not a replacement for your own creativity. The best results come from a partnership.

You are the Curator: The AI might generate ten ideas. Your job is to pick the one with the most potential & develop it. The AI can be a firehose of content; you need to be the one with the taste & vision to shape it into something great.
Fill in the Gaps: Use AI to handle the grunt work so you can focus on the fun stuff. Let it generate a few different plot twists, & then you can write the one that truly excites you. Let it come up with a list of character names, & you can choose the one that feels right.

Where This New AI Paradigm Shines: The Rise of Specialized AI

While the general-purpose creativity of models like GPT-5 might be a subject of debate, this new focus on accuracy & reliability is a HUGE win for more specialized applications. Think about the world of business communication. You don't necessarily want a whimsical, storytelling AI handling your customer service inquiries. You want one that's fast, accurate, & helpful.

This is where platforms like Arsturn come in. Businesses are increasingly using AI to create custom chatbots trained on their own data. The goal isn't to write a sonnet; it's to provide instant, 24/7 support to website visitors. An AI that is less prone to hallucination & can stick to the facts is PERFECT for this. Arsturn helps businesses build these no-code AI chatbots that can answer specific questions about products, schedule appointments, & even help with lead generation. In this context, GPT-5's "intellectual flattening" is actually a feature, not a bug. It ensures the chatbot stays on-brand & provides reliable information.

By using a conversational AI platform like Arsturn, businesses can build meaningful connections with their audience through these personalized chatbots. It’s a practical, real-world application where the AI’s newfound discipline is a massive asset, helping to boost conversions & provide a seamless customer experience.

Final Thoughts

So, is GPT-5 less creative than GPT-4o? The answer is… it’s complicated. It seems to be less whimsical & more disciplined. For those of us who fell in love with the freewheeling, conversational nature of earlier models, it can feel like a step back.

But it’s not the end of the story. By understanding why this shift happened & by becoming more intentional with our prompts, we can still use these powerful tools to augment our own creativity. We need to move from being passive recipients of AI-generated content to active collaborators in the creative process. It requires more effort on our part, but the results can be just as, if not more, rewarding.

The key is to remember that AI is a partner, not a replacement. It can be a tireless brainstormer, a world-class researcher, & an endlessly patient editor. But the spark, the vision, the soul of the work… that still has to come from us.

Hope this was helpful. I'm curious to hear what you all think. Have you noticed a difference in GPT-5? What are your go-to tricks for getting the best creative work out of it? Let me know.