If You Care To Know

Menu

Creativity vs LLMS

The SaaS Boom and the ChatGPT API Shortcut

In the past few years, we’ve witnessed an explosion of SaaS products that are little more than a thin wrapper around OpenAI’s API. A whole new generation of developers and founders are churning out AI-based products that, at their core, are little more than API calls to ChatGPT with some spin—maybe a different interface, a niche prompt engineering trick, or some lightweight automation added on top. And while this has led to a proliferation of AI-based tools, many of them have a strangely familiar feel, which has led to an oversaturation of essentially identical tools.

It’s easy to see why. Large Language Models (LLMs) like GPT-4 offer powerful text generation capabilities at a low barrier to entry. Anyone with basic coding knowledge can integrate OpenAI’s API and create a product overnight. But this shortcut comes at a cost: a lack of true differentiation and originality. Instead of creating innovative solutions, several companies are simply re-skinning ChatGPT and rushing to market in an attempt to grab a share of the AI gold rush.

This overreliance on LLMs to generate product ideas has resulted in a homogenized space where we’re seeing the same AI writing assistants, chatbots, and summarization tools over and over. If you’ve tried one, you’ve tried them all. The problem lies not just in the implementation but in the very creativity—or lack thereof—that LLMs have to provide.

LLMs: The Ultimate Brainstorming Partner or a Creativity Killer?

LLMs have an unprecedented ability to generate ideas at scale. Give them a prompt, and they can produce thousands of ideas in seconds, many more than a human could by hand. This can be incredibly useful, especially for overcoming creative block or rapid prototyping. But there’s a deeper, more unsettling question: What does it do to our own creative minds when we rely too much on LLMs?

Divergent thinking—the ability to generate many, novel ideas—is one such crucial element of creativity. Traditionally, this involves attempting on different perspectives, making unusual connections, and engaging in deep reflection. That being said, when LLMs become the default option for ideas, individuals might start depending on them instead of developing problem-solving skills themselves.

In case you have used an LLM to generate ideas, you might notice that it comes up with good, albeit predictable, answers. While it can generate hundreds of blog topic ideas, business ideas, or product names, many of them are formulaic. This is because LLMs have been trained on enormous amounts of existing data—they reframe and recreate what’s already out there rather than developing genuinely new ideas. As a result, over-reliance on LLM-generated ideas might discourage people from entering the messy, non-linear process of true creative discovery.

Why LLMs Aren't That Creative

The creative illusion in LLMs arises from the fact that they are capable of producing plausible answers that appear to be novel. True creativity, however, tends to encompass unpredictability, personal intuition, and disregarding set conventions—areas where LLMs falter.

  1. LLMs Lack Subjective Experience
    • Creativity is inseparable from human experience, emotion, and perspective. LLMs do not have lived experience, personal biases, or emotions—they predict text based on statistical likelihood. The poet’s personal anguish, the entrepreneurial dream, or the musician’s intuition are all nonexistent in LLM-generated content. They can mimic creativity, but they do not feel it.
  2. LLMs Default to Averageness
    • Because they’re trained on huge datasets, LLMs produce outputs that are the statistical average of what they’ve encountered. In other words, their responses will follow existing norms rather than challenging them. True creativity often involves contravening conventions—think of Picasso’s abstract art or Steve Jobs’ minimalist design ethos. LLMs, on the other hand, are designed to confirm patterns, not shatter them.
  3. LLMs Lack True Originality
    • If you ask an LLM to generate 10 startup ideas, it might generate good but also obvious ones—AI-powered content platforms, automation firms, or marketplace apps. Yet, revolutionary ideas, the kind that reshape industries, are typically a byproduct of humans making non-obvious connections between seemingly disparate disciplines. The most disruptive ideas often start out as outliers, something that an LLM will not be prone to proposing.
  4. The Echo Chamber Effect
    • As more people use LLMs to generate ideas, they are increasingly training their creativity on AI-generated content. This creates a loop where AI-influenced thinking feeds back into AI, powering sameness. The more people use LLM-generated ideas without injecting personal insight, the more homogeneous our innovative outputs become.
How to Cultivate Creativity in the Age of LLMs

While LLMs can be useful tools, we must ensure that they do not stifle original thought. Here are some ways to maintain and enhance human creativity while utilizing AI:

  1. Use LLMs as a Starting Point, Not a Crutch
    • Instead of going with the first AI-generated idea, make it a launchpad. Challenge the suggestions, combine some of the concepts, and add your own ideas. Treat an LLM like an assistant, not a substitute for creativity.
  2. Exercise Analog Creativity
    • Some of the best concepts are thought of when you are not in front of screens. Diary, read non-domain books, draw, or have deep conversations. Inspiration generally comes from unexpected quarters, and overuse of LLMs lowers the exposure to really new thinking.
  3. Practice First-Principles Thinking
    • Instead of relying on an LLM to generate ideas, start by breaking a problem down to its fundamentals. What are you trying to solve? What assumptions can you falsify? By thinking from first principles rather than relying on AI-generated output, you’ll be more likely to develop creative solutions.
  4. Combine AI with Human Experience
    • Some of the most beneficial applications of AI are in human-AI collaboration. Instead of letting an LLM dictate creative direction, use it as a tool to enhance human capabilities. In writing, design, or business strategy, inject personal experiences, intuition, and insights that AI can’t replicate.
  5. Encourage Open-Ended Exploration
    • A lot of creativity comes from following tangents and unexpected directions. Don’t simply say to an LLM “Give me 10 startup ideas,” for instance. Pose deeper philosophical questions. Push into related fields, merge fields, and push beyond the usual prompts.
  6. Create, Don’t Just Curate
    • One of the largest dangers of LLMs is that they encourage passive consumption instead of active creation. Instead of merely using AI to summarize, rewrite, or generate content, focus on original thought. Experiment, prototype, and build things that go beyond what AI can create.
Conclusion

LLMs are powerful, but they are no substitute for novel creativity. The SaaS sector’s inundation of ChatGPT-powered tools highlights the detriments of excessive dependence on AI-powered solutions. Although LLMs possess the potential to scale concepts, they end up reinforcing existing trends rather than exploring new frontiers.

In order to preserve human creativity, we must be mindful of the application of AI. By utilizing LLMs as tools and not replacements, by maintaining independent thinking, and by embracing the messiness of real creativity, we can allow our ideas to remain innovative and productive.

The future of creativity isn’t in outsourcing it to AI but in finding a means of getting AI to collaborate with human ingenuity—without compromising what it is that makes creativity truly remarkable.

AI Ethics 1 Climate 0 Data Science 0 Machine Learning 2 Projects 1 Software Engineering 0 Uncategorized 0