The Challenge of AI: Too Much Work for the User
The fact that AI still requires significant effort from users is a sign of how early we are in its development. As Community Leadership Core founder Jono Bacon points out, even something as basic as choosing between large language models (LLMs) is a confusing task for most people. Once a model is selected, users still have to tweak, refine, and optimize to get meaningful results—and even then, consistency remains elusive. Today’s AI models are powerful, but they demand a level of technical expertise and manual intervention that limits their accessibility for everyday users.
Despite this complexity, AI hasn’t lost its momentum. When asked whether generative AI had already peaked, RedMonk co-founder James Governor dismissed the idea. Instead, he framed the current skepticism as part of a natural technology adoption cycle—one that moves from indifference to hype, then to disillusionment, and finally to widespread use. While some software developers are already embracing AI tools, others are still waiting for the technology to mature. Like other transformative innovations, AI needs time before it reaches its full potential.
AI’s limitations are particularly evident in creative tasks. While tools like Midjourney can generate visually stunning images, their outputs often lean toward kitsch, as Governor notes. Similarly, AI-generated writing lacks the depth and originality of human authors. As Grady Booch points out, AI doesn’t actually reason or think—it merely processes statistics. Unlike human cognition, which is rich with experience, intuition, and emotional depth, AI can only approximate creativity but never truly replicate it.
However, AI still excels at specific tasks. Summarization, for instance, is an area where AI shines. When given unstructured input—such as multiple opinions on a business problem—ChatGPT can quickly synthesize key takeaways with impressive accuracy. Similarly, AI-driven tools are proving invaluable to software developers, helping them generate boilerplate code, explore alternative implementations, and think through technical challenges. As Kelsey Hightower puts it, “writing code should be the last thing a developer does.” In this way, AI isn’t replacing developers but rather enhancing their ability to solve complex problems.