We are still in the early stages of AI’s evolution, and one clear sign of this is how much work users still have to do to make AI tools function effectively. As Jono Bacon, founder of Community Leadership Core, points out, even something as simple as choosing the right large language model (LLM) to run a query can be confusing for most people. Once you’ve selected the model, there’s still plenty of work left to fine-tune the results, and consistency is far from guaranteed. Current AI models require a lot of manual intervention to ensure the output is relevant or aligned with user expectations.
Despite these challenges, RedMonk co-founder James Governor is optimistic about AI’s future. Although the technology is currently in what he calls the “trough of disillusionment,” he believes this phase is part of the natural trajectory of any groundbreaking technology. Governor compares AI’s current journey to the lifecycle of other revolutionary technologies, which tend to experience phases of overhype, skepticism, and then eventual widespread adoption. Some developers are already moving past the disillusionment phase, diving into the practical applications of AI, but for many, it will take more time before the technology feels fully polished and universally accessible.
It has been clear for some time that AI will need time to mature. James Governor’s comment on the “kitschy” nature of much AI-generated art reflects this, as it highlights the ongoing limitations of AI in areas that require human intuition and creativity. Grady Booch, a well-known AI skeptic, argues that we often overestimate the reasoning abilities of AI systems. LLMs operate on statistical processes, whereas human thinking is much more nuanced, involving layers of understanding that machines simply can’t replicate. AI may be able to mimic human work, but it can never replicate the depth or uniqueness of human cognition.
That said, AI is still incredibly useful in many contexts. For instance, when I asked a group of friends to help solve a complex business issue, I input their responses into ChatGPT and asked for a summary. The result was impressively accurate and insightful. This shows that while AI can’t replace human creativity or reasoning, it excels in certain tasks, like summarizing information and handling repetitive work. In the world of software development, AI tools can be especially helpful without replacing the core of a developer’s job. GenAI can assist with tasks like generating boilerplate code or offering alternative solutions, freeing up developers to focus on more meaningful, creative problem-solving.