The Irony of AI (at least for now)
After decades of obliterating our ability to focus, technology has a powerful new tool, LLMs — but unlocking their value requires the one skill we destroyed: attention.
Benedict Evans recently shared a post highlighting his perception that LLMs are in an early stage of technology development in which the usefulness is not quite there. Basically, he’s saying LLMs are overhyped as relevant to the modern professional as they stand right now.
I think he may be missing something.
I acknowledge that for some, the fall off or lack of usefulness is indeed the direct result of overhype. I believe that much of the real value to be unlocked for many real world use cases will come later with the advancement of AI agents/assistants and whatever else comes next. But I don’t think that what many people are experiencing as “this is wonderful, but I don’t have a use case” is actually a lack of usefulness or relevance.
Instead, I think it’s a lack of skill. Rather than attributing the AI usage falloff to a lack of meaningful use cases, we should instead be looking to the fact that many, if not most, users lack a strength in a core skill necessary to gain the full value and benefits of AI tools like ChatGPT, Claude and Gemini in the first place.
Without this critical skill people are only scratching the surface of what AI tools can do, never reaching the deeper levels of orchestration that can augment a wide variety of work and knowledge tasks.
These people are stuck in a cycle of first draft results. Never going further than a follow up or two. Never layering in their example outputs, custom data and contextual information. And as a result they are left unimpressed and progressively disengaged.
The missing skills? Paying attention.
AI has unlocked a new level in the game of life—but the ticket for entry is a skill that most of us have actively destroyed for far too long: the ability to pay attention.
I’ve been working with generative AI tools in one form or another since 2021 when I first began using VQGANS+CLIP image generation notebooks in Google’s Colab to create abstract AI artworks. Since then I’ve moved into designing and developing AI applications with the latest LLM APIs. All in, I’ve spent months worth of time working hands on with LLM technology.
And there has been one consistent learning:
The key to unlocking the power and usefulness of AI lies in the ability to give your full and undivided attention to AI-aided tasks, for extended periods of time—working back and forth to refine and develop the desired outputs. Without this kind of deep work and collaboration, you simply can’t get great results with today’s AI models.
The irony of course, lies in the fact that we in modern society have spent the last decade or more absolutely obliterating our attention spans through the use of mobile technology and in particular social media. This puts many, if not all, of us at a disadvantage.
Like a computer or calculator, modern LLMs are a bicycle for the mind. When used correctly they exponentially accelerate productivity. When used incorrectly, they cause confusion, distraction and frustration.
Without attention and dedicated use, AI is just another useless tool.
The race to gain an edge with AI is underway and those best positioned to leverage its benefits will be those who can overcome outside distractions and dedicate themselves to engaging with their AI collaborators.