AI is a Tool, Not a Solution

In high school, I worked at our local Best Buy. This was during the company’s rise to the top — as the affordability of personal electronics, video games, and compact discs dovetailed with the nation’s financial good fortune — and that put Best Buy in an interesting position: it became the arbiter for consumer electronics trends.

This was the late 90s — the internet was present, but not commonplace. Mobile technology had not quite hit its tipping point, and we were a few years away from the ubiquity of online shopping, so the staff at Best Buy were often the most qualified people you knew when it came to technology. They helped you piece together your first computer. They helped you understand the correct layout for your surround sound speakers. They filtered through features and helped you make a reasoned decision.

They also tried to sell you stuff.

Shrouded in the cloak of expertise, Best Buy staff were trained to push high-margin accessories and long-term warranty plans — not because YOU needed them, but because THEY needed you to need them.

Every generation’s technology has its own version of accessories and warranties. Every advancement is a new opportunity to claim undiscovered land. Every futuristic dream is someone’s quick ticket to wealth and success. And right now, that dream is “artificial intelligence.”

A few weeks ago, I attended Designing with AI 2024, a conference about helping us understand where design and AI intersect. The conference program focused less on the endless promise of unfettered AI and more on the logical and practical use of AI to incrementally enhance our existing practices. After hearing how designers use large language models to help flesh out the opening stages of organizational design and how information architects lean on AI to quickly assess the taxonomy of a news feed, one main point stood out.

AI is not the solution.

AI is important, sure. It will change how we work. AI is a tool that can help provide shortcuts and a new perspective. But AI is not a singular solution on its own — and, given content and design’s need for human-driven context and nuance, we’re a very long way from it ever being a singular solution.

This may feel a bit like a warning. In reality, it’s a plea for level-headedness. The narrative around AI isn’t focused on small incremental updates and thoughtfully considered uses; instead, it’s focused on how AI will craft perfect user interfaces and replace the need for custom illustrators. That’s where things get dangerous — we’re too often trying to figure out how to shove AI into our process, to create use cases that can warrant the technology instead of finding ways to help integrate AI into our existing process.

AI works best when it is used for augmentation, rather than automation.

This probably comes as a blow to the “let’s do AI on everything” set because it removes the land grab. But remember that anyone pitching AI as a kind of solution in and of itself is doing so because there is money involved. Like those long-term warranty plans, someone’s trying to sell you something. Instead of throwing a thousand ideas at AI hoping one of them will be the world’s game changer, our focus on AI should swing more in the other direction: it should be measured and specific, beholden to a real use case.

If your first thought is “I wonder how AI can help us,” I’m here to say you are already walking down the wrong path. The real question should be “How can I improve this process,” with AI augmentation as one of many potential tools. Instead of wondering how AI will help you hire fewer people, or create meaningful communication out of thin air, think more about how large language models will help you pair critical human analysis with the added benefit of more data points.

AI will be a powerful ally — just as there were numerous situations where a specific accessory or warranty was incredibly valuable to the customers at Best Buy. But AI for AI’s sake? Ask all of those people forcing NFTs into every nook and cranny of the internet how that’s going these days.

More than anything, remember that user-centered design requires just that: users. Humans. People, with an understanding of the nuance of interaction. Give them the shortcuts they need to be faster and more productive. But don’t expect that there’s a solution that goes around them.