AI Magic Is Out; AI Realism Is In

[10,000 feet] We don’t have an real AI intuition yet. 

I had a client who was obsessed with his hockey equipment website.

In the middle of every ideation, he’d say something like, “well, on my hockey equipment website, you can compare prices really easily.”

At the time, I’ll admit I was annoyed by this hockey website—how great could it have been?!

But I now realize that I was taking something for granted…

Over the last two decades, we collectively built a deep digital intuition. As we fiddled with our apps and programs, we developed a hands-on understanding of what the technology could do and, more importantly, how we could use it to solve user needs. We don’t have that real intuition for AI yet.

We basically spent the first year of the AI Era saying, “look at what ChatGPT can do!” over and over again. This created an allure that LLMs are deeply powerful magic bots. It also minimized our ability to critically examine their unique capabilities.

And this is understandable! We’ve been thrown into a new AI era; like newborn babies, we’re just trying to figure out what is going on here.

But, in Year 2 of the AI Era, we need to build a real intuition for what this technology can do. Otherwise, we’ll be paralyzed from shaping the next wave of AI applications.

[7,000 feet] There is a risk behind AI inhibition

I have a feeling that many designers and innovators are misinterpreting their lack of AI intuition as a lack of relevance with the technology.

It seems like there is a lot of 'self-sidelining' going on, where we understand the importance of this new technology, but we feel like we are missing authority and intuition…or like we need to wait longer to figure it out.

As a result, we’re letting the first wave of AI be driven by engineers and business goals alone.

There’s a real risk to this, detailed by Jakob Nielsen, because we need AI to be designed in a human-centered way.

Our role, whether we’re focused on innovation, UX, or UI, is to shape products and experiences in a way that delivers the most value to customers.

Without our helmsmanship, a company risks its foray into AI being a big womp-womp—or, more precisely, a waste of investment, energy, and organizational excitement.

[5,000 feet] 12 unique capabilities of GenAI 

ChatGPT seems like it can do everything, but it’s not really good at anything. By itself, it won’t fulfill enough of the requirements to actually solve a nuanced customer need.

The core question at the heart of an AI intuition is this: what customer needs can GenAI reliably solve, after a foundational model is tailored to a narrow use case?

To begin to answer this question, I summarized 12 unique capabilities of GenAI. These will evolve over time as LLMs become more powerful; however, this is the starting point that I use to ideate AI functionality with teams today. 

For example, Capability #1 is: ANSWER EXTRACTION

User need: Today's digital products bury data within files, pages, and tabs, turning the simple act of answering a question into a challenging search for a needle in a haystack. Imagine a user trying to locate their company's sick day policy hidden within a document on the company intranet. Or picture a benefits manager needing to reference the specifics of a patient's insurance policy.

Capability: GenAI can quickly search across data sources, accounts, and even products to retrieve accurate answers, all without the need for a single click.

[Ground]  A workshop to build AI intuition

You can review the capabilities mentioned earlier, and I'd like to offer a suggestion for taking the first step toward putting them into practice: an AI exploration workshop.

Before the workshop, we pinpoint 5 to 6 customer needs that will be our focus for solving with AI. It’s important to reinforce the idea that, even in the age of AI, we must still lead with the customer need…at the risk of creating valueless functionality. 

In the workshop, every participant is assigned 2 to 3 of the unique GenAI capabilities. These become the capabilities that they must learn, own, and love throughout the workshop.

We tackle one customer need at a time. Armed with their capabilities, each participant tries to solve the customer need—creating a real AI feature to show to the group.

For example, imagine we are trying to solve a sales/training need: ‘as a salesperson, I need more real-world training before I go out into the field.’ 

Here, I might use capability #3, roleplay to create a tool that enables sales people to have practice conversations with synthetic customers.

By the end of the workshop, our primary goal is to break through the inhibition that is paralyzing design/innovation teams right now.

What an adventure we’re on; you can feel a new version of the world emerging underneath our feet. This newsletter is a new endeavor for me, and I’d appreciate your help. 

If you’re an innovation leader, hungry for a GenAI strategy, you can schedule a call. 

• We offer a Experience 3.0 Sprint to envision your future, partner-based UX/CX, powered by AI.

• We also offer a 3-hour workshop where you can build your AI intuition

If you’re a designer, researcher, or practitioner, consider subscribing or sending this to someone who could benefit from this content. 

Have a human week,
Alex