Design, Fast and Slow

**If you haven't read Thinking: Fast & Slow by Daniel Khaneman, you should. It helps you understand your own mind. But you can still get the benefit from this post if you haven't.

Are you singing AI praises, Hanan?

“The easiest way to increase happiness is to control your use of time.”

No. I know everyone is sick of this AI did this…that AI did that. I’m going to make my case on why slow & fast design are both valid an acceptable.

If you’ve been online anywhere in the tech scene these past 12 months you must have noticed the growing tug-of-war between designers advocating for new AI assisted procceses, and designers trashing every new AI tool. This war won’t end ebcasue both are perfectly proven methods of deisgn. The key is strategy.

Slow Design: reliable & expensive.

The nervous system consumes more glucose than most other parts of the body, and effortful mental activity appears to be especially expensive in the currency of glucose.

Traditional design is effortful and therefore expensive. Yest it is the one that catches what fast design does not. The outliers and things not found witout deep thinking.

Example: User Interviews
You recruit participants for research. You write the questions, and if you're doing it properly, you interrogate your own questions for bias first. You moderate the sessions in real time, reading body language. You study the video or transcript slowly until you find the golden bits.

This process is slow but the slowness is what catches the gold: the quiet contradiction between what a user says and what they do, the outlier that turns out to matter more than the majority.

Slow design is expensive because it's thorough. Like a Rolex made by hand vs. a less expensive watch brand manufactured in masse.

And it needs resources that not every project has. Time. Budget. Access to real users. A team that isn't already stretched across four other things.

Fast or AI assisted design: tempting & cheap. Good in combo with slow.

"Loss aversion is a powerful conservative force that favors minimal changes from the status quo”

AI-assisted research doesn't replace manual UX research. It attempts to predict the market’s response to your idea by finding patterns across volumes of data no human could read in the same timeframe. The strength is speed.

A human team can do just the same, given the time. Even better. But that is exactly the point: when a project doesn't have six weeks for a research phase, when you have to deliver in one sprint, not a quarter, the ability to get a fast, pattern-level read on user sentiment is the only research of value you can provide. A different tool for a different constraint.

The Caveat: beautiful, wrong things. Confidently & quickly.

“System 1 is radically insensitive to both the quality and the quantity of the information that gives rise to impressions and intuitions.”

But here is where Kahneman's caution matters most: AI in design is confident even when it's wrong. It doesn't pause to check itself. It builds a coherent story from whatever's in front of it and moves on, and that story feels true whether or not it actually is.

An AI tool will design you a flow, a screen, a prototype, based entirely on the prompt you gave it, the data you fed it, the patterns in its training. It will not tell you what it doesn't know about your specific users, your edge cases, your specific market. It cannot see what it hasn't been shown, and it will not pause to wonder about that gap the way a good researcher instinctively does.

The speed is real. The need for a slow check on top of it is also real.
Fast design without a slow check is how you end up building the wrong thing, quickly.

Therefore I admit to using both, the resources decide which.

I don't think traditional design is more legitimate than AI-assisted design. I don't think AI-assisted design is the future that makes the traditional process obsolete. I think that's the wrong thought process entirely.

The actual question, every time, is: what does this project have room for?

• Quick projects & low budget: AI base in 1-2 days, rest of the sprint I take over with slow deisgn.

• POCs: unpaid work, especially cold emails are the place AI can do the most benefit. Though never deliver what it spits out first. It is always a draft.

• Big projects with reasonable budget: Slow, manual design. AI is only used for ideation and planning. Clients get what they pay for.

What about you? what do you think?

——

'Till next time, lots of love and #keepdesigning.
Disclaimer: This post references ideas and quotes from Any direct quotations are kept brief and attributed; the framework and its application to design practice are my own reading of the book's ideas. The opinions about design process are entirely mine. as are my Figma files.