Against Magic — Introduction

Enchanted by AI

There is a magical sparkle that’s surrounding AI. This is the introduction to a mini series on concepts and mental models when working with Generative AI.

Katharina Köth

Follow

Published in

Creative Complexity

4 min readJan 14, 2024

--

At a magic performance, you can be either of the following: You can be the magician who enchants the audience. Or you can part of the enchanted audience, the one who’s curiosity is sparked and wants to figure out how the trick works or the cynic who knows how the trick works and might even spoil it for the others afterwards.

This series is written by the cynic for the curious.

When it comes to Generative AI, magic is everywhere. On Microsoft, on Google, on Figma, on Miro, on Linkedin — and of course ChatGPT; on the largest platforms you can see magical wands and sparks to indicate that some output will be created by the hands of AI.

Hide and seek: Search for the sparkle and you’ll find the AI.

Within just one year, magical sparkles became the de facto iconography for Generative AI. (Thanks to Johannes Klingebiel for being one of the first people to point it out.)

But of course tech is not new to obfuscating infrastructure behind feel-good metaphors. The “cloud” is made of vast over- and underground data centers where our information is stored, and deep-sea cables that connect continents.

UX and UI courses talk about creating “intuitive design” that minimizes cognitive effort and friction in using digital products by adapting real world interaction patterns (think: the nostalgic skeuomorph UI trend that shaped the first wave of mobile app design).

And while these metaphors and approaches made technology more accessible to the majority of people, I’d argue it also made them more ignorant about the nature of how things work.

On a side note, this is not just some hunch. First analysis show that Gen Z — grown up in walled Apple and Android product ecosystems and familiar with “intuitive” UI app environments — lack basic tech knowledge. (To their defense, they’re willing to learn and adapt.)

But it’s fine. Not everyone needs to understand.

Launching this series was motivated by observing too many people (on LinkedIn) that I consider being tech-savy, who — imho — should and need to understand how Generative AI works, being either willfully ignorant, grifting or paying into storyline of magical emergence of a new intelligence.

They recommend doing user research via ChatGPT (tbh, this one triggered me the most), building your business case with ChatGPT, developing your app ideas with ChatGPT. Way too often I find myself baited to correcting misreferenced (public) AI-research papers; or false explanations of the technology.

Coming back to the magic of Generative AI, I’d describe their approach with the words of the Wizard of Oz: “Pay no attention to that man behind the curtain.”

But I’m inviting you to do so. Be more Dorothy, and I promise you that understanding the basic ideas and concepts behind Generative AI will make you more self-sufficient and confident in evaluating more or less valuable, scalable and adaptable applications of the technology.

Disclaimer about myself

While I’m not a trained machine learning engineer, I spent the greatest chunk of my spare time for the past 4 years with actively learning and applying machine learning and data science from scratch. First and foremost, with 15 years of professional experience, I consider myself a UX architect and strategist.
Due to my expertise, I will focus on creative, innovation, design and media industries. There are many more industries and applications of Generative AI, but I know too little about them.

And now: Please follow me to Part One: AI as a Strong Signals Machine

Against Magic — Part One: AI as a Strong Signals Machine

A guide to Generative AI for non-machine learning people

medium.com

https://media.giphy.com/media/KeQ3dG4POE37DNAtVm/giphy.gif