05.12.2024

You're Using AI Wrong - And That's Why You're Frustrated

Tobias C. Kaechele
By Tobias C. Kaechele
Robot

The tech world is experiencing what feels like an AI hangover: After the initial euphoria, company shares are dropping, users are getting frustrated, and there's a growing push for heavy regulation and even outright bans on AI-generated content. But here's the thing: The problem isn't with AI tools - it's with how we're using them.

Let me share a recent experience: My partner needed a complete brand overhaul - new name, logo, and slogan. Instead of spending weeks (and a lot of money) creating something from scratch or paying various agencies, we sat down with our laptops and started ChatGPT & Co. But here's the crucial part: we used them as accelerators, not substitutes for human creativity and decision-making.

Understanding What “AI” Really Is

First, let's clear up a common misconception. Terms like "Artificial Intelligence" are misleading. What we're really working with are probabilistic “models”, sophisticated and trained algorithms with a defined input and output. In the case of large language models, you give it a prompt (which will be encoded numerically) as an input and it will spit out the next “word” with the highest probability (encoded numerically) as an output. There's no actual thinking process, no real understanding, and definitely no "ratio" behind their outputs. They're incredibly advanced machine learning algorithms, yes, but they're still just tools - powerful hammers looking for nails.

ChatGPT, Claude, Midjourney, DALL-E, and so on aren't magic boxes that produce perfect results at the push of a button. They're more like highly skilled brainstorming partners that process information in the form of numbers and probabilities.

How to Actually Use AI Tools

Our little project is a perfect example of effective AI tool usage. We spent about an hour going back and forth with various AI tools. The process began with using language models to generate dozens of name ideas, continuously providing feedback to the tools and reiterating over the drafts. We used the same process for the various logo concepts using image generation tools like DALL-E3, Midjourney, or Flux. Finally, we returned to large language models to brainstorm and refine slogan options.

The result? A beautiful logo draft, a powerful slogan, and an amazing new name. We immediately secured the domain! But here's the thing: none of these came from a single prompt. Each was the result of multiple iterations, careful refinement, and human ingenuity!

The Professional Perspective

As someone who manages teams and regularly interviews candidates for various IT-related roles, I'm often asked about my stance on AI tools in the workplace. My response? “If you're not using AI tools for your work, you're actually wasting company time and money!”

AI is here to stay and this means for you and your employees that you need to know which tools are available, what you can achieve with them, and how to use them effectively. Perhaps most importantly, you must develop the ability to seamlessly integrate them with human expertise and creativity, creating something faster and greater than either could alone. These tools excel at generating initial ideas and concepts, helping break through creative blocks, speeding up repetitive tasks, and providing different perspectives and approaches. However, they still need us humans to produce truly valuable results.

ChatGPT & Co won't craft you a masterpiece - they're amplifiers of your craft, but like any hammer, their potential lies in how skillfully you wield them!