Nat’s Space

Everyone Is Button Mashing

If you have ever played a fighting game, you know you can get by with button mashing. Hitting as many inputs as possible, just to maybe get some of the hits in. It’s possible you might even win a match this way! However, anybody who dives deep into fighting games knows that it is very easy to outplay someone like this. They will spend the time learning the combos, figuring out the strategies and practicing what they learn. In a surprisingly small amount of time, they can very quickly beat a button masher in a match.

AI has started to seep into my workplace. I am surrounded by non-tech people. The leadership at the company are encouraging people to "play around" with AI, and we all were given access to Claude. The more I see people using AI outside of a programming context (which is a whole other can of worms), I am more convinced of the following statement:

Any productivity gains that people are getting by using AI could be matched or surpassed if they just learned how to use their tools effectively.

The software we use is more complicated than many give credit for. Most things people do are learned behavior from years of clicking around and fumbling through. Often, this is good enough. People are able to guess, and brute-force their way to getting what they want. It isn’t hard to type a document into Microsoft Word, but Word, just like almost every program, has a deep well of features that often go untouched by most users. As soon as the user’s needs expand to more than just “basic formatted text”, it can be hard to find the right way to complete their task. They don’t go digging into the user manual, instead they improvise. They accidentally create a ton of friction for themselves, as they push up against the limits of their knowledge of the tool.

AI is a tempting solution to this problem. Remove the friction by having the computer understand the needs with a request and in a matter of minutes and the job is done. Though, how often did the AI mess it up? How often did the AI do it almost right? Once the job is done by AI, it is unlikely that the user by themselves would be able to fix the output of the AI, because they don’t know how to use the tool deeply. They ask the AI again, and maybe it gets closer to what they want. Or maybe not. They spin around for a while until they get something close enough, but learn nothing. The next time they do the same thing, they do the same song and dance, wasting time and money going back and forth with the computer. Essentially, I see them as button mashing. The AI assistant will rarely do the same thing twice in the same way. Nobody using AI is learning how to use their tools.

If someone learns their tools, then they don’t have the problems of a button masher. They can go from idea to end result seamlessly, quickly and with minimal friction, because they put in the initial work of learning how their tools work, their functions, and their limits. Why would they need an AI assistant to button mash for them, when they already know the keyboard shortcuts and button combos to get the job done well?

My advice (if it’s worth anything) is to just learn a little more about the tools you use. Find the documentation and start reading it. Maybe the software you use at work is more capable than you think. This is, of course, larger than just using computers. Go read the owner’s manual for your car or microwave! It’s possible all the little quirks in your cars infotainment can be fixed with strange settings. You'll never know until you look and learn about it. I think in a world of button mashers, those who learn how to play their games will still come out on top.