and why it makes me confident that my future as a designer is going to be OK.

I’m a graphic designer as you might know, so of course I’ve had so many meetings where coworkers or friends ask me if I think my job will be replaced by AI. Sometimes it’s not even a conversation. People just say “you guys are so screwed.” I just shrug and say “we’ll see.”

So you might ask how I can be so non-chalant about all this. Until recently the best answer I could give is, “I just have a feeling that most people aren’t really that satisfied by what AI produces.” Generalists and know-nothings absolutely drool over what AI can do. If you aren’t that good at anything, then of course it’s better than you are. But if you look at any particular field, the skilled practitioners really aren’t that impressed. Call it AI Slop or Spaghetti Code, or the overuse of em-dashes, we can all spot the AI fingerprints in the things we really do well.

The state of AI design in practice today:

Show a designer something you just “AI’d” in Grok. Maybe you love it — but you probably aren’t a designer or you wouldn’t be in this situation in the first place. Now ask them to really critique it and listen to their input with an open mind the way you usually expect designers to do for things they spent hours or days making. You will get two possible responses:

1. The design is sound, but not very interesting. AI is good at the fundamentals when producing rudimentary, emotionally-vacant work.
2. There’s a concept but the execution is incredibly sloppy, not even at the level of a junior designer. AI tends to frankenstein complicated design jobs when left to itself rather than finding elegant solutions.

So that’s the state of AI-created design work when a designer isn’t actually behind the prompt. Of course the situation is different when someone who knows what they are doing is in charge of the AI, because then it’s a force-multiplier. You find that this dynamic holds up in every field. Decent writers write better than AI. Decent developers code better than AI. Solid executive assistants summarize notes and manage schedules with more intuition than AI. It’s only the bland dabblers who are blown away by what it can do with a one-click, gimme prompt.

Nobody plays chess anymore?

As it stands, if there is no human brain dictating what AI should produce, the results are always poorer than what a skilled human could do instead. However, until recently I couldn’t quite articulate why I felt that this was going to be a long-term competitive weakness for AI. That’s until I read a book called The Chess Revolution by Peter Doggers. His book on the history of chess provides a perfect corollary to the modern AI situation. Let’s dive in.

Dogger’s chapters on the development of computers that play chess traces the growing sentiment from the 1950s to the 1990s that eventually there would no longer be any point in playing chess because computers would eventually perfect the gameplay and render human input obsolete. However history shows that this didn’t happen. Chess — played imperfectly by humans against other humans — is at its all-time zenith today. That’s more than two decades after the last victory of a human over a computer in a public competition using standard time controls.

Moreover, Doggers describes a currently-existing cloud-based competitive chess league populated exclusively by AI chessbots that play each other at skill ratings that are hundreds, or even thousands, of ELO points higher than the highest-rated human players of all time. If the sentiment that dominated chess from the days of the first computer and which dominates every field now in relation to AI were true then by all metrics chess died in 2005 and nobody should be playing it.

But we still play chess. So what’s up?

Robot chess just isn’t chess.

Magnus Carlsen, the best chess player in the world today, and arguably of all time, described playing against AI as playing an idiot who always beats you. Basically, AI chess moves don’t make any sense to a human because every single move is a statistical calculation based on millions of down-stream possibilities every second. So those AI vs. AI matches happening in the cloud? Watching them doesn’t even look like you are watching chess.

In fact, when you compare chess move recommendations made by AI-driven chess engines they often look like blunders but the engine rates them much higher than the move that looks common-sense to you. The latter move is now called a “human move” in chess circles. Chess players now use AI models to validate and inform those moves when developing strategy, and this is the key point for all industries. Once AI got so good at chess that humans couldn’t beat it, we stopped looking at it as being a contender and treat it as a tool instead. The value of chess must therefore lie exclusively in humans playing chess.

We can actually derive a general axiom from the lesson of chess:

Specific axiom: The value of chess is always grounded in human-played chess and any non-human means of surpassing human chess can only have value in aiding human chess.

Replace chess with anything else, and you have a general axiom that applies to AI:

General Axiom: The value of X is always grounded in human-guided X and any non-human means of surpassing human-guided X can only have value in aiding human-guided X.

So what does this mean for someone who is a designer?

Robot design just isn’t design.

You might now be saying “this was just a roundabout way of saying that AI is just a tool” but the story of chess has really taken us further than that. It takes us beyond such an opinionated stance and provides some conclusive statements about the future general impact of AI:

First, it says that AI will not knock down the wall between those that have a skill and those that don’t. It isn’t just that Designers will use AI to design better than non-Designers. It’s that non-designers will never be able to assess the AI’s results with clarity and will always settle for second-rate work. Think of it like this: you can’t beat Magnus Carlsen at chess even if you have AI because any deviation from the AI’s moves and you’re sure to make a bad move. Your only hope of victory is not to be involved

Second, it proves that AI in the bigger picture is a lot less exciting than we think. It’s one thing for AI to dominate chess, which has fixed rules and closed conditions. It’s another thing for AI to dominate everything else. It will either plateau before surpassing human capabilities, effectively becoming the new Adobe suite, or it will blow past our capabilities to such an extent that it’s output is far too esoteric to be useful.

Third, it proves that the human connection is where the value lies, not supposed speed or efficiency. A world where every graphic, line of code, and every written word is dictated by AI without human guidance is like the world of AI chess. Everything will look idiotic and sloppy because it will be optimizing for conditions far too complicated to convey meaning to humans. AI, even at it’s absolute peak, is no replacement for humans.

Conclusion

Ultimately, we’re all going to have to contend with people being overawed with AI showing us the bland work it could do better than them, but worse than us. That’s OK. They are probably having the same experience in their turn — I’m certainly no great ad copy writer or marketing strategist but I bet ChatGPT can give me something I think is amazing. That’s just the phase we’re in right now. Chess already went through the whole cycle and humans came out on top in the bigger picture. Designers will too, and likely so will you in your field. After all, this game has been played before and we know all the moves.

Thanks for reading. If you enjoyed this post, subscribe to DesignRight for regular design-related topics.

‍