Miyamoto’s AI Lesson: Good Tools Still Need Good Ideas

I love retro games. That probably explains why Shigeru Miyamoto’s thoughts on artificial intelligence caught my attention more than another promise to produce a hundred times more content. A game can stay enjoyable long after its graphics stop looking impressive. Getting more material onto a screen is only part of the job.

Shigeru Miyamoto speaking on stage at the 2007 Game Developers Conference.
Shigeru Miyamoto at GDC in 2007. This archival photograph does not show the 2026 press conference. Photo: Vincent Diamante, CC BY-SA 2.0.

Miyamoto’s recent remarks put that distinction into words: better tools can take over production tasks, while the intention behind a piece of work still matters. I find that a useful way to think about AI, both as someone who enjoys games and as someone who writes this blog.

What Miyamoto actually said

In the full press conference published by Nintendo DREAM WEB on October 23, Miyamoto compared generative AI with earlier changes in creative tools. Computers and word processors made some technical tasks easier, allowing individual judgement and design ability to become more visible.

He also described a trap: enjoying the process of doing the work can make us confuse effort with quality. If AI handles more of those tasks, he sees an opportunity to think harder about what creating something really means. His comments are a personal view of creative work, rather than a new Nintendo product announcement or a guarantee about anyone’s job.

I care about what a game gives me

When I pick up my Anbernic while travelling, I am looking for a game I want to spend time with. I do not need its appeal to be explained by the number of assets, the size of the team or the latest graphics technology. A clear idea, satisfying controls and a reason to keep playing can go a very long way.

That is where Miyamoto’s argument makes sense to me. Generating another level is useful only if that level belongs in the game. Generating more dialogue helps only if the characters have something worth saying. The difficult decision may be choosing what to remove.

The same question applies to my blog

For Routes, Bytes & Budgets, a longer article is not automatically more helpful. I want a reader to leave knowing whether a device fits their needs, which detail needs checking, or whether keeping what they already own is the sensible option.

AI can help with the work around that goal. But a polished paragraph still needs a sound argument, accurate sources and a reason to exist. My own experience also needs to remain mine: an attractive description cannot substitute for something I actually tried.

This connects with my earlier look at AI video editing and the time spent on setup and corrections. Speed becomes valuable when the finished result holds up to review.

The part I want to keep practising

I like tools that remove tedious work. I also want to keep practising the decisions that make the result useful: noticing when something feels wrong, explaining why, and knowing when to start again.

Miyamoto’s perspective leaves room for optimism without asking me to admire everything generated by a model. My test stays pleasantly ordinary: did this make something worth reading, playing or using? If the answer is yes, the tool has helped.

Comments

Popular posts from this blog

Obsidian’s New Search: Less Friction for Notes in Two Languages

OPPO Find X9 Ultra vs Find X10 Pro Max: Is the Older Ultra Still Worth Buying?

OPPO Find X10 Technical Preview: Cameras, Design and Missing Specs