What is Creativity?

As always, opinions are my own, not those of Lichess.org.

February 2025 Joe Rogan interviewed Magnus Carlsen (thanks GoMagic). I summarize:

Carlsen: I find it interesting to read what people were saying about chess computers 50-60 years ago, when people were having an actual discussion about whether computers could defeat a grandmaster.
Rogan: People were having the same discussion about go, right?
Carlsen: Go is much more [complex] than chess.
Rogan: Go engines created moves which top-level players are playing today.

Rogan's interviewing style aside, there are a couple tractable topics here (tractable: challenging, but possible):

1. Is it fair to compare what people said about chess computers and go computers?
2. How creative are chess and go engines?

I read about chess computers over the decades, and about machine learning in recent years. But before arguing about theory, let's list practical creative uses of technology:

• Pythagorean blacksmith hammers (striking tuned metals against each other to generate chords)
• Aleatoric music (throwing dice to randomly generate music)
• Astrology (tarot cards etc.)
• Monte Carlo simulations, MECC educational simulation software (The Oregon Trail etc.)
• GM Browne vs. Belle (tablebase) queen vs. rook exhibition
• Databases (spelling autocorrect, ChessBase, Google autosuggest)
• Deep Blue rematch (positional sacrifce), AlphaGo, AlphaZero, etc.
• Automated computer analysis
• Leonhardt Gambit Song (I don't know how much of this is computer-generated)

Over time, what happened? We found more complex ways to use tools to solve challenging problems, to a point where absent careful scrutiny people may imagine that technology seems indistinguishable from magic.
When people realize that technology follows a set of natural laws, the illusion that technology runs on magic is broken. The video game The Talos Principle asks the player whether value is created, or value is realized. Let's avoid that philosophical debate, since technology can't solve problems on its own: technology needs us to creatively make use of it in order to solve real-world problems.

As for the real-world problem of teaching players to improve, The Perpetual Chess Podcast guests have opined about this, including GM Friedel noting that this is a challenging research area. It's difficult to distill conventional wisdom into natural language, both because said wisdom is in the minds of masters (not in the memories of computers) and because natural language presents its own challenges.

Upon reflection, I am a bit surprised that Google DeepMind / Scholar haven't yet invented a new way to teach the royal game, since there is such an abundance of literature and media at their disposal. On the other hand, what profit can be generated from teaching (other than learning how to teach and applying that to other domains)?

I had intended to research what people said about go computers and about chess computers, but I ran out of time so I'll leave that as an exercise for the reader.

EDIT 2026-02-28: Naviary documented some of the above technological developments:
https://youtu.be/vpE7u6ya1k8

---

Image credit: ameenfahmy