Nathan Curtis

Weighing "I should have a clue of [common type] in this cryptic and this is one of the only entries it works for" versus "the clue of [other type] I already have is clearly better"

New variety cryptic "This Old House" now available for free at www.patreon.com/posts/160568...!

Here's a free Pathfinder to start off the month: www.patreon.com/posts/free-p...

I plan to make a more fleshed-out argument in another medium, but I wanted to put together some thoughts on the use of generative AI in puzzles. Tl;dr: don't. Apart from everything else about GenAI, it is fundamentally opposed to the whole reason for solving and writing puzzles.

Constructing puzzles is also an exercise in sharing your values. What kind of insights do you value? What kind of vocabulary do you find interesting? Even the wording of a clue says something about your values as a constructor. AI has no values, and no voice.

I plan to make a more fleshed-out argument in another medium, but I wanted to put together some thoughts on the use of generative AI in puzzles. Tl;dr: don't. Apart from everything else about GenAI, it is fundamentally opposed to the whole reason for solving and writing puzzles.

The same goes for using AI to help write puzzles. Writing a puzzle is as much a process of discovery as solving a puzzle, even more so most of the time.

I plan to make a more fleshed-out argument in another medium, but I wanted to put together some thoughts on the use of generative AI in puzzles. Tl;dr: don't. Apart from everything else about GenAI, it is fundamentally opposed to the whole reason for solving and writing puzzles.

I am not opposed to using other tools to help you with the thinking and working out. If you're looking up answers to crossword clues, it still takes thought to figure out what synonym to use, or spot a misleading definition, or turn an oblique trivia reference into something searchable.

I plan to make a more fleshed-out argument in another medium, but I wanted to put together some thoughts on the use of generative AI in puzzles. Tl;dr: don't. Apart from everything else about GenAI, it is fundamentally opposed to the whole reason for solving and writing puzzles.

To enjoy puzzles is to relish the flashes of insight, made all the sweeter by the effort it takes to get to those insights. Using AI to solve puzzles, however, is by is very nature ceding the process of insight and discovery.

I plan to make a more fleshed-out argument in another medium, but I wanted to put together some thoughts on the use of generative AI in puzzles. Tl;dr: don't. Apart from everything else about GenAI, it is fundamentally opposed to the whole reason for solving and writing puzzles.

Puzzles are, at heart, toy problems that we solve because it feels good to use our brains to overcome obstacles. Everything else is just theory.

I plan to make a more fleshed-out argument in another medium, but I wanted to put together some thoughts on the use of generative AI in puzzles. Tl;dr: don't. Apart from everything else about GenAI, it is fundamentally opposed to the whole reason for solving and writing puzzles.

There's also the issue that using AI makes you bad at the thing you are doing. It makes up fake answers to basic factual queries, and it has been demonstrated that people who rely on AI for a task get markedly worse at doing the task themselves in a matter of months.

I plan to make a more fleshed-out argument in another medium, but I wanted to put together some thoughts on the use of generative AI in puzzles. Tl;dr: don't. Apart from everything else about GenAI, it is fundamentally opposed to the whole reason for solving and writing puzzles.

So, for starters, if you are considering using GenAI tools to help solve or create puzzles, you have clearly decided not to care about issues like the large-scale theft of intellectual property and the massive environmental damage done in the name of AI.

I plan to make a more fleshed-out argument in another medium, but I wanted to put together some thoughts on the use of generative AI in puzzles. Tl;dr: don't. Apart from everything else about GenAI, it is fundamentally opposed to the whole reason for solving and writing puzzles.

(How do I know it would have taken <5 minutes? It took me <10 minutes, and I am not an experienced coder.)

I plan to make a more fleshed-out argument in another medium, but I wanted to put together some thoughts on the use of generative AI in puzzles. Tl;dr: don't. Apart from everything else about GenAI, it is fundamentally opposed to the whole reason for solving and writing puzzles.

Spent an hour issuing over 100 refunds, 1 at a time, because Patreon messed up my billing and their support was not helpful. Helpful tip if you're a creator on Patreon: if you want to pause billing for a month, do not pause on the first of the month.

I occasionally get emails from a music store I frequent about new trombone-related items or sales. The fact that the latest one includes both trombone cases and wine cases reminds me how much of an outlier I am among trombonists.

One time in college we were playing a piece that had a bass trombone solo and the conductor wanted me to play more loosely. "Have you ever gone out and had a little too much to drink?" "No" Everyone in the brass section turned to stare at me like I had two heads.

I reeeealllly don't want to write clues today but I reeeeealllly gotta.