Easy way to monetize Sims fans who will actively purchase all of the expansions. Sims 4 is particularly bad about it from what I recall because there's like 4 levels of DLC which range from the lower end themed stuff packs to like, $10 to implement laundry mechanics. At least back in previous Sims, those would generally just be implemented in larger expansions.

Right, the fact that ideally, the game itself would handle managing information would be the primary appeal over something like AI Dungeon or ChatGPT.

With that said though, AI Dungeon, Hidden Door, etc. primarily seem to be 'choose your own adventure' type deals where you can hash out the details to get the story flowing in a direction you want. It's probably easy enough to make a system that starts with broad story arcs (e.g. Act 1, Hero leaves X hometown, Act 2, Hero explores Y cave, etc.) and then hash out the details inside those specific chapters (e.g. Act 1.1 Hero says goodbye to Mom, Act 1.2 Hero travels through Windy Meadows and meets the Sage, Act 1.3 Hero and Sage find Y cave).

Even with that in mind however, I think at the moment, the scope would still be fairly limited if you're trying to do anything beyond a CYOA type of game. With just a generic fantasy RPG for example, unless you program in a large variety of assets, I could easily see a lot of games have a similar template like 'John the warrior left town, then met Jane the sage in the Dark Forest, then went to the Evil Castle and defeated the Demon King. The end' and it could easily get stale.

Assuming it happens eventually where the AI can generate its own assets, NPCs, etc. and the story lines are actually internally consistent, sure. The concern I have with an AI-generated game being released now is that even with the current technology, story telling via AI does have some limitations. To my knowledge, AI Dungeon itself still does require some babying to get it to do what you want it to since it only remembers the most recent parts of the story being generated; you have to update its memory bank with new information periodically, otherwise the AI produces weird results, gets caught in loops, etc.

Depends on your perspective, really. If you believe in the idea that game developers generally don't want to spend all their time and energy consciously releasing a product that they know is horrible, you end up mostly with a bunch of games (that were popular enough for reviewers to consider scoring for clicks) that score around 6 through 10.

When we're talking about products that take hours to process and are generally more expensive to consume than other mediums like movies and books, a 6/10 game is just gonna seem less appealing at a glance than one that's at least 8/10. This isn't to say that people should avoid the 6 or 7s as generally, that's also a score given to relatively niche games without broad audience appeal.