I keep seeing people talk about using generative models with very low accuracy and success rates. I have to resist the urge to point out that with rates that low, it’s possible they’re just seeing randomness in action and the feeling of productive utility is down to their own perception