We revisit our 2025 Southern California sales tax forecast a year later and find that an honest range beat a confident guess (and that averaging our two models would have beaten either one alone).
Most forecasts include a range, but not all ranges answer the same question. This post breaks down the difference between confidence and prediction intervals, and why it matters for real-world city decisions.
If your forecast is just one number, it’s probably telling an incomplete story. Here’s how prediction intervals add the context city leaders actually need to plan ahead.
A vague prompt gets unpredictable results. A specific one gets reliable results. Here's how to write prompts that actually work, using techniques you already know from managing people.
LLM hallucination is manageable, but not by waiting for smarter models. The real solution is giving the model less room to guess. Here are four techniques that make AI outputs dramatically more reliable.
When ChatGPT responds to your prompt, it's not retrieving an answer from a database. It's predicting the next word, one token at a time, based on probability.
