Hacker News Reader

You can't make this claim from pushing numbers around in an excel spreadsheet.

>As previously noted, the metrics from OpenAI and Anthropic are imperfect proxies for AI risk and usage, while still being the best available.

Seems they're just coming out and admitting they refuse to measure it themselves. Not a good sign.

View on HN

If it's the right numbers (called measurement data) and the right excel sheet pushing (called running a validated model) that is exactly the way you can make these claims. Overall it's called the scientific method.

What is your null-hypothesis and how does your data actually refute your null-hypothesis? And how is your sample representative?

Reply on news.ycombinator.com