The proper comparison is not a human, it’s a computer. Or even a human with a computer. But my point wasn’t to judge LLMs on their (in)ability to do math - I w

The proper comparison is not a human, it’s a computer. Or even a human with a computer. But my point wasn’t to judge LLMs on their (in)ability to do math - I was only responding to the parent comment’s assertion that they’ve gotten better in this area. It’s worth noting that all of the major models still randomly decide to ignore schemas and tool calls, so even that is not a guarantee.