Item — HN Reader

I implemented a receipt scanner to Google Sheet using Gemini Flash.

The fact that it is ”intelligent" it's fine for some things.

For example I created structured output schema that had a field "currency" with the 3 letter format (USD, EUR...). So I scanned a receipt from some shop in Jakarta and it filled that field with IDR (Indonesian Rupiah). It inferred that data because of the city name on the receipt.

Would it be better for my use case that it would have returned no data for the currency field? Don't think so.

Note: if needed maybe I could have changed the prompt to not infer the currency when not explicitly listed on the receipt.

2 comments

> Would it be better for my use case that it would have returned no data for the currency field? Don't think so.

If there’s a decent chance it infers the wrong currency, potentially one where the value of each unit is a few units of scale larger or smaller than that of IDR, it might be better to not infer it.

I think most tools in this space do the "infer a bunch of data and show it to the user for confirmation", which lowers the pain of a miss here.

> Would it be better for my use case that it would have returned no data for the currency field?

Almost certainly yes.

Except in setups where you always check its work, and the effort from the 5% of the time you have to correct the currency is vastly outweighed due to effort saved from the other 95% of the time. Pretty common situation.

Even in those setups it's better to leave the currency field blank instead of hallucinating something.

You have the option to prompt it to do what you say. Of course it will not be 100% deterministic, but that's what evals are for.