> Nearest X place" is a task for map apps, not broad Web search.
Why? I have a big search bar and I dump everything into it. That’s the reality of how I search the web.
Every search engine worth its weight has an integrated mapping feature and it will punt you to a map for certain queries.
I tried chatGPT for restaurants and it made up places. I personally don’t need it to be a concierge telling me about places appropriate for certain occasions (although that would be useful to many i admit) when I can’t trust the results anyways. Fool me once… and I fall back to Google because I don’t have patience to be fooled twice.
> restaurant reviews in general are problematic.
Not related to chatGPT but I think everyone gets this wrong. Yelp et al are good for auditing a restaurant selection, but are terrible for searching. I find that those listicles from major publications that list “Best X in $Location” are great for discovering if you don’t want to walk around and actually look at storefronts. So I think there is trustworthy (enough) review sources, but the ML models can’t know that.
Why? I have a big search bar and I dump everything into it. That’s the reality of how I search the web.
Every search engine worth its weight has an integrated mapping feature and it will punt you to a map for certain queries.
I tried chatGPT for restaurants and it made up places. I personally don’t need it to be a concierge telling me about places appropriate for certain occasions (although that would be useful to many i admit) when I can’t trust the results anyways. Fool me once… and I fall back to Google because I don’t have patience to be fooled twice.
> restaurant reviews in general are problematic.
Not related to chatGPT but I think everyone gets this wrong. Yelp et al are good for auditing a restaurant selection, but are terrible for searching. I find that those listicles from major publications that list “Best X in $Location” are great for discovering if you don’t want to walk around and actually look at storefronts. So I think there is trustworthy (enough) review sources, but the ML models can’t know that.