←back to thread

3 points scal | 2 comments | | HN request time: 0.399s | source
Show context
ibdf ◴[] No.45026464[source]
"To plan your trip to Rome, we first need a city in Italy." Eh?
replies(1): >>45026533 #
ibdf ◴[] No.45026533[source]
"The trip duration should be 10 days or less." What? Why?
replies(1): >>45027297 #
scal ◴[] No.45027297[source]
The trip you were trying to plan might be a little confusing . I’m currently fixing the few bugs encountered. Please note that this is a beta version and, for now, trips are limited to 10 days.
replies(1): >>45027682 #
1. ibdf ◴[] No.45027682[source]
Exposing my search here is probably not the best call, but that's fine since I was trying to break it.

1. You don't have enough safe guards in place for the AI. I could chat with it about any topic.

2. I can get better results from ChatGPT directly

3. There's clearly no privacy

replies(1): >>45028031 #
2. scal ◴[] No.45028031[source]
Thanks for your feedback. I’m currently making the necessary adjustments to keep Milo focused on travel. As for the results, at the moment the AI model being used is not the most advanced, mainly for cost reasons since the app is free and still in beta.

Regarding privacy (if that’s what you meant), when an error is logged, the content of the last exchanges is indeed included so that we can better understand the source of the problem. That said, you’re right — this point should be mentioned clearly under the chat modal, and it will be implemented in a future update.

Thanks again for the constructive criticism, it’s always appreciated.