I've been toying with this for a few years now. Here's what I've learned:

1. There's no tool out there (currently) to do exactly what you're looking for
2. In order to achieve the results you want, you'd need a tool that's able to query flight prices in real-time. This is problematic due to rate limiting and ChatGPT won't have access to real-time flight prices.
3. You used to be able to build your own, but be prepared to spend some cash on it. Here's how I did it years ago:

• Create matrix of airport distances from your origin airports. For example, if I live near SFO, I'll store the values of distance between SFO and all of the other major airport codes worldwide. Would copy those values for SJC and OAK as those are also viable alternatives
• You generally have 3 variables for searching: destination, date, and carrier. This is where the rate limiting comes into play. You'd need to search hundreds, if not thousands of different date/airport combinations to get pricing.
• Once you have the pricing for a particular pair, you can calculate a cost-per-mile based on the distance matrix you calculated earlier and sort the results by cheapest.

Back in the day (2017?) I created the above rather easily in Python and paid for API access to Google Flights (think it was QPX express)
Google Flights no longer offers an API which led me to give up on that little hobby.

If anyone has been more successful using any sort of AI tool, or building their own, I'd like to compare some notes.