This is such a great idea. A tip if you are not already aware (and interested in music) is to check out “The List”, a Bay Area concert guide which has been maintained since the 90s: http://www.foopee.com/punk/the-list/

Thanks, I didn't know about this yet and this is A++++

EDIT: Got me some Death Grip tickets!

Jealous, have been hoping to see them for years but missed out on the last show they played in my country. Here’s hoping they add it to their tour.

Yeah, I rarely go to shows anymore just because I've been to so many in my life that it's not that thrilling anymore.

But I think Death Grips is worth a show, given their reputation for live shows.

I'm already dreading standing in GA after they start two hours late, holding my pee to maintain my good spot LOL

Ahh yes the list, it is shared to whoever doesn’t know about it. It’s the opposite of gatekeeping and a pleasure to see.

Amazing. I used the list 20 years ago religiously.

Please opensource this - or even a pastebin would work.

Totally agree, would love to have a local instance running

I built a site to scrape events from my city and show them all on one page. It was amazing for finding things to do, until Facebook shut down their events API and the site died overnight.

The secret is to not use any apis and just scrape with abandon (and then to only build it for yourself so that you don't get sued)

I tried, but scraping ended up being too hard, so I binned it.

Best real world example of something actually useful for ChatGPT I’ve seen yet.

Not to be contrarian but from what I could experience with chatgpt I wouldn't want it recommending things to me.

It's not controlling a pacemaker. If the events it chooses don't sound fun just don't go.

Lived in the bay the whole time, did the same thing. Multiple sources, categorization, distances, etc. Chaining GPT for rating is a nifty feature add though.

This is brilliant and could be useful in general. Very "Jeeves" like.

That's a startup product you got there.

Sorry for being a noob. But I am not sure that I understand how a chat gpt can decide this. As I thought that it just uses probability to find the next likely word for sentence completion. How does this even work? And can we trust what the ChatGPT outputs?

You should try it, which it sounds like you haven’t! :-)

I asked ChatGPT:

Is chatgpt a stochastic parrot?

No, ChatGPT is not a stochastic parrot. ChatGPT is a large language model that uses deep learning algorithms to generate text based on the input it receives. While it can sometimes produce nonsensical or inaccurate responses, it is not simply repeating phrases or patterns without any understanding of the meaning behind them. Instead, ChatGPT is designed to analyze and learn from a vast corpus of text data, and it uses that knowledge to generate responses that are intended to be relevant and meaningful based on the input it receives.

What's your prompt? I've been doing something similar but I've found it difficult to get chatgpt to output a consistent data format.

I use the chatgpt35-turbo "system" field for instructions, then put the event info into the "user" field

--- PROMPT ---

You are given a meetup event description. For each of the following attributes, return the percent likelihood that the event involves has that attribute, in CSV format:

Attributes:

- Technology

- Pets

- Exercise

Example:

Technology,0.3

Pets,0.1

Exercise,1.0

(The example values should be in percentages, can't edit anymore)

Always give good examples in the prompt, preferably a few (2-4). Still not 100% reliable though, especially with models below GPT-4.

Ask it to return only valid JSON and show it an example, best GPT trick I know. I use this for my cocktail recipe generator https://yourdink.is

Now that's an embarrassing typo. I meant https://yourdrink.is

Are you willing to share the source code? I wanna do this for London

How did you measure the driving distance?

I asked chatgpt4, and it did some insane formula that uses lat/long and the radius of the earth to measure distances, then asked it to add a bonus 20 minutes for any destination in SF, 15 minutes for any event in Berkeley/Oakland (and it correctly put bounding rectangles on the three jurisdictions).

Cool, thank you for the insight.

I stumpled upon a similar quest where my app had to calculate the walking distance between two coordinates (the client and the restaurant).

OSM seem to be the only good option, I could not find any better alternative in my case (Ruby).

Can you share the promo you give to gpt?

hey i'm building something very similar, pls put this on github would love to hack on it!

this is brilliant, please open source it. I need AI to tell me which AI events to go to.

+1; would love code for this!!!

Same