site banner
ArjinFerman  Tinfoil Gigachad  4mo ago (text post)   734 thread views

Tinker Tuesday for November 19, 2024

This thread is for anyone working on personal projects to share their progress, and hold themselves somewhat accountable to a group of peers.

Post your project, your progress from last week, and what you hope to accomplish this week.

If you want to be pinged with a reminder asking about your project, let me know, and I'll harass you each week until you cancel the service

1
Top Bottom Old Controversial Comments
Jump in the discussion.

No email address required.

I'm automating Hinge. Android emulator, pyautogui, PIL, GPT-4o. It's almost too easy.

The flow is:

  1. pyautogui: take a screen shot. With a 1:8 aspect ratio on the emulator this gets the whole screen. Earlier versions scrolled and stitched together, which mostly worked, but boy do I feel stupid not thinking of it sooner.
  2. AI: prompt to extract information from the info section (height, job, age, education, etc), an assessment of personality (nerdy, travel loving, high fashion, etc), and a physical description (weight, race, hair color), and an overall assessment of if she's my personality/physical type. (Note: the goal is nerdy but hates travel and isn't high fashion!)
  3. python: ignore literally all of #2 except the job and education. If either matches a whitelist of terms that signal smarts, proceed.
  4. PIL: split the screenshot into sections, using the like buttons.
  5. AI: transcribe (for prompts) or describe (for images) each section (separately) and provide a response. (My favorite part: I have it refer to her as The Candidate, which is how we have to write interview feedback at work.)
  6. AI: given all the transcriptions/responses, pick the best one.
  7. pyautogui: click heart button, type response, send.

Costs me about $0.04 to reject, $0.10 to message. I think I can get that down some. I only ran it for one batch, and it got a match faster than I normally do. Small sample size, but I am optimistic.

As to why #3 is so simple - I initially had a hand written weighted average of all the things, but looking at the actual behavior realized that:

  1. Really all I care is that she's smart and not terribly fat
  2. GPT-4o is not good at telling me if she's fat, so far.

Favorite kerfuffle: it messaged a woman, shown in a photo by a giant 10ft novelty planted pot: "is that enormous, or are you tiny"? It...was certainly not the latter.

This raises some questions for me:

  1. Do I let it do more than the first message? Probably not - it's just the endless swiping/messaging into the void I dislike. Conversion rate match -> date is tolerable.
  2. Could I let it literally do everything up to and including putting a calendar event on my calendar? Probably so - I'd say it'd cut my conversion rate in half.
  3. Do I admit I'm doing this? n=1, but I did, immediately, and it went well for me.
  1. I'd say no, a second message is a key part of the funnel and I wouldn't outsource this. Please say you have decent game and can actually perform better than a robot here.
  2. You could have it schedule for you while scanning recent messages - you handle the human part, it handles the paperwork.
  3. I would, at the first date. If she thinks it's cool, you've found a good match. If she's offended, she's probably too solipsistic and sensitive to be a long-term play.

What's your goal here? Sex or love?

Side note: I fucking love that you did this for many reasons. Congrats!

  1. I doubt I will outsource, I find that part easy and almost enjoyable. That said, the AI is better than you'd think at it. I think if I gave it enough information about me, it'd be fine. I figure the formula is roughly: respond to what she said, ask followup questions, tie in something about yourself. 1/2 are very AI friendly; 3, moderately.

\3. Interesting, I think saving it for the first date might be a good call. I think it's a good story at any time, but probably more useful there.

Goal is love for sure. If I just wanted to get laid, the algorithm would not need the "is very smart" filter, and even the "not very fat" filter could be relaxed...

Side note: I fucking love that you did this for many reasons. Congrats!

Thanks, I've wanted to do it for a while. I really thought it would be harder. I'm curious to try about o1-preview as well, although it's 3-4x the cost. GPT-4o-mini was not adequate for sure.

Curious to see how Claude sonnet 3.5 (new) performs. Feels like it has more emotional intelligence than ChatGPT. You might have invented a new AI benchmark.

How does it do for opening messages? I remember when I was dating and my lack of creativity really held me back there.

It's plenty good. The prompt basically just says not to be too enthusiastic, and to keep it short. I've been writing down my manual messages to prompt it with as examples, but am not even using that yet.

E.g. the match was responding to a picture of a woman standing in front of a field of sunflowers, captioned "My happy place": "Sunflowers might just be the happiest flower there is. What's your favorite outdoor place to relax?" A little extra, but it shows "I" looked at the picture, makes it about her, is easy to answer/start a conversation with - not bad imo.

If messaging is your bottleneck, you can use basic chat interfaces for response ideas. I would not overestimate how much the specific message matters - I'm pretty sure it's more of a rule-out than a rule-in kind of thing. Just don't be a creep basically.

What is this place?

This website is a place for people who want to move past shady thinking and test their ideas in a court of people who don't all share the same biases. Our goal is to optimize for light, not heat; this is a group effort, and all commentators are asked to do their part.

The weekly Culture War threads host the most controversial topics and are the most visible aspect of The Motte. However, many other topics are appropriate here. We encourage people to post anything related to science, politics, or philosophy; if in doubt, post!

Check out The Vault for an archive of old quality posts. You are encouraged to crosspost these elsewhere.


Why are you called The Motte?

A motte is a stone keep on a raised earthwork common in early medieval fortifications. More pertinently, it's an element in a rhetorical move called a "Motte-and-Bailey", originally identified by philosopher Nicholas Shackel. It describes the tendency in discourse for people to move from a controversial but high value claim to a defensible but less exciting one upon any resistance to the former. He likens this to the medieval fortification, where a desirable land (the bailey) is abandoned when in danger for the more easily defended motte. In Shackel's words, "The Motte represents the defensible but undesired propositions to which one retreats when hard pressed."

On The Motte, always attempt to remain inside your defensible territory, even if you are not being pressed.


New post guidelines

If you're posting something that isn't related to the culture war, we encourage you to post a thread for it. A submission statement is highly appreciated, but isn't necessary for text posts or links to largely-text posts such as blogs or news articles; if we're unsure of the value of your post, we might remove it until you add a submission statement. A submission statement is required for non-text sources (videos, podcasts, images).

Culture war posts go in the culture war thread; all links must either include a submission statement or significant commentary. Bare links without those will be removed.

If in doubt, please post it!


Rules


Recommended Posts And Communities

Recommended Realtime Chats