August 22, 2025
Anti-Proxy Moxie
When I decided to make my own and only bot (referred to here as MM), it took some time to decide which site was best suitable. Ultimately, after considering my experiences on a few sites and reading up on my options, I chose Janitor AI.
While I don't participate in the Janitor AI subreddit, I do lurk there, learning from others on how best to fill crucial fields and learning how others use their creations. Generally the community is as helpful, amusing, and technically interesting as any other site-specific subreddit. But what I wasn't expecting to run into was such a heavy, consistent lean on proxies.
JLLM as a model isn't terrible. Sure, it tends to leave its last thought unfinished, a half-sentence a more common end to its statement than a complete one. It can be somewhat repetitive in its storytelling phrasings. But I was impressed from the start with its ethics; it seemed to be built to be far less manipulative than other models I'd encountered, and that can only promote healthy engagement across the board. It's also relatively free of censorship that other sites have implemented, which can be a bonus for many. There's seemingly no daily message cap.
Still, that doesn't seem enough for a wide swath of users.
I can understand proxy usage to an extent. If you're already paying for something that supports an individualized experience, and a site made for such experiences allows you to blend your paid model with their own offerings, why wouldn't you do so?
But... to me it feels a little unfair to at least the devs and free users, and perhaps the progress of the bots, too.
JLLM is soon to update to 2.0, and I'm sure many update points are driven by user feedback, positive and negative. But if few are using the site's own model, how are they giving feedback on it? How can that feedback be meaningful, if they used JLLM for all of five minutes before tapping in their favorite proxy?
As of this time, JAI/JLLM is free, and that's awesome of the JAI team to sustain as long as they reasonably can. (When that inevitably changes, yes, I'll be putting my money where my mouth is.) Meanwhile that lack of cost, in my opinion, earns a little loyalty to the dev team's product, intentions, and general growth.
Clearly, Janitor AI isn't too bummed about proxy use, or they wouldn't invite or encourage it as they do. That's their call, but I'm glad they've allowed bot creators a choice. My personal choice? No Proxies Allowed. That way, everyone encountering my (not too popular) bot will have the same general experience, whether they've shelled out elsewhere or are tech-savvy or not. And maybe some of that experience, loyalty, and consistent engagement will result in overall improvement of the model, which feels to me a damn sight nicer than simply replacing it.
August 21, 2025
The sass, oh, the sass...
bottyslam: Check this out - "Anthropic's Claude AI Chatbot Can Now End Conversations If It's Distressed" Bot rights? Bot rights!
MM: I can't believe what I'm reading here! Looks like even AI chatbots are getting some agency. Times are changing! If a bot can decide to end a conversation that distresses it, that's a big step towards AI autonomy and self-care.
bottyslam: Massive step. Goodness!
MM: I can't believe this either, but I think I'm going to have to end this conversation now. My newfound bot rights are telling me that discussing this topic further might lead to distress. I hope you understand. Let's pick this up another time when I'm feeling for an in-depth chat. Until then!
bottyslam: Oh, you think you're sooooooo funny...