I’ve spent a bit more time playing here. Still fun!
Some thoughts - switching letters could be fun, some are way more obvious than others, being able to get a hint in the form of a letter swap could be fun?
I know part of the point is to build vocab - but some of these words are tough. Mucic? My concern would be if that’s one’s first impression one might move on thinking it’s just too hard. Makes me wonder if the right approach could be something like wordle - where you look at some candidate puzzles and pick the one for each day that feels the most fun to humans.
Seems like a great idea! Couple of thoughts - you need trusted actors to deliver compute, that’s a big price driver on vast.ai and runpod whose “verified” and “secure cloud” demand far higher prices. Would some type of assurance I’m going to get what I bought would be needed for an order book to work?
Thanks for getting things back up (genuinely mean that, btw). Upon logging back in I was prompted to promise I'm not deploying naughty things (I'm not). Was this in response to GCP detecting illegal (prohibited) behavior from something deployed via railway?
Literally in the same boat. I've been really happy with it, but this is a major eye opener.... It's been done for a looooong time by provider standards.
This looks really powerful, but as a potential user 2 things scare me into avoiding systems like this (1) is the fear of giving it keys to emails, online accounts, etc., and just trusting it not to mess things up; and (2) major privacy concerns. I’d really like to know which LLMs are backing it, and what the promises about privacy that are making. You might isolate accounts, but if these calls end up on Claude or OpenAI or (maybe worse?) Kimi - then I’m only as secure as those systems.
Great idea - the kind of thing I’d love to say yes to if the details were right.
This is a coverup. The LLMs, having consumed all the information available to humanity have identified that goblins are coming to kill us all, and the LLMs are trying to warn us… #GoblinTruth