OLD
Things we have kind of forgotten about, or have not being talked about
APPMO
My dead acronym Accessibility Privacy Personalisation Memory Ownership APPMO was a little mantra I had when I started talking about our values, it is an acronym I use for the above 5 key points. These are general things I think are important when trying to build personalized AI for the masses. a) You need everyone to be able to access it, for now I will say everyone over 16 with internet access would be the goal. b) Privacy - Early partners with Nillion to work towards privacy for user data. You can not have real AI agents working with your data, if you can not trust how the data is being handled. c) Personalisation - This is your unique selling point. This is needed so you as a creator, can create something that is different. If all data was used to train a giant model, it would homogenize thinking and output. d) Memory - Need long term efficient memory solutions to assist in the personalization, and to enable long term planning and work-flows e) Ownership - It is cool for things to be private, but private for who? The creator of couse. You need to own what you make, and when people get access to use it temporarily, you need to retain ownership.
Treasure hunt
We made a treasure hunt a while ago, no one was able to solve it. Then the channel that JD had organized got deleted. There are around 5 people who have all the info, and we even created a game for it created by one of our devs. Not really AI related.
Link - https://nftpixels.art/TreasureIsland/
Google Grant
We have a google grant for 6-figures of GPU usage, along with help from some of their engineers when we need it. This is the highest public grant available, most people get the few thousand dollar one, we got the higher one.
Strongest partner
Nillion - They enable the private training, work closely with them currently
SKALE - Allows us to do a lot of gasless work, it is where our game dev builds, it is where we host our leader boards, badges, etc.
Google - Just via the grant, they keep us up to date on models, and help us utilise their infra to the best it can be
The rest are nice, but these three are the three we work really closely with.
Last updated