happy wednesday, tunecasters here's my favourite song about Frankfurt https://www.youtube.com/watch?v=BtyjaSqdh2I
pretty booked up for next little while but a farcaster meetup would be fun sometime!
Maybe AI companies "get away with" using their training data up until now But then maybe many people adopt licenses (or enact laws) that prevent this use If so, then the "pre-2023 training set" might forever be special, the default umwelt of AI before online content starts to become inaccessible/expensive to train on
if you haven't seen it already, this is great: https://youtu.be/KRvv0QdruMQ
Link to the full doc! (whoops) https://venkatesh-rao.gitbook.io/summer-of-protocols/
have any of the "AI safety concerns are overblown, it won't kill us all" people written a good account of their argument?
useful Dune dashboard here: https://dune.com/johnrising/erc-4337
I like it for personal email, wouldn’t use it for work. Newsletter reader function is nice, and I like the “collections” feature that lets me bundle threads together by topic and view as one timeline.
welcome @jameszaki from the Privacy & Scaling Explorations group! (https://appliedzkp.org)
welcome @nloureiro, designer on the ethereum .org team!
happy wednesday, tunecasters https://youtu.be/EVZcc_iy4DA
possible! But: - Security isn't free! Such an L2 would have to start paying for its own security, e.g. thru inflation - L2 clients may optimize more for performance vs decentralization, making it challenging to become an L1 again - You might not keep your TVL if you are suddenly far less secure than every other L2