phone alarm:
- boring
- can snooze through it
- can silence it and keep sleeping
robo vacuum on a morning schedule:
- noise doesnβt let you fall back asleep
- makes you stand up and pick things off the floor
- makes your home cleaner!
phone alarm:
- boring
- can snooze through it
- can silence it and keep sleeping
robo vacuum on a morning schedule:
- noise doesnβt let you fall back asleep
- makes you stand up and pick things off the floor
- makes your home cleaner!
interesting that bot accounts here look so similar to x: attractive female profile pics with real names. i guess ppl just plugged in a new api to the same algorithm
I like to visit the Google Cambridge office when visiting Boston for nostalgia reasons but this year, 2.5yrs after I left, it looks like everyone who could've invited me as a guest has now moved/quit. Tech turnover rates are crazy.
Examples:
- GCG, PGD and other token-space optimization algorithms (optimize discrete F: token-sequence -> R)
- SAEs (why does a simple sparseness hack work so well?)
A lot of modern AI research feels kinda similar to operations research, in that:
- most interesting problems are in general NP-complete
- but approximate algorithms work remarkably well in real-world cases
- and we don't know what structures precisely make the approximations work
my experience using the new claude computer use
so any environment where there's a reliable "action done" event that you can catch programmatically works much much better
example trace from one of my experiments:
- claude clicks "expand"
- expand takes 2.5 seconds, so the next screenshot sent to claude (after ~2s) has no changes
- claude clicks again, but by that point expansion already happened, so the menu gets collapsed
- claude gets very confused and gives up
but also, some manual integration remains necessary. claude doesn't see a video, it sees screenshots with long gaps in between (the reference impl is like 2 seconds). how does claude know the application finished responding to the click, vs is still loading, vs the click didn't go through? it's hard
sorry i meant that desktop has those automations already so the marginal improvement is smaller than mobile.
but the killer use case will prob be smartphone app automation, it's something that's kinda hard to automate nowadays through other means. on desktop it's easier with things like applescript or windows desktop automate
imo it's still a little too dumb to be useful. we might need 6 months.
how do yall manage to use two social apps at once? i got decently good at writing out my thoughts but it feels really weird to just, idk, split it 50:50 between here and twitter