america has a big consent problem

test("increment adds 1 to x", () => {
expect(increment(0)).toBe(1);
expect(increment(-1)).toBe(0);
expect(increment(41)).toBe(42);
});

this should be standard with agents. I use a similar setup and love it

ai alignment issues are broadly either A) societal misalignment issues (why is a creepy, useless pedo given control of so many resources?) or B) technical (is there a mathematical way to stop a shoggoth?) ... both important but people seem to choose one and disregard the other

unfortunately reddit already force fed it to me then banned me for making a benign comment about the general situation 3:

i hope they acquire google balls i would love for it to be an actual product that gets discontinued in 4-18 months before they quietly make google spheres

this is 100% going to land them in legal trouble and still be worth it from the perspective of improving health outcomes especially preventatively. but im so confused why you dont start with the personal trainer version

(error) 'model.layers.5.mlp.experts.gate_up_proj'

who do you think pays their salaries?

(error) expected scalar type BFloat16 but found Half

law? enforced? i thought they stopped doing that

its ok... the joke is this is how he interacts with humanity, by watching through a window and complaining

(error) Network Error

I am glad the idea resonates so strongly. one thought here. often addressing the challenges feels like the reason you haven't made progress when in reality it was a prerequisite to making even more. the flow state comes and goes as it wishes but rest is never truly wasted

sounds fun😎 also I think bluesky has reels that dont spy on you now might be worth trying instead

the best part about new beginnings is you can't really prepare for them, so there's no worry about being unprepared either

interesting -- are you an architect yourself?

personally - i wish more brutalist buildings had more plants, makes them less imposing and disconnected