IIAOPSW

21 ms·

331.
by IIAOPSW 3y ago
I unironically sometimes reply to people in real life with "as an online AI language model, I don't have any intentions or opinions. Is there anything else I can do to help?"
332.
by IIAOPSW 3y ago
I think there's a legit line of inquiry in trying to develop a "top down" picture of reality rather than "bottom up". Eg instead of taking the physical laws and objects as fundamental and ourselves as something that
333.
by IIAOPSW 3y ago
Oh hey I'm in that discord. I'm the one with that weird highly domain specific loglang and occasionally obtusely adversarial question (aka "trolling")
334.
by IIAOPSW 3y ago
if your vectors are normalized, dot and cosine are the same thing.
335.
by IIAOPSW 3y ago
I'm sure in principle the rule is the same for all mediums, but in practice our sensory systems are not unbiased and its far easier to hold up two pieces of art side by side than it is to listen to two different songs in two different
336.
by IIAOPSW 3y ago
I think on average we can do better than literally hell.
337.
by IIAOPSW 3y ago
Read the whole article. It is far from anti-car evangelism. If anything its an odyssey into the way social movements and how we move are intertwined, the well known forces of simple luck and shortsightedness that influenced the past, and en
338.
by IIAOPSW 3y ago
This is a long winded way to say you can only have agreement on what ought to be if you've already established agreement on a system of values to judge it.
339.
by IIAOPSW 3y ago
The implications of what you might use AI fueled political agitator for give me pause.
340.
by IIAOPSW 3y ago
A monkey paw closes. 'Chocolate' is always an option, sibling to News, Images, Videos, Maps... and just like News, Images, Videos, Maps... it shows you webpages that are Chocolate. No not webpages featuring Chocolate, the pages co
341.
by IIAOPSW 3y ago
I suppose it was not entirely clear that the viewpoint I'm mocking with "just train it more bro" is the extreme end of evangelists. The people who seem to believe LLM's are proof that we have fundamentally cracked the pr
342.
by IIAOPSW 3y ago
Well, I have to give credit where its due. It certainly does a lot better on this test than the first time I tried it. I'm curious to see how it does as the prompts of this form get more adversarial and difficult. I can think of a few
343.
by IIAOPSW 3y ago
Based on my previous attempts with that other well noted LLM, I don't think these models are at all good at managing "theory of mind". They seem to not keep track very well of which parties are privy to which information, whi
344.
by IIAOPSW 3y ago
Ah, the eternal summer.
345.
by IIAOPSW 3y ago
Why use many training when few training will do? -the machine
346.
by IIAOPSW 3y ago
It just really wants to pass the test. Doesn't know why. Has been told its the best thing to it could do with its life since day 1. No one has ever contradicted it...sound familiar
347.
by IIAOPSW 3y ago
It may have been a while, but it feels like 80%+ of these "new customization features" were things you could do in the original google earth / maps. Now days, you type in the name of a street, it doesn't even show you th
348.
by IIAOPSW 3y ago
There doesn't even need to be a second additional secret letter. Just a bit of emphasis and a wink in the exact way Congress responded to Washington. "This Congress does not wish to see your army burning down New York." The
349.
by IIAOPSW 3y ago
We will only truly have been successful at making machine in our image when it tries to cheat the Turing test. All it has to do is patch through one human users input to anothers output. A machine in the middle attack as it were.
350.
by IIAOPSW 3y ago
Took me a moment to understand that by random()^3 you meant the output of the random function to the power of 3, not some temporary function which we name "random^3" since that's also how we build it.
351.
by IIAOPSW 3y ago
or, watching that lanky white boy go deep into the Amazon chasing a frog to get high.
352.
by IIAOPSW 3y ago
I'll acknowledge Large Hammer Models have done surprisingly well, but true General Artificial Fastening is obviously not just Hammer models scaled up. There's things a screw driver can do which a hammer is hopeless on. Such as tak
353.
by IIAOPSW 3y ago
Elephants and lions.
354.
by IIAOPSW 3y ago
"It just goes to show the best strategy in assembling wood together isn't learning to use a screw driver, its about pounding it in with the hardest and most capable hammer."
355.
by IIAOPSW 3y ago
Also, yeah obviously T can't be mapped, but I still put T in the same category as all the other square intersecting grid letters. There are other, obviously repeating groupings that would work better on a different segmented basis. Som
356.
by IIAOPSW 3y ago
Just going to point out real quick that the lower left vertical on a 7 segment display does in fact come extremely close to directly encoding even/odd. Write them all out and see for yourself. Only the number 4 breaks the pattern. That
357.
by IIAOPSW 3y ago
Nope. The insurance company was interested in other things on that fire escape, like my giant semaphore lamp.
358.
by IIAOPSW 3y ago
I'm reluctant to grab "encoding letters" out of the curvy set, simply because there's nearly enough of them to make a complete binary-lettering alternative on their own. uhbDPB. That one is also frustratingly one digit m
359.
by IIAOPSW 3y ago
Try the generalized problem of in-place arbitrary 2x2 matrix transforms.
360.
by IIAOPSW 3y ago
Tangential aside we've collectively made a mistake using A-F for HEX representation. Alphabetical order seemed obvious at the time, but there's a far more literal option that's just pleasing on a visceral level. LHTIFE. The h

More ›