Cookie Thread Act 3: The Cookie Strikes Back

General
#19901

I would suspect that new forms of machine learning will come along, wouldnt be that worried

#19902

currntly working on Machine Learning 2

#19903

a machine that is flawed from first principles is never going to answer “Yes” to “Am I flawed from first principles?” and it’s never going to accurately explain why

#19904

Chess has like 7 major engines that analyze one another

#19905

“ But it’s not obvious that Socrates’ interpretation is the right way to understand the oracle’s message. In fact, the priestess didn’t say that Socrates was wiser than everyone else, she said that no one was wiser than Socrates. This is compatible with the following propositions:

  1. Everyone is as wise as Socrates.

  2. Socrates is no wiser than anyone else.

  3. No one is very wise.

  4. No one is at all wise.

So the oracle may have been saying that Socrates is not at all wise and in that respect is no different from, and certainly no better than, anyone else. But Socrates immediately turns this into the pronouncement that he is the wisest – meaning (a) that he has some kind of wisdom (which the oracle didn’t say) and (b) that he has more of it than anyone else (which she also didn’t say”
This is the funniest interpretation

2 Likes
#19906

yeah fun fact we still don’t really know what knowledge is. edmund gettier has a pretty cool two page paper in which he overturned the entire field of epistemology and it hasn’t been figured out since https://fitelson.org/proseminar/gettier.pdf

#19907

the strongest chess engine is a learning algorithm that they just left running overnight

#19908

even if you do believe knowledge is “justified true belief” there’s the question as to whether the “justified” part is even needed. i have known things with completely inaccurate justifications for why

#19909

all of which function according to very similar principles
i accept that chess engines are very neat algorithms, but i actually don’t think the chess engines can meaninguflly report on each other even if they can meaningfully make decisions

#19910

the strongest cheese engine is a mouse named story

#19911

good. if we ever know what knowing is we will explode. Stop looking into the brain. We should NOT know how it works. we will all perish

#19912

personally I think that instead of making false criticisms of AI companies (“OpenAI isn’t even vaguely trying to understand how their AIs work”) we should make true comparisons (“OpenAI made a big deal about how much they care about safety, wrote policies that prohibited the use of their AIs for military, warfare, or weapons development, then deleted those policies and started working with the literal Pentagon”)

there are lots of valid criticisms of AI companies!

1 Like
#19913

but that first point is fucking true

#19914

Wait what

1 Like
#19915

you cannot prove your system isn’t fundamentally flawed using that fucking system

#19916

This ai doesn’t tell me how I can rob a bank but it works with the them

#19917

no yeah that’s true they’re training models to perform spotting for artillery

1 Like
#19918

“We care about public PR until the governments pays us a lot of money”

tale as old as time

1 Like
#19919

imagine working with the literal Pentagon

1 Like
#19920

i liked the part where people started intentionally poisoning ai learning out of spite

1 Like