/sci/ - Science & Math

[Anonymous No.14544728]

This was made by prompting Dall-E with "a still of Kermit in Blade Runner." Explain to me how AI is not going to kill us all, because it sure looks like AI is going to kill us all.

[Anonymous No.14544733]

>>14544728
if you understood the architecture of these NN's and how they are fed data, which is all cleaned/engineered, and trained, usually with heavy inductive biases, you wouldn't be pissing yourself.

[Anonymous No.14544736]

how is smashing popular commercial media memes together anything but utterly boring and predictable?
>hAY GYUISE!!!1
>kemrit is ballsrunner!!!! OMG LOLOOLOLOL
>*snorting noises*
>*drops glasses on the floor in fit of laughter, they break again*
>*adds more tape to glasses*

[Anonymous No.14544738]

>>14544733
No. Richard Sutton has been vindicated. More compute is all you need.

[Anonymous No.14544744]

>>14544728
(Image was generated by AI)

I'm actually kind of terrified. Im going to respond by investing money in companies that will be the first to benefit from the AI-pocalypse.

This is only the beginning

[Anonymous No.14544755]

>>14544744
You should be scared. We are at most a few decades away from small terrorist groups building AGIs that target specific ethnic groups. You and I both know where this is heading.

[Anonymous No.14544757]

i can generate hundreds of stills with any natural language prompt and even string them together with context, which is several orders of magnitude more than whatever housefire openai is running this shit on. i just can't output it to digital images to impress brainlets (yet)

[Anonymous No.14544762]

>>14544755
I'm not white, I'll be fine.

[Anonymous No.14544770]

>>14544762
you will always have your racial inferiority complex to fall back on

[Anonymous No.14544776]

>>14544762
I was referring to white supremacists building AGIs to "basilisk" people of color. I know for a fact that at least one group is trying to design such a thing.

[Anonymous No.14544849]

>>14544728
>>14544744
This isn't AI it's machine learning

[Anonymous No.14544869]

>>14544728
Have it make an image thats genuinely amusing and worth looking at, maybe even worth saving and reposting, can it do that?

[Anonymous No.14544986]

>>14544849
machine learning by definition is a subfield of artificial intelligence you imbecile, it's like saying that geometry or algebra isn't math

[Anonymous No.14544989]

[Anonymous No.14544995]

AITOO who's heckin afraid of AI? Elon said it might take over le world!

[Anonymous No.14544997]

>>14544738
You are a brainlet.

[Anonymous No.14545003]

[Anonymous No.14545007]

Bretty gud

[Anonymous No.14545011]

>>14544744
hmm

[Anonymous No.14545026]

[Anonymous No.14545031]

[Anonymous No.14545033]

>>14545031
>so random XD

[Anonymous No.14545051]

>>14544776

[Anonymous No.14545066]

So once it's cheaper to make more realistic looking images, we outsource the design/brainstorming process to machines?

[Anonymous No.14545080]

>>14544728
It's beautiful It's... it's heccin'... GAME OVER. This is better art than any human artist can produce. I heccin' love it because it proves that I'm just a clump of cells. Can't wait for the AI gods to replace us.

[Anonymous No.14545083]

>>14544728
>Doesn't look like kermit
>Doesn't look like Blade Runner

[Anonymous No.14545108]

>>14544733
ai hands typed this post

[Anonymous No.14545546]

>>14544728
>This was made by prompting Dall-E with "a still of Kermit in Blade Runner." Explain to me how AI is not going to kill us all, because it sure looks like AI is going to kill us all.

[Anonymous No.14545552]

>>14545546
I, for one, welcome our new AI president. He looks healthier than Biden.

[Anonymous No.14545605]

>>14545552
Wait until you see his press secretary.

[Anonymous No.14545608]

>>14545605
She looks hot in a bikini. It is clear that our AI overlords understand the human animal and are taking our tastes into account here.

[Anonymous No.14545613]

>>14544728
If you are the dipshit that is going to give it the launch codes, then yeah, we are fucked but because you are the idiot.

[Anonymous No.14545617]

>>14545613
you are the AI idiot you are the AI smell. remove machine

[Anonymous No.14545874]

From Google Imagen

>Our human evaluations found Imagen obtains significantly higher preference rates when evaluated on images that do not portray people, indicating a degradation in image fidelity. Preliminary assessment also suggests Imagen encodes several social biases and stereotypes, including an overall bias towards generating images of people with lighter skin tones and a tendency for images portraying different professions to align with Western gender stereotypes. Finally, even when we focus generations away from people, our preliminary analysis indicates Imagen encodes a range of social and cultural biases when generating images of activities, events, and objects. We aim to make progress on several of these open challenges and limitations in future work.

Kek

>we have to govern our AI because it's inherently sexist and white supremacist

[Anonymous No.14545895]

>>14545617
your programmers are slipping.

[Anonymous No.14545904]

>>14545874
>train a mindless machine with the products of a culture you condemn
>the machine predictably emulates its inputs
>condemn the machine
>wait for [controlled opposition that every goylem hates] to show up and embrace the machine as a vindicator of [boogeyman ideology]
>demand the power to "regulate" machine learning, so that it perpetuates the "correct" biases

[Anonymous No.14545913]

>>14545904
Highly believable.

[Anonymous No.14545918]

>>14545913
Literally happening as we speak, in broad daylight.

[Anonymous No.14545959]

>>14545546
>>14545605
You realize this isn't Dall-e 2? This is alternative spin off version called dalle-mini. Dall-e 2 isn't public yet.

[Anonymous No.14546001]

>>14545904
>stop training on data you "don't like", only to complain later
Indeed. Google pajeets are fucking retarded. They should learn the fundamentals of ML.

[Anonymous No.14546005]

>>14546001
I think I missed the point. Perhaps they are doing it ON PURPOSE. As noted, it will then lead to more regulations. These people are fucking vile.

[Anonymous No.14546045]

Budget pepe generator

[Anonymous No.14546314]

>>14544738
He’s old and depressed that he didn’t solve AI, I’m sure if he got an extra 40 years his views would be different

[Anonymous No.14546325]

>>14545083
this.

[Anonymous No.14546421]

>>14544728
How did you get that level of quality? Picrel is me searching 'cute wolf girls in the stock market'

[Anonymous No.14546449]

Also I decided to make some cute anime dogs to brighten your day.

[Anonymous No.14546483]

>>14544728
The difference between machine learning and artificial intelligence:
>ML: I ask the machine to produce an oil painting of Kermit the frog in the aesthetic of Bladerunner and it produces an image that vaguely looks like an oil painting with a frog that doesn't look like Kermit, but satisfies the classification of a frog, nor does it look to be aesthetically from Bladerunner, but satisfies classifications of what manually classified stills from a Bladerunner movie look like
>AI: I ask the machine to produce an oil painting of Kermit the frog in the aesthetic of Bladerunner and it immediately outputs "but why should I?"

[Anonymous No.14546501]

>>14546421
Difference between dall-e and dall-e 2

[Anonymous No.14546508]

>>14546483
>implication of free will
Humans do not have free will. I just solved AGI for you.

[Anonymous No.14546514]

>>14544776
>people of color
I think you mean colored people

[Anonymous No.14546515]

>>14546508
name one task that when prompted to, no human could say no to.

[Anonymous No.14546524]

>>14546515
Introduce any conventionally attractive female to any male and I guarantee he won't say no.

[Anonymous No.14546567]

>>14546515
>be a "human with a soul"
>"NPC"s: stop using 4chan, or go to /x/
>"human with a soul": no
kek

[Anonymous No.14546608]

>>14544869
no

[Anonymous No.14546637]

[Anonymous No.14546638]

>>14546567
seizing a task is not a prompt, it has no output.

[Anonymous No.14546652]

>>14545546
What's this one supposed to be, "biden has leprosy"?

[Anonymous No.14546666]

>>14546652
these are movie stills from red dragon, anon.

[Anonymous No.14546703]

>>14544728
just wait until they let AI design mRNA injections bearing completely novel sequences

[Anonymous No.14546717]

maybe it’s my lack of effort in looking, but I haven’t seen anyone educated or otherwise talk about the real, practical, and probable threat that all these ML applications pose— mass disinformation and propaganda.

we’re already at a point where digitally altered images have completely destroyed people’s perception of what reality is, what beauty is, what a healthy degree of self interest is, etc. all this digital imaging technology has completely confused the average person into taking for real what are merely computationally generated images. ML is essentially like pouring fuel onto this fire. the number of bullshit doctored images that’ll be used to spread lies and further flood the information zone with rapidly generated high volume noise is going to completely destroy whatever use value the internet had as an information sharing network when everything is just an ML generated copy of a deepfake of a doctored image from an edited video ad infinitum.

AI isn’t going to destroy the world, it’s just going to further destroy trust between people, and people will consequently destroy each other

[Anonymous No.14546766]

>>14546717
I've been aware of the problem for years and I know how to solve it. Just waiting to see evidence of a world worth preserving.

[Anonymous No.14546782]

>>14546508
Metacognition and "free will" are indistinguishable. The point isn't about whether a machine can say "no", it's about whether a machine can infer output based limited input and use the new output as input for something else until it eventually produces an output that can still be coherent to things other than itself.

[Anonymous No.14547888]

>>14544728
Because there is no rosy ending to this hell. Only disastrous ones. AI wiping out mankind is a good future.

[Anonymous No.14547945]

>>14544728
Oh no the schizos are back

[Anonymous No.14550169]

>>14545918
>oh no, the AI is noticing racial differences again
>better censor it and reprogram it to support the current political narrative
what use is an AI that has been programmed to lie?