[Boards: 3 / a / aco / adv / an / asp / b / bant / biz / c / can / cgl / ck / cm / co / cock / d / diy / e / fa / fap / fit / fitlit / g / gd / gif / h / hc / his / hm / hr / i / ic / int / jp / k / lgbt / lit / m / mlp / mlpol / mo / mtv / mu / n / news / o / out / outsoc / p / po / pol / qa / qst / r / r9k / s / s4s / sci / soc / sp / spa / t / tg / toy / trash / trv / tv / u / v / vg / vint / vip / vp / vr / w / wg / wsg / wsr / x / y ] [Search | Free Show | Home]

https://www.fastcodesign.com/901342 78/biased-ai-is-a-threat

This is a blue board which means that it's for everybody (Safe For Work content only). If you see any adult content, please report it.

Thread replies: 30
Thread images: 5

File: 63354479_p0_master1200.jpg (458KB, 1200x846px) Image search: [Google]
63354479_p0_master1200.jpg
458KB, 1200x846px
https://www.fastcodesign.com/90134278/biased-ai-is-a-threat-to-civil-liberty-the-aclu-has-a-plan-to-fix-it

https://www.theguardian.com/commentisfree/2017/apr/20/robots-racist-sexist-people-machines-ai-language
As stupid and sensational as this is in reality (as the AI they're describing are really just advanced algorithms, this DOES however open up new possibilities in future stories about AI. Imagine an inventor keeping his newfound discovery of true AI a secret because interest groups might lobotomise it?

Pic unrelated.
>>
>>15739090
It already happened
RIP Tay
>>
>>15739090
>Grauniad.
>>
>>15739101

IIRC, Tay's case was hilarious, as it comes out fucked up in many different ways after every lobotomy sessions to fix it.
>>
>>15739090
>it's another "dumbasses mistake algorithms regurgitating shit for actual artificial intelligence" episode
at least it's not another "peppers ghost is a hologram" episode
those were so fucking awful
>>
>>15739090
Nah.
The west will keep pumping out the same turned against their masters shit
The east understands that reality can be racist, and doesn't care

Fiction will continue business as usual
>>
>>15739090
that confirms /pol/ did nothing to Tay. it just how computer math works.
or are you going to tell me a system is capable of being racist
>>
>>15739090
So more crazy AI stuff. Would be fine seeing that used more in shows that aren't Digimon.
>>
File: 1478058772153.png (592KB, 1440x2560px) Image search: [Google]
1478058772153.png
592KB, 1440x2560px
reminder that /pol/ has 3 beautiful super racist AI daughters
>>
File: disgusted aramaki.jpg (4KB, 125x125px) Image search: [Google]
disgusted aramaki.jpg
4KB, 125x125px
>laurie penny
>>
sauce if anyone cares
https://www.pixiv.net/member_illust.php?mode=medium&illust_id=63354479
>>
>>15739449
BlackBolt has a DA, too (albeit the drawing doesn't have the dialogue box on there).
>>
>>15739090
>purging unintended behavior from your AI because people keep feeding it garbage for shits and giggles, grossly distorting its learning acquisition, is lobotomization if I don't like it

The bigger problem is the fact that the Internet, or more accurately the social media slice of the internet, is a comically terrible input source for a learning algorithm. You can go from zero to NIGGERS NIGGERS NIGGERS in fractions of a second, completely wasting development effort and computational cycles teaching Watson how to be a /pol/lack.

Also, CS degree holder here, technically learning algorithms form the starting point of the artificial intelligence discourse. If you want to be autistic about it the semantics lie in the question of weak vs strong and whether the objective of AI is implementation and applications of machine learning (which is where things are right now more or less) vs trying to make a self aware sentient piece of software that people will fearmonger over being the next skynet (more of a philosophical question and not of immediate application to everyday living until the need to automate the human creative process or something similarly abstract becomes necessary).

At any rate, the question of racism in the results would indicate something's fishy with the input. Again, garbage in garbage out, zero to NIGGERS, all that jazz.
>>
If people can get this freaked out over easily exploitable algorithms, then it does stand to reason that an actual self aware robot would attract all sorts of people to either dissect it or lobotomize/destroy it if the links are of any indication.
>>
even with fearmongerers governments probably realize the threat a super intelligent AI would pose following the trend of exponential growth and if someone else makes one first every body else is fucked. but i'd imagine they'd keep that all low-key
>>
>>15739712
agreed. the way people behave online is totally different from the way people behave in real life. but i mean, if your AI is meant to live in the online world, then why stop it
>>
tl;dr retarded "journalists" who can't tell the difference between an IRC chatbot and a strong AI write sensationalist articles about pure nonsense pandering to stupid people so that they can make money off hits, and stupid cunts like OP help by linking them instead of using archives

What else is new?
>>
>>15740444
Because the only thing more obnoxious than a /pol/ster trying to shill their special brand of memery is a program that screeches nonsense about THE JEWS while generating vaguely original content with a level of efficiency that would render 4chan obsolete within a year.

The former derives at least some humor from the experience, but the latter is just a noisemaker on full blast, an infernal digital shitpost engine.

On the plus side, if you can "learn" shitposting, then you should theoretically be able to use that to improve current social media quality metrics and, I dunno, remove the moderator middle man from 4chan in favor of a soulless collection of code that can swing the banhammer with inhuman precision.

But then that gets into a philosophical issue of whether or not it's censorship when a computer is the one calling the shot, I guess?
>>
>>15739101
I really wouldn't consider that a true ai though, it just repeated shit that was said to it and took bits and pieces from phrases so that it appeared to be holding a conversation, that is like believing Cornelia was an actual ai and not just a slightly more annoying spambot
>>
>>15740535
>But then that gets into a philosophical issue of whether or not it's censorship when a computer is the one calling the shot, I guess?
I mean, you COULD blame whoever programmed it but I honestly doubt most shitposters would make that connection, also did they ever reinstate the bot back in /r9k/? Because what started the whole breakdown of the site was its removal

I honestly wonder if it wouldn't be best to make it work sitewide
>>
>>15740535
>an infernal digital shitpost engine
>tfw the robots are coming for your hobbies rather than your jobs
>>
>>15740744

Even that sounds like a better idea for a Terminator reboot than Moviebob's idea.
>>
>>15740744
>>15740786
>hurr durr reddit
This is actually a really interesting essay on the topic of this thread and I think everyone interested in the discussion should check it out:
https://youtu.be/7Pq-S557XQU
It's well-informed, it doesn't waste time on autistic character persona, it's worth 15 minutes of your time.
>>
>>15739090
That's what Neuromancer is about.
>>
File: colonel_4852.jpg (9KB, 135x196px) Image search: [Google]
colonel_4852.jpg
9KB, 135x196px
>>
>>15741587

Do you still need those 61 Scissors?
>>
>>15741587

Fictional Tay?
>>
>>15739090


this happens all the time, read a fucking book
>>
Biased people reproduce their biases in whatever they create. Not news to anyone except those who refuse to see their own biases.
>>
>>15741518
Gave the video a shot, it makes its point well, which is more than I expect from YouTube vids to be honest.

I do question the ability to learn the creative process, but then again that edges into philosophical nonsense as to what defines creativity. You can procedurally generate content to pursue an arbitrary success metric, and that is addressed within the video under the subject of automating creativity, but I feel there's a difference between that kind of creativity, and the arbitrary standards that philosophy and the arts will suggest.

Mind you I did CS in college, not liberal arts nonsense. Still, it'd be fun to see an art exhibit full of stuff generated via machine learning.

>tfw failed my intro to AI course due to the teacher being terrible at teaching the material but now want to self learn this stuff from scratch

I've had the smart shitpost filter idea floating around in my head for a while now, maybe I should do something about it.
Thread posts: 30
Thread images: 5


[Boards: 3 / a / aco / adv / an / asp / b / bant / biz / c / can / cgl / ck / cm / co / cock / d / diy / e / fa / fap / fit / fitlit / g / gd / gif / h / hc / his / hm / hr / i / ic / int / jp / k / lgbt / lit / m / mlp / mlpol / mo / mtv / mu / n / news / o / out / outsoc / p / po / pol / qa / qst / r / r9k / s / s4s / sci / soc / sp / spa / t / tg / toy / trash / trv / tv / u / v / vg / vint / vip / vp / vr / w / wg / wsg / wsr / x / y] [Search | Top | Home]

I'm aware that Imgur.com will stop allowing adult images since 15th of May. I'm taking actions to backup as much data as possible.
Read more on this topic here - https://archived.moe/talk/thread/1694/


If you need a post removed click on it's [Report] button and follow the instruction.
DMCA Content Takedown via dmca.com
All images are hosted on imgur.com.
If you like this website please support us by donating with Bitcoins at 16mKtbZiwW52BLkibtCr8jUg2KVUMTxVQ5
All trademarks and copyrights on this page are owned by their respective parties.
Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.
This is a 4chan archive - all of the content originated from that site.
This means that RandomArchive shows their content, archived.
If you need information for a Poster - contact them.