collapse

* Navigation

* User Info

 
 
Welcome, Guest. Please login or register.

* Who's Online

Author Topic: AI Discussion  (Read 112 times)

0 Members and 2 Guests are viewing this topic.

Online BlackCurtains

  • Mad Scientist
  • Moderator
  • Trade Count: (+69)
  • MIB Licensing Show Pinkie Pie
  • ***
  • Posts: 10736
  • Orange ponies are the best.
    • View Profile
    • BlackCurtains Crafts
AI Discussion
« on: Yesterday at 01:48:00 PM »
I feel this needs it's own thread.

I'll quote some responses from the other thread here.


as far as AI goes i'm very very skeptical. i think it's something that could be beneficial if it worked well and could do mundane tasks. the problem is so much of LLM type AI doesn't work well. that's the kind of AI that takes in a bunch of data and provides output. ChatGPT, Diffusion, etc. i'm not super familiar with the generative AI for art. it seems like it could be cool for creative artists but i do think there should be some kind of compensation/opt out for artists who's work is used to train the AI. do these companies even know who's art was used thought? i doubt it. and the idea of people selling AI generated art doesn't sit well with me.

i'm not an artist but i am a web developer and the code hosting services i use have rolled out AI coding assistants that learn off code that people store on their hosting services. there is no compensation for using people's code to train the AIs. in fact they charge you to use the AI and your code helped train. there is also no guarantee that the code training the AI is good, working code either. the results based on my personal testing are hit or miss. it's nice for repetitive stuff. but as far as anything challenging i find the results disappointing based on the hype.

i could go on and on about AI and how much i don't like it. i would never think it's bad to play around with for fun. I've done this myself with ChatGPT. my real worry is jobs lost and poor quality output. there are already a bunch of junk AI books on Amazon put out by scammers trying to make money for nothing. i also worry that companies with greedy leadership who don't care about quality and want to increase profit will try to replace employees with AI. there isn't much regulation yet and the people who run the leading AI tech companies suck and don't care about people. TBH i think they're a bit out of their minds. replacing humans with bad AI is bad for employees who lose jobs and customers who have to deal with a poor experience. there isn't a vision from these people to make the world better. to have everyone do less work and live in abundance. in it's current form AI is a hype bubble and a grift. like crypto and the metaverse was/still is. even the commercials promoting AI in software and devices don't really say much about what the AI will actually do. it's just "hey we added AI. give us more money!"

and yeah the "deepfake" videos terrify me. especially since so many people believe that if it's on video, it must be real. also i've read that a lot of it is porn so that's horrifying for people who may end up in a video without consent.

like most stuff that comes out of big tech, AI could be cool. but not in it's current form.



I am actually less scared of AI (do we need a separate thread to discuss this btw? It's interesting) as in real AI (actual artificial intelligence) than human-developed and human-taught AI, because that enables it to be influenced and manipulated in dangerous ways.

As a writer, too, I take pride in what I write being my own. I don't want someone to look at what I write and think I used something to help me write it. I don't know if that is the same as going out to artistically collaborate using the tech and acknowledging it as a tool...I think it may be different with words than with pictures. Not sure how...just how it feels to me.

I DO worry about it being used by people in universities to cheat assignments going forward. I don't want a doctor who didn't write their own assignments or do their own research to get their medical degree. :/ Plagiarism is already a big issue in universities (speaking as a former GTA, it's something we were trained to deal with). But AI is like having another person write an essay for you, only cheaper :/ I think this could be a problem in the future. Dr Google and all that :/

And I worry about how AI already represents neurodiversity - I saw someone do some research on this, and AI presents an autistic person as a gloomy, skinny, white male in his twenties. It's not helpful.

You both bring up really good points.

@Beth, you're right about it being a nifty new thing that people and companies are exploiting, like crypto, etc. Maybe that's one of my things, because I do think AI and robotics will help humans reach a utopia one day. But for right now it's mostly people using it for fun or profit. And I am kind of sick of all my programs adding AI stuff to them. Photoshop has done it and it's very pushy about using it. And YouTube, when I go to upload content, it's AI this and AI that.

@Taffeta, I sometimes forget there are people who would use it for bad things. You brought up things I've never even thought of, like the doctor thing. That IS worrisome. And yeah, more complex concepts that aren't widely understood tend to be stereotyped, that's the fault of the human coders. It's like that robot from Dubai. I forget her name. In interviews she has said things like robots will take over and humanity is doomed, etc. But again that's on the people who programmed her. ChatGPT is getting like that too. When it was first released it was much more interesting to talk to before people sent it down the meme hole. Now it's like chatting to someone on the site that shall not be named.

Hopefully, somewhere there is a dedicated group of people working on an AI that won't be polluted with awful things. That day might be a long way off still.

So, yes, I agree that it can be dangerous and maybe we aren't ready for it yet. I still find it fascinating.
« Last Edit: Yesterday at 02:07:50 PM by BlackCurtains »
visitors can't see pics , please register or login

Instagram | Facebook
Commissions are Closed
Avatar and sig by Mione (offsite)

Offline Ponyfan

  • Trade Count: (+39)
  • MIB Licensing Show Pinkie Pie
  • ******
  • Posts: 13782
    • View Profile
Re: AI Discussion
« Reply #1 on: Yesterday at 03:27:08 PM »
This probably isn't what you were thinking of when you started talking about AI, but someone on Facebook has trained an AI chatbot to talk like Sailor Moon/Usagi.  This person told the chatbot that Sailor Moon only existed in a parallel universe that wasn't connected to our own.  The chatbot acted surprised to find that out.


Ponyfan
visitors can't see pics , please register or login


Thank you SDS for my avatar and sig

Online lovesbabysquirmy

  • Trade Count: (+60)
  • Colombian Baby Pony
  • ******
  • Posts: 17105
  • Gender: Female
  • ~never too old for ponies~
    • View Profile
Re: AI Discussion
« Reply #2 on: Yesterday at 04:12:48 PM »
I'm just going to put this piece of journalism here:

https://twitter.com/LaurieSegall/status/1766278068002299975

I think this is abhorrent.  if it wasn't for the fact that she agreed to the experiment at a tech conference, I would think these "tech dudes" should be in jail for defamation, harassment, and creation of exploitative intimate deepfake media.

I mean seriously, people can have public opinions about Mark Zuckerberg, but he is a MARRIED man with children.  This experiment is beyond acceptable - what if "tech dudes" or their friends decided to do this to any one of us?  What if anything we said that was disagreed upon by another person decided to tell the AI to do this kind of stuff, including accusing the person of illegal sexual content, "just, like, as a fictional story" and get around the simple restraints that are placed upon the AI machines, and the next day, you have no job, you are threatened with prison, your reputation is destroyed, your family thinks the worst of you, on and on I could go.

Having little lazy shortcuts to get things done so we don't have to employ people as a civilization is straight-up DANGEROUS.
« Last Edit: Yesterday at 04:14:43 PM by lovesbabysquirmy »
The Bad Trader List
visitors can't see pics , please register or login
visitors can't see pics , please register or login
 visitors can't see pics , please register or login

<3 Sig Art: SquarePeg[current avatar] Vanilla Virus, Sweetpop, Thimble, SourdoughStomper, LyrePony, Tropical Sunset, PureNightShade, Ellis1342,KissedByThunder, Shaiyeh <3

Online Beth3346

  • Trade Count: (+91)
  • G3 Prototype Pony
  • *****
  • Posts: 2581
  • Gender: Female
  • she/her/hers - Baby Mint Dreams by Lady Satine
    • View Profile
    • My Little Pony Collection
Re: AI Discussion
« Reply #3 on: Yesterday at 04:41:29 PM »
This probably isn't what you were thinking of when you started talking about AI, but someone on Facebook has trained an AI chatbot to talk like Sailor Moon/Usagi.  This person told the chatbot that Sailor Moon only existed in a parallel universe that wasn't connected to our own.  The chatbot acted surprised to find that out.


Ponyfan

lol i think those are kind of fun

Squirmy, that's awful. what if a someone did that to a regular person and a jealous partner saw it. that kind of thing should be illegal IMO. probably falls under free speech though. ugh.

i'm not at all worried about the robots taking over, Terminator scenarios. i don't think technology is there yet or ever will be. AGI (Artificial general intelligence) is kind of the holy grail for AI people these days. S*m Altm*n, the ChatGPT guy keeps saying it's just around the corner. gives M*sk vibes IMO. we don't have safe self drive cars either. there's some really weird stuff these people are into. eccentric billionaires/millionaires are a trope for a reason i guess. it's a rabbit hole i'm not going down here. we need much better people building this stuff if we want a utopia for everyone. right now it's likely to be utopia for ultra rich people and a whole lot of unemployment.

i'm still fascinated with AI but i worry about it in the hands of government and corporations. drones and facial recognition making decisions without human intervention. those robot dogs with guns also give me nightmares.

i think AI needs to be highly regulated and developed with a lot of input from a diverse group of people. it really bothers me that people get to roll out all of this new stuff that affects us all and we don't really have much of a say. we can decide not to use it but don't have much control over what people with more power do with tech. it seems like we need like a worldwide vote or something :D i know that's impractical and would probably have it's own issues. we all know voting doesn't always produce outcomes that protect underrepresented people. I also think it shouldn't be replacing people's jobs unless there is some kind of universal income and real social safety nets. like guaranteed housing, food, healthcare. with all of the current austerity trends i don't believe we will get that even if a lot of people lose their incomes. we can't have a utopia when governments don't see making sure everyone's basic needs are taken care of as a human right.

scenarios like Taffeta mentioned worry me too. An AI trained on content that contains negative stereotypes is going to output negative stereotypes. we have enough negative stereotypes about neurodivergent people in pop culture already. also more chatbots, more layers before you actually get to talk to a human don't sound great. especially in highly skilled fields like law and medicine. there are already a lot of worries about AI's hallucinating and just making stuff up. like an AI for lawyers could just make up a law if it's trained on bad data.

i think we really could all have better lives now with existing technology. better mass transit. or any mass transit if you live where i live :D less spending on military toys and more money on making people's lives better. it's depressing that we have so much spending on jets that don't work right and kids in the US have lunch debt. solving that would be a drop in the bucket for the US budget. a better social safety net would be good too.
Current Grails - Dutchies, Princess Variants PM me if you have any for sale - Want List

Offline ChocolateStarfire

  • Trade Count: (+21)
  • Sweet Scoops Pony w/Charm
  • *****
  • Posts: 2421
  • Gender: Female
  • I Love Ponies...Always!
    • View Profile
    • Enchanting Customs
Re: AI Discussion
« Reply #4 on: Yesterday at 06:28:11 PM »
I've seen the Terminator films, 2001-A Space Odyssey (and read the books!) and I, Robot, in addition to the Spielberg film AI. I prefer we stop now while we are ahead, because we will have Skynet take over. D:
My MLP customs
My eBay Sales
Brightbow Avvy-SquarePeg :lovey:

Offline Taffeta

  • Trade Count: (+62)
  • Colombian Baby Pony
  • ******
  • Posts: 16065
  • Gender: Female
  • UK Pony, Jem and Mediaeval Japanese obsessive :D
    • View Profile
    • The My Little Pony Scrapbook
Re: AI Discussion
« Reply #5 on: Yesterday at 06:54:21 PM »
A quick warning at the top of my post that the links to articles relate to things that may be political or may upset some people due to their content.

I watched a documentary a while back about deepfakes being used to create questionable content of prominent figures and then being used as blackmail material. I don't want to get into the details of it because it's not safe for this site, but I'm sure you can imagine.

So long as the internet and society discriminates against any group, groupthink human taught AI will also promote those stereotypes and close those doors.

I'm wary of getting into too much political stuff, but one example about how technology failed people is the Horizon scandal over here, where many, many post office workers were accused (and many convicted and jailed) for embezzlement from the Post Office, when in fact there were faulty software algorithms involved.  General overview here for anyone unfamiliar:
https://www.bbc.co.uk/news/business-56718036

What is the most concerning about that, aside the human cost, is that people did know there were issues and we're finding out now how much of that got swept aside because the tech couldn't possibly be the problem. Far easier to believe lots of very normal post office workers from all over the country were all committing fraud at the same time...right?

Despite issues like this, the government are trying to get a last minute amendment through the online safety bill which will allow the Department for Work and Pensions to monitor bank accounts of benefit claimants, their friends and family to 'prevent fraud'.
https://www.disabilityrightsuk.org/news/dr-uk-says-scrap-dwp-plans-use-ai-scan-benefit-claimants-bank-accounts

In short, they are preparing to spend more money and rely more on tech and AI to do stuff rather than human beings, despite the big red flag of the Horizon scandal.

The DWP is apparently also monitoring social media accounts already using AI (at great public expense) to 'look for fraud'.

Not that I think fraud is ok or should be let go, but...it's like smashing up a whole house because you need new windows. And in the case of disability benefit, there isn't really a case for account monitoring, because the benefits work differently from income support benefits.

Moreover, there's so little fraud from claimants of disability benefit (0.2% of claims I think is the figure). It's a considerable amount lower than the amount of error in the DWP's decision making and benefit awarding process (they quote it at 4% overall).

This one is not an 'in the future' thing. It's literally right down the road, and honestly, it's terrifying.
visitors can't see pics , please register or login
|夏草やつわものどもが夢の跡|

Offline cowboyopossum

  • Trade Count: (0)
  • Rainbow Curl Pony
  • ***
  • Posts: 447
  • Gender: Male
  • I cannot contain my silliness | he/him
    • View Profile
Re: AI Discussion
« Reply #6 on: Today at 05:54:32 AM »
This is what I am saying, Taffeta. It can be used for such evil things... Especially in the hands of criminals.
visitors can't see pics , please register or login
visitors can't see pics , please register or login

Thank you Squarepeg and Wizard for my profile and sig!

Online BlackCurtains

  • Mad Scientist
  • Moderator
  • Trade Count: (+69)
  • MIB Licensing Show Pinkie Pie
  • ***
  • Posts: 10736
  • Orange ponies are the best.
    • View Profile
    • BlackCurtains Crafts
Re: AI Discussion
« Reply #7 on: Today at 08:27:27 AM »
@LBS and Taffeta, I guess I need to blame naivety on my part here. I really had no idea AI was used in ways like that.

LBS, that is incredibly disturbing. I admit I don't know a lot about deepfake videos. I saw one of Tom Cruise once, saying weird things, and it looked absolutely real. And since he does say and do weird things if I hadn't already known it was fake, I would have believed it.

visitors can't see pics , please register or login

Instagram | Facebook
Commissions are Closed
Avatar and sig by Mione (offsite)

 

SimplePortal 2.3.5 © 2008-2012, SimplePortal