OpenAI Debacle - Update (03/01): Elmo sues OpenAI for...uhhhh..."reasons"

[Jason]

50 minutes ago, Kal-El814 said:

Discord is great for small communities, chat with the same, and with certain games / industries, troubleshooting or assistance. Anything beyond that it's fucking terrible, and the fact that a lot of people use it as de facto wiki / website replacements is absolute shit. I appreciate that it's easier to build and maintain for a lot of people, but it's still awful.

 

My main problem with Discord is that it's absolutely fucking terrible to manage once you're in more than a couple of servers, even if each individual server is a small manageable size by itself. 

 

I routinely miss DMs because if you have it open on a computer it won't give you push notifications on your phone, which they apparently consider a feature not a bug, and I usually have a Discord tab open somewhere. And then it's not obvious when I go in that I have an unread DM because I've wound up in so many servers they don't fit vertically on screen even on my phone, which pushes the unread DM indicator off screen. 

 

Meanwhile if you're on desktop, it beeps and flashes at you that you have a notification but doesn't actually indicate which server/channel it's coming from. It does have a notification tray...with an icon that absolutely does not suggest that it's a notification tray, so I was using Discord for years before I found out about that. But it's useless because as far as I can tell it just piles up notifications even from servers/channels you've muted notifications, and it groups them by server instead of just being reverse chronological order, so you can't readily tell which one the beeping and flashing was about. 

 

And of course you have to mute notifications pretty much the second you join a server or else your phone is immediately going to get flooded with notifications if the server is even remotely active. 

 

One of my hobbies used to use Cisco Spark/Webex/Teams (it's been through a bunch of rebrandings) and that was actually great because you had a unitary list of chats, and any chat with new activity was automatically floated to the top of your chat list. When you went in, it'd reliably take you to the new content location like we have on the boards here. So it was very easy to keep up with all the chats without missing anything. But we abandoned it for Discord and one of the reasons was that getting people onto the Cisco platform required going through one guy who works at Cisco (the reason we were on it in the first place, solved a problem for the hobby and I think Cisco got free testing out of it basically) and he's largely checked out of the hobby. 

[Jason]

tl;dr you can tell Discord has become a bloated unwieldy mess that they just keep piling new features onto when it was originally intended primarily as a TeamSpeak kind of application

[Commissar SFLUFAN]

6 minutes ago, Jason said:

tl;dr you can tell Discord has become a bloated unwieldy mess that they just keep piling new features onto when it was originally intended primarily as a TeamSpeak kind of application

 

Gotta justify that VC money somehow!

[Kal-El814]

11 minutes ago, Jason said:

 

My main problem with Discord is that it's absolutely fucking terrible to manage once you're in more than a couple of servers, even if each individual server is a small manageable size by itself. 

 

I routinely miss DMs because if you have it open on a computer it won't give you push notifications on your phone, which they apparently consider a feature not a bug, and I usually have a Discord tab open somewhere. And then it's not obvious when I go in that I have an unread DM because I've wound up in so many servers they don't fit vertically on screen even on my phone, which pushes the unread DM indicator off screen. 

 

Discord DMs are might as well exist in a black hole.

[Keyser_Soze]

On 11/27/2023 at 12:34 PM, Kal-El814 said:

Tired: WE'RE ALL GONNA BE PAPERCLIPS, SKYNET, REJECT MODERNITY

 

Wired:

 

Sports Illustrated Published Articles by Fake, AI-Generated Writers

FUTURISM.COM

---

Sports Illustrated was publishing articles under seemingly fake bylines. We asked their owner about it — and they deleted everything.

 

 

Sports Illustrated publisher fires CEO Ross Levinsohn after AI scandal

WWW.BBC.COM

---

The head of the magazine was fired not long after Sports Illustrated reportedly published AI-written articles.

 

[unogueen]

On 11/27/2023 at 6:06 PM, Commissar SFLUFAN said:

 

Gotta justify that VC money somehow!

Infinite profits means infinite productivity. No matter how asinine.

[ort]

I don’t have a lot to add besides piling on the total shit show that is discord. 
 

I don’t really use it much, but there are a few communities for games I play where it is the only source of information… and woof, the UI feels like it was designed by a million monkeys with a million typewriters. It’s so bad I’m not even sure where to begin to criticize it… but as a super casual user without a lot of experience with it, I find everything about to be completely unintuitive, baffling and user hostile. 

[elbobo]

Just in time for the election openAI has made a video creating AI

 

It is insane we have gone from DALLE 2 to this in less than 2 years

 

Sora: Creating video from text

OPENAI.COM

 

 

[CitizenVectron]

3 hours ago, elbobo said:

Just in time for the election openAI has made a video creating AI

 

It is insane we have gone from DALLE 2 to this in less than 2 years

 

Sora: Creating video from text

OPENAI.COM

 

 

 

Prompt: "greasy nerd posting on the day one patch internet forums on a computer about AI "  

(yes, this is actually generated by Midjourney right now...it's crazy how good these are)

[Commissar SFLUFAN]

We are so very, very screwed.

[Xbob42]

50 minutes ago, CitizenVectron said:

 

Prompt: "greasy nerd posting on the day one patch internet forums on a computer about AI "  

(yes, this is actually generated by Midjourney right now...it's crazy how good these are)

That dude's not greasy at all! I can get ten times greasier than this!

[Keyser_Soze]

Greasy nerd has a very chunky keyboard.

[ThreePi]

2 hours ago, Commissar SFLUFAN said:

We are so very, very screwed.

 

Are we? Perhaps this just leads to a scenario where the media/internet is so flooded with cheap AI generated nonsense that people will start to inherently distrust anything on social media and bring about a return of reputable human-curated content. On a small scale, looking at stuff like the Wirecutter being one of the few bastions of seemingly honest consumer product reviews.

[Ricofoley]

2 hours ago, CitizenVectron said:

 

Prompt: "greasy nerd posting on the day one patch internet forums on a computer about AI "  

(yes, this is actually generated by Midjourney right now...it's crazy how good these are)

[legend]

10 hours ago, ThreePi said:

 

Are we? Perhaps this just leads to a scenario where the media/internet is so flooded with cheap AI generated nonsense that people will start to inherently distrust anything on social media and bring about a return of reputable human-curated content. On a small scale, looking at stuff like the Wirecutter being one of the few bastions of seemingly honest consumer product reviews.

 

Humanity will eventually be able to navigate this. But the the first few years will have growing pains and this is coming out right during a presidential election with one of the most corrupt and incompetent candidates the US has seen. So it's not great!

[chakoo]

The sora stuff is impressive but I think people are over blowing how good it is. There is tell tell signs in all the videos it was done with AI. I'm a strong believer in the 80/20 % where its quick to get to that impressive 80% but it will take a really long time to ever finish that remaining 20%. 

[legend]

Just now, chakoo said:

The sora stuff is impressive but I think people are over blowing how good it is. There is tell tell signs in all the videos it was done with AI. I'm a strong believer in the 80/20 % where its quick to get to that impressive 80% but it will take a really long time to ever finish that remaining 20%. 

 

For the malicious uses, I don't think that final 20% matters. Too many people are already willing to believe more blatantly obvious propaganda.

[chakoo]

1 minute ago, legend said:

For the malicious uses, I don't think that final 20% matters. Too many people are already willing to believe more blatantly obvious propaganda.

 

I normally don't factor those people into my thoughts because they don't need this in the first place for their propaganda. We've learn that just making shit up and repeating it is all you need.

[ThreePi]

38 minutes ago, legend said:

 

Humanity will eventually be able to navigate this. But the the first few years will have growing pains and this is coming out right during a presidential election with one of the most corrupt and incompetent candidates the US has seen. So it's not great!

 

Yeah, initially I think its going to be a mess. But I don't think it'll mature enough to strongly influence this election. It seems easy enough to spot that anyone trying any funny business will get called out right away. Going into the future I think it'll just be like how no one trusts Google searches anymore because its predominantly slop that's been SEO'd to hell.

[elbobo]

2 hours ago, chakoo said:

The sora stuff is impressive but I think people are over blowing how good it is. There is tell tell signs in all the videos it was done with AI. I'm a strong believer in the 80/20 % where its quick to get to that impressive 80% but it will take a really long time to ever finish that remaining 20%. 

 

This was the pinnacle of AI video 9 months ago

 

 

Now we have this

 

 

 

Less than year to go from silly gibberish to pretty close to passable. Where the hell is it going to be in 5 years.

 

 

[legend]

31 minutes ago, chakoo said:

 

I normally don't factor those people into my thoughts because they don't need this in the first place for their propaganda. We've learn that just making shit up and repeating it is all you need.

 

I'm not worried about it capturing more people than the crazies, I'm worried about it giving the crazies more reason to do something insane.

 

4 minutes ago, ThreePi said:

 

Yeah, initially I think its going to be a mess. But I don't think it'll mature enough to strongly influence this election. It seems easy enough to spot that anyone trying any funny business will get called out right away. Going into the future I think it'll just be like how no one trusts Google searches anymore because its predominantly slop that's been SEO'd to hell.

 

I sure do hope you're right about it not affecting the election!

[chakoo]

38 minutes ago, elbobo said:

Less than year to from silly gibberish to pretty close to passable. Where the hell is it going to be in 5 years.

 

I've been messing with Generative AI (diffusion) models for quite some time and even GANs for a while before that. It looks impressive on the surface but there is still a lot more that is needed and that remaining 20% is going to take some time.

[Jason]

Air Canada must honor refund policy invented by airline’s chatbot

ARSTECHNICA.COM

---

Air Canada appears to have quietly killed its costly chatbot support.

 

[Reputator]

Just starting to see some of these videos getting posted. Yeah there are flaws but ot the untrained eye, ie most people probably wouldn't be able to tell. And if we got this far this fast, I don't see those flaws lasting very long.

 

And yeah frankly it's a bit terrifying. I don't know that humanity collectively has the intelligence to cope with this new reality.

[Ominous]

1 hour ago, Reputator said:

Just starting to see some of these videos getting posted. Yeah there are flaws but ot the untrained eye, ie most people probably wouldn't be able to tell. And if we got this far this fast, I don't see those flaws lasting very long.

 

And yeah frankly it's a bit terrifying. I don't know that humanity collectively has the intelligence to cope with this new reality.

Just run it through a low rez filter and post it to IG or TikTok and on a phone most people would buy it 100%. 

[GeneticBlueprint]

 

Thanks I hate it

[mclumber1]

22 minutes ago, GeneticBlueprint said:

 

Thanks I hate it

 

I dunno.  I'd love a movie/miniseries version of the World War Z book.  The Brad Pitt movie didn't do the book justice.  

[Keyser_Soze]

 

[unogueen]

18 hours ago, mclumber1 said:

 

I dunno.  I'd love a movie/miniseries version of the World War Z book.  The Brad Pitt movie didn't do the book justice.  

Actual movies require storyboards, lighting tests, set design, etc. that seems unexplained how Sora plans to fulfill with an actual narrative.

[unogueen]

If this thing is scraping my behaviour over time everything would be tits and ass for me, no matter what I ask for.

[Signifyin(g)Monkey]

21 hours ago, Reputator said:

Just starting to see some of these videos getting posted. Yeah there are flaws but ot the untrained eye, ie most people probably wouldn't be able to tell. And if we got this far this fast, I don't see those flaws lasting very long.

 

And yeah frankly it's a bit terrifying. I don't know that humanity collectively has the intelligence to cope with this new reality.

The most interesting aspect of the situation to me is that there’s a paradox nested in the automation AI provides that was not present in prior technologies that caused large waves of disemployment via the automation of labor.

 

Namely, that AIs require the labor they are built to automate.  To take a simple example, an image generator AI can automate the creation of images, but it also can’t work without lots of people providing it lots of source images.  An image generator, in a sense, can’t completely automate image generation without killing itself.

 

Cars, by contrast, did not need horse-buggy drivers to work.

 

So the dynamics of the coming labor disputes and legal battles over AI are going to be unlike anything we’ve really ever seen.

 

We live in interesting times.

[unogueen]

3 minutes ago, Signifyin(g)Monkey said:

The most interesting aspect of the situation to me is that there’s a paradox nested in the automation AI provides that was not present in prior technologies that caused large waves of disemployment via the automation of labor.

 

Namely, that AIs require the labor they are built to automate.  To take a simple example, an image generator AI can automate the creation of images, but it also can’t work without lots of people providing it lots of source images.  An image generator, in a sense, can’t completely automate image generation without killing itself.

 

Cars, by contrast, did not need horse-buggy drivers to work.

 

So the dynamics of the coming labor disputes and legal battles over AI are going to be unlike anything we’ve really ever seen.

 

We live in interesting times.

Assuming that the substandard isn't already good enough. Synthetic data solves that no?

[Signifyin(g)Monkey]

14 minutes ago, unogueen said:

Assuming that the substandard isn't already good enough. Synthetic data solves that no?

Not really, because an AI that just recycles pre-existing data will eventually lose the capacity to adapt to new changes in its domain. (newly invented words/images/sounds/etc.)

 

That’s not necessarily the case when it’s working with a closed system with fixed rules, like chess—but it’s certainly the case when it’s working with continuously evolving systems like human language or visual design.

[unogueen]

20 minutes ago, Signifyin(g)Monkey said:

Not really, because an AI that just recycles pre-existing data will eventually lose the capacity to adapt to new changes in its domain. (newly invented words/images/sounds/etc.)

 

That’s not necessarily the case when it’s working with a closed system with fixed rules, like chess—but it’s certainly the case when it’s working with continuously evolving systems like human language or visual design.

Eventually. But this is a one trick pony with a lot of power, money, and layoffs behind it. How much does it tear apart in a sprint and how do we reconcile the consequences.

[finaljedi]

On 2/15/2024 at 9:59 PM, Xbob42 said:

That dude's not greasy at all! I can get ten times greasier than this!

 

Microsoft Copilot is here for you, same prompt

 

 

19 hours ago, GeneticBlueprint said:

 

Thanks I hate it

 

This reminds me of those crypto bros that tried to sell NFTs in games as "Imagine you don't have to find a gun in the game, you could buy it, and then sell it when you're done with it to another player".  Those were people who don't play games and I'm pretty sure these people don't read a whole lot of books.  This does answer the burning question I've always had:  Who the fuck reads those terrible AI generated Kindle books people shit onto Amazon?