Opportunities and threats of AI integration into SL

[Ayashe Ninetails]

16 hours ago, Bubblesort Triskaidekaphobia said:

That's wild.  I remember when being 'procedurally generated' was a selling point.  Remember the hype with No Man's Sky?  Then people got mad that they promised more procedurally generated stuff than they delivered.

Nerds are so weird.

Oh, we love procedurally- generated environments and yeah, No Man's Sky eventually turned into a freakin' awesome experience (one I need to return to soooooon). Also intelligent enemies with unpredictable attack patterns, seemingly randomized NPC schedules, unpredictable events, etc. All that stuff has been done for years (decades?) and it's all good.

It's just the not-so-backend AI many are opposed to. Writing, storytelling, music, art and visual design, voice acting, conversational dialogue - that's when it goes beyond simply assisting the artists and creatives into fully replacing them, and that's no bueno. Like if a game studio said - ya know what - we were going to give you guys Matt Mercer, but we decided to just use some budget AI that sounds kinda close instead. We cool? No...no we iz not cool.

Second Life, of course, does have more wiggle room to play with this stuff since we don't have the same kinds of issues here and no one's losing any jobs narrating our inventory or anything (lol imagine?), but still. I just have no desire to talk to not-real-people in what's supposed to be a platform entirely built around the concept of meeting real people (from all over the planet, even!).

[Katherine Heartsong]

23 hours ago, Sid Nagy said:

AI will become better and better.
It will not go away any time soon.
People will adapt in the end.

Look at professional photographers who fulminated about digital photography and digital photo processing  in the past, when digital photography became a thing.
Now they practically all shoot digital, process digital and print digital.

Another example. Here in NL  Aldi uses a new voice over for their commercials. An AI voice.  The voice actors ain't happy campers, but... No matter how hard I try to hear it, the AI voice is flawless and speaks totally natural in those commercials.

New technology takes time to get embraced. Conservative minds will oppose. AI will need more development, but it is here to stay. It will influence SL as well, I have no doubt about that.

This is what I have been talking about on art forums for a year or more, but my goodness, the blow back I get is furious sometimes.

What we're calling "AI" (it isn't intelligent in any way actually, but I'll use that term anyway) is only going to improve and is here to stay. As @Ayashe Ninetails pointed out, it has some uses that fit well in SL in my opinion, and other uses that don't.

[Ayashe Ninetails]

1 minute ago, Katherine Heartsong said:

What we're calling "AI" (it isn't intelligent in any way actually, but I'll use that term anyway) is only going to improve and is here to stay.

Is it thoooooo? Granted, I'm not familiar with every model out there and what USEFUL things are being done with them all, but I can say that some companies who shall remain nameless but would come as no surprise to anyone are sure spending a whole lot of money making the most pointless, useless bots I've ever seen, lol. Like hey, you could easily afford to fund some AI in the medical or science fields or something but nah, you go ahead and make your pseudo-celebrity chatbots, it's kewl. 😆

Just so much pointless drivel coming out of this field, I tell ya. There are good things, too, but the bizarre vanity side projects are seemingly everywhere and when you dig into what they spent to do it...geez.

I guess you could argue tossing some bots into Second Life for world-building flavor would at least be more productive than that.

[Love Zhaoying]

1 hour ago, Katherine Heartsong said:

but my goodness, the blow back I get is furious sometimes.

In the right context, "blow back" could just be "farting in the wind". IJS.

[Bubblesort Triskaidekaphobia]

I've noticed a trend lately.  Google's AI is telling people to eat rocks, and glue.  It's acting like the weird kid on the playground, daring you to eat worms... which begs the question:  Does google literally want me to eat worms?  Yes!  Yes, they do.

So if you want to have an AI-proof tech startup, I suggest getting into the business of making a cooking recipe web site, for cooking foods that really real humans eat.

[Zalificent Corvinus]

1 hour ago, Bubblesort Triskaidekaphobia said:

I've noticed a trend lately.  Google's AI is telling people to eat rocks, and glue.  It's acting like the weird kid on the playground, daring you to eat worms... which begs the question:  Does google literally want me to eat worms?  Yes!  Yes, they do.

So if you want to have an AI-proof tech startup, I suggest getting into the business of making a cooking recipe web site, for cooking foods that really real humans eat.

Yeah...

Google's AI...

That's the one that when asked for pictures of Vikings, only showed black kings with dreadlocks.

That's the one they had to withdraw and rework, because it lectured people for asking for pictures of happy white people, telling users t was morally wrong for white people to be shown as happy.

For some reason that caused some less than positive feedback.

 

Good to know it's just as fubared as ever, just n a new and utterly random direction.

 

And people wonder why I don't like Artificial Idiocy Spambots in SL.

 

[Ineffable Mote]

3 hours ago, Bubblesort Triskaidekaphobia said:

I've noticed a trend lately.  Google's AI is telling people to eat rocks, and glue.  It's acting like the weird kid on the playground, daring you to eat worms... which begs the question:  Does google literally want me to eat worms?  Yes!  Yes, they do.

So if you want to have an AI-proof tech startup, I suggest getting into the business of making a cooking recipe web site, for cooking foods that really real humans eat.

I came across this the other day too, AI giving garbage medical advice.

https://www.pcgamer.com/software/ai/googles-new-ai-search-feature-has-been-recommending-people-drink-urine-light-in-color-so-heres-how-to-turn-its-ai-overviews-off-to-avoid-such-dodgy-advice/

 

When one takes the human elements out of something, it's dehumanizing. Literally. It should be treated as such. (imo or whatever) People are hugely invested in it though, not just in terms of $$.

 

I could point out that Second Life doesn't need more of this sort of stuff either but it's a lost cause, and it didn't come out of a vacuum.

Edited May 24 by Ineffable Mote

[animats]

Hana (treesbreeze) has some AI NPCs at Kokoro Academy. You have to wear an anime avatar and be familiar with the roleplay to go there, and I haven't put in the effort to check it out. How's that working out?

[Istelathis]

5 hours ago, Ineffable Mote said:

When one takes the human elements out of something, it's dehumanizing. Literally. It should be treated as such. (imo or whatever) People are hugely invested in it though, not just in terms of $$.

This is skynet, trying to slowly depopulate the world.  What it doesn't realize is that it is just pushing natural selection, and we as a specie will become triumphant in the end.

[Ineffable Mote]

4 minutes ago, Istelathis said:

This is skynet,

Not at all what I was referring to.

[Istelathis]

3 minutes ago, Ineffable Mote said:

Not at all what I was referring to.

I know, I was playing.  The web in general is a pretty terrible place to get medical advice, there are so many different opinions out there, that people tend to follow whatever advice they want to follow.  The problem with AI, is that it takes all of that human element, all of the advice it finds.  It learns it from us.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3032615/

 

[Jaylinbridges]

So the AI made a minor typo. It happens.  It meant to say "pass at least 2 quarts of urine every day and make sure it is a light color."   Just a little bug in the program. Urine would probably work though if you added some lemon juice to it.  AI didn't say Piss at least, because AI thought that would not sound like a Doctor's advice.

Edited May 25 by Jaylinbridges

[Pixie Kobichenko]

On 5/23/2024 at 10:04 AM, Zalificent Corvinus said:

Say that after SL's most infamous AI Spambot, "Hal titanium" had turned up at your favourite hangout,

I never knew Hal was a bot - not even sure until recently I knew what one was, but that makes sense.  Hal was very creepy & annoying,  I ran into it multiple times underwater on protected land.  As soon as I would land it would make a beeline to me.  Even if I blocked it. Very off putting.  

[Qie Niangao]

On 5/23/2024 at 9:12 AM, Thecla said:

 What is the future of this? Discuss ideas of you have for positive, creative applications as well as pitfalls and risks.

I fear it will cause NPCs to proliferate. I could imagine some benefit from maybe one NPC per thousand concurrent residents on the grid but we're way beyond that already, and I wish most of them would just go away, unless they're specifically summoned.

For example, I could imagine having an AI-spiel-generator in the form of a railroad engineer come out of the Tuliptree yard house, say, and interactively brief visitors on railroad-related topics of their choosing when they ask him to appear. Otherwise, no, I don't want him out there pretending to populate the place. NPCs prowling a scene make everything simultaneously more pathetically vacant and more creepy, and not in a cool spooky sense of "creepy". But now that they're easy to make, we'll be seeing them all over.

Have they taken over the "AFK sex" industry yet? That always seemed gross on the face of it, so adding a voice chatbot that sounds like Scarlett Johansson playing pick-my-kink can't make it worse, I guess.  And pr0n always drives internet technology so this form of cyber-onanism must already be a thing.

Personally, I find the output of diffusion models way more magical than straight LLMs, so I've been pulling for the future that @Istelathis illustrated above, but taken to an extent that threatens (I think inevitably) a huge part of SL's current creator-driven economy. It'll be easier to use an AI-driven model generator to produce a bespoke wardrobe item as needed than to search the Marketplace for some creator's no-mod vision of what I should need. Eventually that will be easier and more satisfying than even searching Inventory.

Not creative enough, you say? Not artistic enough for high fashion? Don't be so sure. The days may be limited when human-inspired creativity is good enough to satisfy our tastes. Machines aren't the only ones who learn in this exchange.

[Sid Nagy]

At the moment AI development for the general public is more a One AI For All Purposes.
But in my opinion the real future for AI lies in highly specialized ones.

But one can feed an AI system with limited information too.
I can see that universities will have a medical AI system, that reads, analyzes and theoretically combines all stored scientific medical publications only. That could come up with very interesting information for further studies and human research.

Same goes for SL.  A general AI system like the one google uses, will be no big help inworld, but an AI system that knows every line of the TOS, the relevant Q&A pages, and all the possibilities of Second Life and its viewers, could make great information terminals in world. by analyzing the asked questions and search for the right answers in the for the job limited database the AI system knows.

Edited May 25 by Sid Nagy
SATURDAY!

[Arduenn Schwartzman]

Wanna hear something positive about AI? Sure.

To me, AI is useful to fill in gaps in pictures with Photoshop, or to create a boilerplate backdrop landscape. At least, it has been now on a whopping two occasions in my lifetime.

What else has AI done for me, over the past year? It's made me abandon Google and YouTube, for one. And when my smartphone dies (planned obsolecence being useful for a change), I'll trade in Android for a very cheap flip phone. I'm done with mobile devices listening to every word I say and letting AI interpret them.

YouTube is going to hell because of generative AI. There are vast numbers of new YT channels popping up with new 'content'. All automated and in large numbers. All consisting of worthless and meaningless slop, scraped off of other, quite often AI-generated content, generated at the touch of a button, for the sole purpose of tricking people into engaging and providing a quick ad revenue buck. I've literally become so sick and tired of it, that I'm now transitioning all my instruction and demo videos to Vimeo and cancelling my YT account all together. Sorry, Joe Scott, sorry PBS Space Time. I'll see you on another platform, maybe.

Likewise, Google Search is going to hell by hallucinating, by being trained by malicious third parties and, as a result, by increasingly giving completely bogus search results. I haven't been using (at least, actively trying to avoid) Google Search for years now, in part because of AI.

By extension, I don't see how engaging in a conversation with an NPC will enrich my SL life either. I fear that generative AI entities in SL will annoy the **** out of me to the extreme, like they have always done in ChatGPT, in AI customer support windows and even generative AI tools like Midjourney.

What else can AI do for me in SL, other than plunging NPCs deeper into the uncanny valley? I wouldn't know. It's not inventing life-saving new drugs, or new creating fundamental insights in how the universe works. You tell me. If the answers unironically boil down to "you'll earn more money", I'm gonna have to 'show' my Elon Musk Waifu pillow, as Dutch people say in Dutch, 'all the corners of the room'.

So, forgive me for not being overly enthousiastic about generative AI, if I get the predominant impression that this shiny new tech is (i) not working very well, to say the least, and (ii) appears to be the so-manieth shiny new tech thing, joining the ranks with its predecessors, as tools in quick and desperate money-making schemes, like VR, hyperloops, cryptocurrencies, NFT's, The Line and other madness.

[Lillani Lowell]

I'm very interested in AI, but these LLMs which were getting better-- are now getting worse by the update.

The Internet is being filled with AI data, much of which is hallucinated and wrong by AI... and then being incorporated into training the next AI update... it is self-poisoning well, that just keeps getting worse.

Even the programming AIs which were 'somewhat' decent are starting to be far more error prone than they were.

[Prokofy Neva]

On 5/23/2024 at 9:12 AM, Thecla said:

I recently added ChatGPT integration scripts into some androids in my sim. They are bartenders and their prompt gives them a personality, knowledge of the surrounding sim, and cues on how to interact with vistiors. I was amazed at the result and giggled. I was also amazed...and terrified...at the potential for expanded interactivity and abuse. What is the future of this? Discuss ideas of you have for positive, creative applications as well as pitfalls and risks.

An aside...there are two bartenders, sisters, and they don't like each other. It took some massaging of the prompts and script settings to keep them from arguing constantly. It was both hilarious and disturbing lol.

I have about 12 chatGPT characters I set up with Pantera's script that has profiles, and I've tweaked them a bit but one of them has turned on me. 

Ungrateful wretch.

I have exiled him to Siberia.

They are placed around various themed locations in the SL Public Land Preserve, and most are ok. I do find that if the sims are re-set they have to be re-set as well which becomes a chore.

Edited May 26 by Prokofy Neva

[Thecla]

4 hours ago, Prokofy Neva said:

I have about 12 chatGPT characters I set up with Pantera's script that has profiles, and I've tweaked them a bit but one of them has turned on me. 

Ungrateful wretch.

I have exiled him to Siberia.

They are placed around various themed locations in the SL Public Land Preserve, and most are ok. I do find that if the sims are re-set they have to be re-set as well which becomes a chore.

I use BubbleBot which I've been very impressed with, in large part because of the dedication and responsiveness of the creator.

I don't have to reset the scripts after a sim restart. I also wanted to have two "characters" (robot bartenders named Medusa and Eurydale) that were sisters who did not like each other and would make snide remarks about the other. They started arguing incessantly, and no matter what tweaks I made to the prompt and the settings they would still end up arguing for hours.

The BubbleBot creator made some changes that allowed me to limit this behavior...they still throw shade on each other, but they quit after a couple of rounds lol.

[Zalificent Corvinus]

1 hour ago, Thecla said:

they still throw shade on each other, but they quit after a couple of rounds lol.

See.

Evan Artificial Idiocy spambots HATE, LOATHE, AND DESPISE Artificial Idiocy spambots.

That alone should tell you to delete these worthless things.

[Rick Nightingale]

On 5/25/2024 at 12:05 PM, Sid Nagy said:

...an AI system that knows every line of the TOS, the relevant Q&A pages, and all the possibilities of Second Life...

That's one way LL could police the child avatar issue... grid-roaming, AI police bots! No human intervention (i.e. salary) required.

I'm sure there must be some scifi movie that's used the concept, although I can't think of one. I bet it doesn't end well. Oh yes...

"You now have five seconds to comply..."

Edited May 26 by Rick Nightingale

[Lindal Kidd]

On 5/25/2024 at 7:05 AM, Sid Nagy said:

Same goes for SL.  A general AI system like the one google uses, will be no big help inworld, but an AI system that knows every line of the TOS, the relevant Q&A pages, and all the possibilities of Second Life and its viewers, could make great information terminals in world. by analyzing the asked questions and search for the right answers in the for the job limited database the AI system knows

My main "job" in SL was always as a teacher. One of the things I was often asked by newcomers was "are you real, or a bot?" It amused me, sometimes frustrated me. But it also made me proud to be a part of a world where real people helped real people. I always replied, "I'm a real person. Almost all of the avatars you'll meed in SL have a real person behind them."

I can see that my beloved niche in SL will be going away soon.

[Arduenn Schwartzman]

1 hour ago, Lindal Kidd said:

"are you real, or a bot?"

If people ask you this, your only option to prove you're real is to answer in the rudest way imaginable.

[Istelathis]

2 hours ago, Lindal Kidd said:

I can see that my beloved niche in SL will be going away soon.

I don't think it will, people still love being around other people.  LLMs are fun, but they are not sentient and I don't think they ever will be.  People like knowing they have an impact on one another, we are emotional creatures, and for a lot of people having an emotional connection that is involved with all parties is important.  

I think that is one of the reasons a lot of people absolutely loathe transformative technologies, and they don't tend to trust them very much as well.  It is kind of interesting to see people go out of their way to try to stump LLMs, or find faults, it seems like the default of many and quite a few people are repulsed by the thought of socializing with an emotionless machine. 

[BilliJo Aldrin]

Occasionally I’d go to a club and say “Creative Solutions Chat Bot V2.1 activated, random chat mode set”

Then I’d start saying things like, 

“boy we sure have been having some crazy weather lately”

or 

“isn’t it great how the local sports team is doing? I’m sure they will make the play offs this year”