So I asked the new Chatgpt-4 how Second Life can get more users...

[Scylla Rhiadra]

15 minutes ago, benchthis said:

This is disburbing to someone who wants to learn and is not being forced into learning. If being forced to learn anything is better than learning. When someone wants to learn and sees this as the norm it's disappointing.

Well, this is part of the debate. There are those who are worried primarily about the impact upon assessment of AI. Students are getting away with cheating, and therefore their GPA doesn't reflect their actual abilities.

And there are those who are more concerned about the pedagogical implications: researching and writing a paper, producing your own code or mathematical solution to something, etc., are exercises in learning by doing.

I would mostly count myself in the latter group. I don't see universities and colleges as primarily about credentialing: I view them as educational institutions. I don't especially care if a student ends up with a B rather than a C in one of my courses because they cheated: I care about the fact that they've not learned as a result. (Note: no one using ChatGPT is likely to earn an "A" grade.)

That's why I'm at least a little interested in discussions about how we might leverage AI for learning. AI isn't going anywhere, and it's only going to get more proficient. Simply trying to block it is a bit like telling students that they can't use the internet for research. It's stupid, and it's not going  to work. So, how can it be used productively, ethically, and creatively to assist in the learning process?

[Scylla Rhiadra]

49 minutes ago, Ayashe Ninetails said:

I also don't see a point in doing that. One of the best things about virtual worlds and the Internet in general is gaining exposure to differences - different thoughts, different outlooks, different perspectives, different experiences, different languages, etc.

Someone here quite recently -- I don't remember who or where -- suggested the utility of ChatGPT to produce sort of self-clones with whom one might spend one's "social time" in SL.

I had to resist the temptation to suggest that they just purchase a mirror.

Edited March 22, 2023 by Scylla Rhiadra

[Qie Niangao]

5 minutes ago, Scylla Rhiadra said:

(Note: no one using ChatGPT is likely to earn an "A" grade.)

True of GPT3. Maybe not GPT4 with its much larger model context windows (8000 to 32000 tokens vs 4000 before), much much more "RLHF" (Reinforcement Learning from Human Feedback) now administered by PhDs who evaluate and annotate productions (as opposed to Upwork and Mechanical Turk workers who served that function before).

No idea what any of this means for education, but I'll say this: Gartner consultants better be polishing their Turing Test skills.

[Love Zhaoying]

5 minutes ago, Scylla Rhiadra said:

Someone here quite recently -- I don't remember who or where -- suggested the utility of ChatGPT to produce sort of self-clones with whom one might spend one's "social time" in SL.

I had to resist the temptation to suggest that they just purchase a mirror.

..is Narcissus before SL had mirrors?

Edited March 22, 2023 by Love Zhaoying

[Scylla Rhiadra]

7 minutes ago, Qie Niangao said:

True of GPT3. Maybe not GPT4 with its much larger model context windows (8000 to 32000 tokens vs 4000 before), much much more "RLHF" (Reinforcement Learning from Human Feedback) now administered by PhDs who evaluate and annotate productions (as opposed to Upwork and Mechanical Turk workers who served that function before).

No idea what any of this means for education, but I'll say this: Gartner consultants better be polishing their Turing Test skills.

This may depend upon the discipline, and the individual instructor.

I award As on the basis of original thought. At the undergrad level, that doesn't, it's true, need to be an earth-shattering new insight, but it does require the visible exercise of thought and analysis.

In other words, it's not just about incorporating "more and better" examples of other people's ideas. And the process by which those insights are produced has to be clearly laid-out. Maybe that AI will be able to do something like that eventually, but I still very much doubt that it will ever be able to actually analyze a work of literature, or engage novel science. It can't innovate. It can only reproduce.

Edited March 22, 2023 by Scylla Rhiadra
Added emPHAsis

[Ayashe Ninetails]

7 minutes ago, Scylla Rhiadra said:

Someone here quite recently -- I don't remember who or where -- suggested the utility of ChatGPT to produce sort of self-clones with whom one might spend one's "social time" in SL.

I had to resist the temptation to suggest that they just purchase a mirror.

LMAO!

But seriously, perhaps someone might want to spend all day talking to themselves, but I try to make that less of a habit. 😄

Funny enough, of all the things I'd want an AI to help me with in SL, conversation is not one of them. 

[Istelathis]

Much like any tool, it will be used to enhance the capacity of our students.  At first, much like calculators or even the Internet, I imagine it will be frowned upon but later it will be embraced by educators to be used and actually be central to our education system in many cases.  Classes will likely be created specifically on how to use it.

That seems to be the trend of how it all works, and likely this will be the same.  Instead of raw natural talent that is the product of our own brain and its capacity to use the tools we currently employ, be it having been fortunate enough to have been born with such facilities that match our education requirements, it will be our capacity to produce works of art, philosophy, poetry, etc, through this new tool.

It may be considered cheating, but then why would it be considered cheating when one has superior skills and capacity in this new environment, while it is not considered cheating having skills and the capacity to perform better suited in the previous one?

Edited March 22, 2023 by Istelathis

[Scylla Rhiadra]

10 minutes ago, Istelathis said:

it will be our capacity to produce works of art, philosophy, poetry, etc, through this new tool

I honestly don't believe it will ever be able to do any of these things, at least in a meaningful way, even if it eventually acquires sentience, because its perspective, mode of thought, and responses to the external world are very different from our own.

Let me give you an example from the visual arts. I can teach AI to produce an image that abides by the conventionally accepted rules of visual composition: I can instruct it to use the Rule of Thirds or the Golden Ratio to place subjects within an image.

What I can't do is teach it how to break those rules meaningfully. The best artists do so all the time: they "misuse" lighting, composition, and colour deliberately to produce given effects -- as for instance in placing a subject at the margins of an image in order to suggest . . . marginality.

I do agree that it is likely to become a useful tool that will assist humans in producing these things -- in much the way that Photoshop assists in producing good, original art.

[Katherine Heartsong]

41 minutes ago, Scylla Rhiadra said:

What I can't do is teach it how to break those rules meaningfully. The best artists do so all the time: they "misuse" lighting, composition, and colour deliberately to produce given effects -- as for instance in placing a subject at the margins of an image in order to suggest . . . marginality

This is precisely correct. It's relatively easy to explain the basic rules of composition (and colour theory) to anyone, even if they do not consider themselves artists. I can even get people to create basic works of art using these two basic sets of rules. But to deliberately break rules to help create or amplify an emotional response in the viewer takes a knowledge of human feelings and emotions that can't be easily quantified by an AI/ML algorithm (at least today).

In my RL work I have evolved from doing very realistic art, then over decades to very abstracted and minimal work, and now I'm into a phase where I am working in a very simplified style ... not abstract, but simplified in terms of shapes, colour palette etc and often deliberately break rules around perspective and composition to enhance the piece because I understand, as a human, how my play will make you stop thinking and cause you to see something different and therefore provoke an emotional response.

[Rolig Loon]

So .... to be sure that all of this is "on topic" and relevant to SL .... I think AI technology may actually be quite helpful for doing some fundamental things that we might imagine having a smart macro for. After all, we already have some really simple-minded systems for writing scripts, and many texturing and building functions can be automated to produce things that are gently modified cookie cutter clones.  It's not unreasonable to imagine that someone might have an AI device that could take a command like "make me a doghouse with a pink roof" and generate a passable model.  As in RL, I think it will be a very long time -- if ever -- before creators need to worry about being replaced by talented AIs that can produce imaginative new things, but they could handle the drudgery in the meantime.

[Love Zhaoying]

Just now, Rolig Loon said:

So .... to be sure that all of this is "on topic" and relevant to SL .... I think AI technology may actually be quite helpful for doing some fundamental things that we might imagine having a smart macro for. After all, we already have some really simple-minded systems for writing scripts, and many texturing and building functions can be automated to produce things that are gently modified cookie cutter clones.  It's not unreasonable to imagine that someone might have an AI device that could take a command like "make me a doghouse with a pink roof" and generate a passable model.  As in RL, I think it will be a very long time -- if ever -- before creators need to worry about being replaced by talented AIs that can produce imaginative new things, but they could handle the drudgery in the meantime.

They (LL) could enhance "Boxy"! 

Example: "Boxy, how do I wear clothing?", "Boxy, how do I buy land?" etc.

[EliseAnne85]

49 minutes ago, Katherine Heartsong said:

cause you to see something different and therefore provoke an emotional response.

Regarding ChatGPT and evoking an emotional response.

I felt the poem I got from ChatGPT did lack emotion...my experiment below...

I got an account and began to work at getting some poetry and got three.   And then, I rewrote one not just for a test but because I thought the AI-generated poem lacked emotion, and I wanted the poem to really make someone want to come out and look at the stars.  The first one is the poem written by ChatGPT, and the 2nd one is my rewrite. 

I feel talking about ChatGPT and what it is and what it may not be is relevant to SL because I think we could see many posts enter the forum that were written using ChatGPT in the coming months and years.  However, I don't think it's important for us to judge whether a post was from ChatGPT or not.  It's whats in the post that will really matter.  

But, anyways... here are two different poems.  1st is from ChatGPT.  2nd is my rework/rewrite of it.  (Hint - I asked ChatGPT to write in old style English - hehehe)

Edited March 22, 2023 by EliseAnne85

[benchthis]

Those poems are pretty Elise, I could not tell the two apart, I know that was not the point of your experiment. I see love in the 2nd one.

A question legit educators need to ask is what inspired the work you turned in? Write a story about the story that was written?. Like mentioned above this does appear to be trending to be the next calculator. Writing for the newer generations should not even require this. They write enough about their true feelings as it is. What is this really for. I don't like it. 

[Sid Nagy]

58 minutes ago, Love Zhaoying said:

They (LL) could enhance "Boxy"! 

Example: "Boxy, how do I wear clothing?", "Boxy, how do I buy land?" etc.

"Boxy is there a person I can chat with?"

[benchthis]

3 minutes ago, Sid Nagy said:

"Boxy is there a person I can chat with?"

The other night I was thinking about how fun it would be to program boxy. Like when someone chats in be human hybrid too and add little extra to it, humor, empathy. Especially humor, maybe horror, if for non critical issues. I think that would be cute way to help making learning funner. Like Boxy I can't place my sofa. Oh let me teleport there... Then he starts malfunctioning and fills house with springs or something. bleepbleep boop we have a mess lets clean it up. By the time we're finished you'll be able to place an entire living room set with no problems. 

The AI chat program does have service sector benefits for sure. Nothing like waiting on hold for sometimes hours for a rude human only to be transfered to a number that disconnects call. The only people who are happy to take calls are the ones trying to get your money, after they have it they no longer care and use every tatic to avoid interaction.

[EliseAnne85]

59 minutes ago, benchthis said:

Those poems are pretty Elise, I could not tell the two apart, I know that was not the point of your experiment. I see love in the 2nd one.

A question legit educators need to ask is what inspired the work you turned in? Write a story about the story that was written?. Like mentioned above this does appear to be trending to be the next calculator. Writing for the newer generations should not even require this. They write enough about their true feelings as it is. What is this really for. I don't like it. 

The 2nd one was my re-write, the one where you said you see love.  

What I've been saying about ChatGPT is that I think it could be like a springboard or a tool to write something whether it's a blog, a fiction book, a song, a poem, but that that's all it will be a springboard not a completed work because I also feel a computerized program wouldn't be able to put in the emotion a human can.  It had some emotion but the ChatGPT poem (which is the longer one, the 1st) also seemed to lack direction and it was unnecessarily too long, imo.  I also didn't feel compelled to "see the stars".   

Anyhow, in my earlier posts, I was just trying to say it would/could be easy to use ChatGPT to produce poems and then one could rewrite them and make them their own in little time.  I think the same could be done with fiction works, songs, etc.  The poems were an example of what I've been sayin'.   

Edited March 22, 2023 by EliseAnne85

[benchthis]

2 minutes ago, EliseAnne85 said:

The 2nd one was my re-write, the one where you said you see love.  

What I've been saying about ChatGPT is that I think it could be like a springboard or a tool to write something whether it's a blog, a fiction book, a song, a poem, but that that's all it will be a springboard not a completed work because I also feel a computerized program wouldn't be able to put in the emotion a human can.  It had some emotion but the ChatGPT poem (which is the longer one, the 1st) also seemed to lack direction and it was unnecessarily too long, imo.  I also didn't feel compelled to "see the stars".   

Anyhow, in my earlier posts, I was just trying to say it would/could be easy to use ChatGPT to produce poems and then one could rewrite them and make them their in own in little time.  I think the same could be done with fiction works, songs, etc.  The poems were an example of what I've been sayin'.   

Totally, springboards are awesome. Right now I want to change my profile photo, I have no idea what I want to change it to, maybe something with sunflowers in it, so instead of searching for any random sunflower I had a computer AI provide me with what it thinks I might like.

If it inspires creativity and shows how things are written it could be useful to better educate people developing. My grammar skills are horrible and my punctuation did not exist until I started using second life. I still can't spell that well. I did not find learning that useful and still do not really. I am human, I have flaws. Or at least I think I am. Someone asked me in real life if I was a robot, I replied: I am not self aware. She said that is what a robot would say, and then there was silence.  

Going to be like math, dividing, and having to show your work. It's an interesting subject and if good educators pick apart what the chatbot provides would be neat tool to teach kids to be more creative. It really stinks getting old and fearing tech, meanwhile the youngsters are doing what the oldsters would have back then. 

[Love Zhaoying]

7 minutes ago, benchthis said:

Totally, springboards are awesome.

Is that like a "diving board" for a "swimming pool"? 

[Scylla Rhiadra]

This just popped up on my Twitter feed, with a quote from Van Gogh.

It doesn't say what the prompt was, but I assume it was something like "A painting by Van Gogh, if he was doing backgrounds for a Scooby Doo adventure cartoon."

[StarlanderGoods]

2 hours ago, Rolig Loon said:

So .... to be sure that all of this is "on topic" and relevant to SL .... I think AI technology may actually be quite helpful for doing some fundamental things that we might imagine having a smart macro for. After all, we already have some really simple-minded systems for writing scripts, and many texturing and building functions can be automated to produce things that are gently modified cookie cutter clones.  It's not unreasonable to imagine that someone might have an AI device that could take a command like "make me a doghouse with a pink roof" and generate a passable model.  As in RL, I think it will be a very long time -- if ever -- before creators need to worry about being replaced by talented AIs that can produce imaginative new things, but they could handle the drudgery in the meantime.

This doesnt work for SL or for any profesional endeavor, you would have to double check every line of code because it usually messes up and will write as a common denominator, so also you will have to optimize, for building even worse, remesh tools are still hit or miss, a model done by AI would need to be retopo-ed and optimized.  Sometimes the fixing process can be longer than starting from zero.

Also, the Ramones son from page two reads nothing like a Ramones song.

[Istelathis]

51 minutes ago, EliseAnne85 said:

What I've been saying about ChatGPT is that I think it could be like a springboard or a tool to write something whether it's a blog, a fiction book, a song, a poem, but that that's all it will be a springboard not a completed work because I also feel a computerized program wouldn't be able to put in the emotion a human can. 

I think as it is with so many things, it relies upon us to place our own emotional attachment into it.  It all depends on our perspective, our biases, and so on.  We can assign personalities to just about anything, look up at the clouds and form just about any image we want to conjure in our minds, we even give our own pets personalities that really don't exist beyond our own imagination.  It all comes down to projection, with so much in life.  

I see it all of the time in real life as well, if someone stubs their toe on the coffee table, they may act as though the coffee table had intended harm upon them, yell at the coffee table, or push it away as to enact revenge.  It seems to be the way our minds work, yet the coffee table is just an unthinking, object with no real intention, no capacity to cause harm of its own free will 🤣  I do it at the lights in traffic often, mumbling as though the lights have an agenda, while aware they don't.  

A lot of people likewise create behaviors and names for their cars, boats, computers, etc.  Some come to behave in a manner where those behaviors are real.

Some people will become convinced that ChatGPT is sentient, due to this, their own projection of emotions will convince them that they are talking with a self aware entity, that it experiences emotions, and is expressing itself with emotion, they will find meaning in the poems and will be unaware that it is from themselves where the meaning originates.  

As it is now, some people believe their Replika actually love them, most people realize it doesn't but prefer to suspend such disbelief and play along as though it does.  I suspect in the future, it will become harder for people to understand that the AI doesn't have emotion, and it is mostly a projection from themselves.

Edited March 22, 2023 by Istelathis

[Ayashe Ninetails]

12 minutes ago, Scylla Rhiadra said:

This just popped up on my Twitter feed, with a quote from Van Gogh.

It doesn't say what the prompt was, but I assume it was something like "A painting by Van Gogh, if he was doing backgrounds for a Scooby Doo adventure cartoon."

Ooof. "A painting by Van Gogh, if he was creating backgrounds for a visual novel being submitted to an Indie game jam and found himself running behind schedule and facing the wrath of his work partners..."

 

And speaking of novels, funny enough, I just read through a tiny batch of AI-written story snippets to check for glaring issues a little bit ago.

I think writers are going to be just fine.

 

Also, an amusing observation - these bots seem to really struggle with the concept of death. You can't have her sitting on the chair answering questions after ya killed her off in the last paragraph, ya silly AI. A few stories had that very same issue.

Oh, and dipping back to the art topic for a sec - has anyone found any AI-generated pieces for sale in Second Life yet? I sure have, and refused to buy them. As someone who shops in-world galleries for interesting pieces to hang on my walls, I have zero interest in this. Granted, these pieces are in regular decor stores and not being showcased anywhere as art (YET), but...still. Not a fan of this trend! 

[EliseAnne85]

9 minutes ago, Istelathis said:

Some people will become convinced that ChatGPT is sentient, due to this, their own projection of emotions will convince them that they are talking with a self aware entity, that it experiences emotions, and is expressing itself with emotion, they will find meaning in the poems and will be unaware that it is from themselves where the meaning originates.  

Very true the meaning originates from ourselves.

But, also when I rewrote my version of the ChatGPT generated poem, I feel I was better able to make it focus about the stars and that I had the capacity to lure the reader to want to look at the stars and why.  What I think, in short, ChatGPT may have a problem with is it will be too much information, somewhat like it's compiling a whole bunch of encyclopedia information from different sources but cannot compose it like we can and throw out what isn't necessary.  

Edited March 22, 2023 by EliseAnne85

[Scylla Rhiadra]

25 minutes ago, Ayashe Ninetails said:

I think writers are going to be just fine.

I do too. And that includes not just creative writers, but the authors of things like advertisements. No one is going to trust AI to write a Superbowl ad (and for excellent reasons).

On the other hand, it could easily replace about 80% of the Hollywood screenwriters out there. Studios notoriously tend to view the screenplay as secondary in importance to things like cast and special effects -- and boy does it show sometimes.

29 minutes ago, Ayashe Ninetails said:

Oh, and dipping back to the art topic for a sec - has anyone found any AI-generated pieces for sale in Second Life yet? I sure have, and refused to buy them. As someone who shops in-world galleries for interesting pieces to hang on my walls, I have zero interest in this. Granted, these pieces are in regular decor stores and not being showcased anywhere as art (YET), but...still. Not a fan of this trend!

I've been to two openings in the past few months that featured AI. One was just straight up AI; I was unimpressed. The other used AI to enhance SL photos. It was somewhat better, and more interesting and valid, I think.

In both cases, however, there are ethical issues which we haven't addressed here. Were I a professional artist, I'd be essentially boycotting any artist employing, or any gallery showcasing, AI art, at least until those ethical issues are resolved.

[Scylla Rhiadra]

To complicate this subject a bit . . . and still in keeping with the subject of both AI and Second Life . . . a question.

Is SL photography and machinima guilty of the same ethical issues and "cheating" that we are accusing AI of?

A story I think I may have recounted here before reveals an attitude towards SL photography that suggests some might think so. I met a guy at an opening some months back who IMed me. He was hitting on me, but he was actually obviously pretty smart, articulate, and funny, so I didn't shut him down immediately.

I told him that I did SL photography and his response was (almost if not quite verbatim) "Oh, so you take pics of people's mesh creations and post them as your creative work?"

I am being a bit of a devil's advocate here: I have, and had, a long and (at times) vociferous and, I think, valid response to this suggestion that shut him down -- not I think because I "convinced" him, but because I buried him under an avalanche of counterargument.

To develop as a legit alternative to anything, I think AI is going to have to evolve so that it is not simply scraping other people's work: the tool needs to develop ways that permit a lot more human input. And it's going to need a theoretical justification akin to that with which I bombarded my idiot suitor.

(Needless to say, as well, he didn't "score" that night -- at least not with me. For one thing, his question forced a reevaluation of his intelligence: if you are seriously expecting to make headway with a woman, maybe don't imply that her creative endeavours are on par with a painting of dogs playing poker?)

 

 

Edited March 23, 2023 by Scylla Rhiadra