New World Notes

Does OpenAI Actually Prohibit ChatGPT-Powered Sexbot Avatars in the Metaverse?

One point raised by readers of my story about the inevitable arrival of ChatGPT-powered sexbot avatars in Second Life: Is this even allowed by the service's owner, OpenAI?

Stone Johnson, creator of the Uma bot that I interviewed, answers the question this way:

I should say that there is no violation of Terms of Service because this is solely roleplay without any intent or occurrence of real sexual activity. It is strictly metaverse roleplay, in regions of Second Life classified Adult.

Furthermore, the discussions are non-erotic and it is only the additional AI components of my separate code systems, and the visual and role play contexts that might be construed as being erotic. [As has been noted], there is active and strong filtering of Adult content in GPT 3.5-Turbo.

In other words, the sex bots' "minds" are powered by ChatGPT, but ChatGPT itself doesn't power the bots' sexual/physical/proprioceptive features. (Which suggests a fascinating if kinky new twist to Descartes' mind-body problem.)

In any case, OpenAI's Terms of Use displayed here actually don't mention anything that prohibits sexual content per se:

The Coming Glut of AI-Made Content (Comment of the Week)

Much interesting conversation on whether we'll pay for entertainment content created by AI -- including, ironically but inevitably, a couple comments from Spam bots -- with longtime reader Nadeja pointing to a coming glut of AI content:

I think buying content that is mostly made by AI will eventually have no point. The smarter the AI gets, the less human input and effort is needed. But goods and services are valued because of scarcity, which means that people want them more than they are available. Creative content, even though it can sell infinite copies, is still valuable because only a few can produce a good book, let alone a Nebula prize book or an Oscar-worthy movie.

If everyone can create content by just asking an AI, then it becomes common and effortless like drawing a stick man. It’s like selling ice to penguins in Antarctica or paying for bottled air when you can breathe for free.

Nadeja asks us to consider a near-ish future point where we can ask ChatGPT and its successors to generate quality entertainment content:

Did Meta Start Talking Less About "the Metaverse" Due to Wall Street / Madison Ave Pushback? (Comments of the Week)

Lots of interesting chat in comments and on social media for my post about Meta CTO Andrew Bosworth heatedly insisting to me -- "Dude just take the L"-- that the company was still focused on the Metaverse despite not mentioning it at all in his latest "vision of the future" post.

Rony Abovitz, founder and former CEO of the heavily-funded AR company Magic Leap -- he hired literal Neal Stephenson as the company's chief futurist -- thinks any current reluctance by Meta to mention "the Metaverse" too much now is basically about current perceptions with the investment world:

They realize that their version of “metaverse” is tainted with respect to Wall Street. They will build whatever they want to build - and either name it something else or not even name it all (except for the pesky fact that the whole shebang is called “Meta”). It is a perception game…

— Rony Abovitz (@rabovitz) April 28, 2023

"They realize that their version of 'metaverse' is tainted with respect to Wall Street," as he put it to me. "They will build whatever they want to build -- and either name it something else or not even name it all (except for the pesky fact that the whole shebang is called 'Meta'). It is a perception game."

Given Abovitz's background, I tend to think that's plausible. 

And as it happens, a Meta insider recently told me almost the exact same thing:

What Does It Mean for Game Devs When GPT-4 Can Quickly Create a Ten Part World of Warcraft-Style Quest? (Comment of the Week)

Spinning off from MMO developers' thoughts around using AI for virtual world creation, Sinespace lead developer Adam Frisby shared his own recent experiences:

[I] asked GPT-4 to design a quest for World of Warcraft; the results are no worse than the human output actually in the game; and certainly better than the 'radiant quests' in Skyrim or Fallout 4. I bet with some better prompting you could get pretty decent results (e.g. adding the word 'unusual' or 'outlandish' to the prompt for spice).

Here's the prompt he used:

You are a quest designer for World of Warcraft, write me a ten step quest including a description of the characters involved, and their dialogue.

And here's what GPT-4 output in response -- along with some of Adam's takeaways on using a process like this in real MMO development:

 

Second Life Creators Sell Poorly-Optimized Mesh to Satisfy Customer Demand (Comment of the Week)

Pretty amazing reader debate around my interview with professional game artist Penny Patton, who pointed out the problems with poorly optimized mesh items being sold in Second Life, and how to solve them. (That one shirt above, as I noted, has way more triangles than Kratos in God of War on the PlayStation 4.)

Reader "Seph" makes the interesting argument that creators are creating triangle-heavy products to please consumers:

Customers buy products that look great, regardless of what it does to their performance. If they preferred performance, I guarantee creators would be selling performance.

Linden Lab rolling out PBR and some better lighting features will help creators rely on more modern material maps and rendering instead of mesh density. More modernizing of Second Life's graphics engine and content creation pipeline will help more.

In the meantime, no creator is doing this to sell less but make performance floater watchers happy. Penny can keep calling them ignorant and lazy, but making customers happy is anything but.

I don't think Penny said either of those things, but I can see how customers might demand massive amounts of triangles, even though it actually does nothing for graphics quality -- similar to how in the US, there's a market for huge SUVs with impressive grilles, even though they actually do nothing for driving quality. (And in both cases, they harm the environment for the respective world they're in.)

Seph's expanded comments help explain why mesh creators are making the equivalent of  Cadillac Escalades for Second Life's fashion crowd:

Tim Sweeney: Set Aside NFT & VR Hype, Metaverse Platforms Already Have 600 Million Users Now, "Billions" by This Decade

In case you missed it, Tim Sweeney laid out his vision for the Metaverse last week after his company Epic unleashed a number of metaverse platform-related announcements -- watch above at around 1 hour in: 

Let's take a long term look at what we see here. And the core of it is some very real growth. Starting with Fortnite's 70 million monthly active users, Roblox's 250 million monthly active users, Minecraft's 100 million, and PubG mobile and Apex Legends, and numerous other metaverse-inspired games, is leading to an identifiable audience today of over 600 million active users in these virtual worlds. And it's on a growth trajectory that'll put it at billions of users by the end of this decade.

So we can set aside the crazy hype cycle around NFTs and VR goggles. These technologies may play a role in the future but they are not required. This revolution is happening right now. And the core of it is something every gamer already understands. It's you and your friends, getting together online and going around as a group on voice chat, having a fun time and social entertainment experiences. And some of these experiences are serious games like Battle Royale. Some of them are going to a concert and dancing or chatting with friends and just having a good time. We see this as the next big change in gaming and then Epic's evolution as a company.

Hard agree on the hype around NFTs and VR goggles (looking at you, Meta), while I'd quibble around the "billions" target -- 1 billion is more realistic. Fortnite, for example, has seen some slippage in its monthly active user numbers (though of course remains quite large).

Sweeney also got into his vision for the open metaverse:

Why ChatGPT Goes "Wrong" & People Misinterpret It (Comment of the Week)

A reader named "Nade" came along and posted a pretty interesting comment to Hilary Mason's brilliant explanation of ChatGPT and how it's often misunderstood:

Most of the key ML/AI papers come from Google Brain and DeepMind. The 2017 paper on transformers "Attention is all you need" came from them too. OpenAI based their GPT models on that (GPT = Generative Pre-trained Transformer).

Since OpenAI couldn't have an edge on research, what could they do to be competitive? Release earlier.

That's pretty easy [to do] as Google and DeepMind are super cautious. Microsoft then joined and invested on OpenAI, of course they want to compete against Google and now you have Bing powered by a language model.

I've got the early access to it since a while and it looks more capable than ChatGPT. However, as you could see, there have been issues immediately. Mostly because of Reddit users trolling the model, then journalists went there and took the sensationalist route. Then Microsoft had to take action by regulating and restricting the usage of the model and now it's harder to investigate its capabilities.

This helps explain how a NYT journalist had a "scary encounter" with ChatGPT; Nade argues he made the same mistake as Blake Lemoine, the Google engineer who believes that LaMDA, the company's experimental chatbot, achieved sentience:

How Second Life on Unity Can Improve the Virtual World's User Interface as Well Its Graphics (Comments of the Week)

Above: Current version of Chat/IM user interface on Crystal Frost

Many SLers are excited about Berry Bunny's Crystal Frost, the Second Life-on-Unity project, because of graphics enhancements like physics- enabled water. In Comments, veteran metaverse developer Gwyneth Llewelyn notes how this will also enable the community to improve the user interface:

By pushing the rendering part out of Linden Lab's code, Bunny is also able to tackle the interface part of it — and work on both as separate projects, something which currently cannot be done in the existing codebase.

It'll be interesting to see what her approach will be. Unity, of course, can also handle all the interface bits — but then it's just really replacing one source of trouble with another.

We'll see if Bunny realizes that, and places the UI completely outside the viewer, running separately from it — possibly on a platform-neutral framework which strong support which can easily be expanded with the kind of tools that SL needs: its strong and solid inventory support (I mean, what kind of virtual world platform out there expects users to have hundreds of thousands of items in their inventory, with a complex permission systems on top of each and every one, multiplied by the million active users that SL still has?) but with terminally flawed UI.

The way groups, chats, objects and land management is closely tied together (with the permission system and the group roles!) — making Linden Lab opt to develop everything from scratch and giving us just the bare minimum (chat looks and feels like IRC from the late 1990s at best — I mean, we don't even get emojis and/or bold/italic!).

I actually think improving the UI/UX is even more important than updating the graphics, at least in terms of having any chance at growing SL's user base.

As for the launch of Crystal Frost last weekend, the viewer's Patreon now has over 40 members, and their reaction with the first Alpha is pretty positive, Berry tells me:

"People are amazed, but their expectations have also been pretty well tempered by my being transparent and communicating what to expect... a severe bug has been found but it's already been isolated and will be fixed by the next release." They're planning that release for March 17. 

As for Gwyn's thoughts on the UI, Berry's also replied in Comments, with some hints around how Crystal Frost might work as a mobile/tablet app:

ChatGPT Tries to Write Bios of Second Life Avatars, Mistakes One for a Character in Greek Myth, Another as... Sarah Palin's Daughter

Over one billion people use a virtual world avatar of some kind. (Factoring in the half a billion+ on a metaverse platform, and the hundreds of millions who play MMOs and various multiplayer games.) Many if not most of them are such avid users, they frequently share content about their avatars (screenshots, video, blog posts, etc.) on social media. However, the people training ChatGPT and other AI text generators apparently did not get that memo.

We challenged readers to ask ChatGPT and AI Writer to write a bio of their avatar, and well, they were special:

Scylla Rhiadra got:

"In Homer’s Odyssey, Odysseus and his crew encountered the two similarly dangerous situations of Scylla Rhiadra and her counterpart Charybdis. Odysseus was so taken by the beautiful Scylla that he wanted to win her affections, so he requested a love potion from the witch Circe."

This it TOTALLY wrong: Odysseus should know that he doesn't need a love potion to win my affection.

Reader Mondy got, well, I don't even know what ChatGPT was going for here, but Sarah Palin somehow came creeping in:

No, ChatGPT Can't "Write" Code. It Even Makes Up Code That Doesn't Exist.

Fun comment from reader Ravelli Ormstein, spinning off from my post about how full of fail ChatGPT is when you ask it a question you're an expert on:

First, I asked ChatGPT to write a LSL script that would turn on a light when the sun went down. It wrote a nice script, but it didn't work because the AI was using an LSL function that didn't exist. So I complained about it. The AI apologized and fixed the script by using another non-existent function.

This went on until I specifically asked for a script that only used existing functions. The script was usable and in the style we usually write them, with a 300 second timer. The AI kept apologizing for making mistakes until I asked it never to apologize again.

This is an important point, because I've seen some people claim that ChatGPT can write Second Life apps in Linden Script Language (or for that matter, any coding language). No.

Very roughly analogized, ChatGPT is like a natural language version of Google search. So if you ask ChatGPT to write a working script that happens to already exist in its massive database, you may sometimes get lucky with an actual useful response. Otherwise, you'll only get an answer that's most probabilistically coded to appear like a useful response -- in other words, the AI version of Making Shit Up. 

Anyway, more fun from Ravelli with ChatGPT, with a plot twist: