New World Notes

Generative AI Survey Results: Overwhelming Majority Say It Won't Radically Change Culture, Just Be Another Creative Tool

Here's the results of last week's survey, asking how much generative AI will change movies, music, and metaverse platforms. By and overwhelming 65%, the majority opinion is that AI's impact will be nominal, and that generative AI programs like ChatGPT and Midjourney will just become one of many other high tech tools in our creative tool chest. Less than 1 in 4 said AI's impact will be "radical".

So I guess that settles it!

Though in fairness, with just 26 responses, this is a pretty small data sample -- most other New World Notes surveys get 200-500 responses. So maybe these results indicate most readers are very unsure on the topic. (Or maybe don't really care either way?)

Survey: How Much Will Generative AI Change Movies, Music, and Metaverse Culture? Read the Debate, Take the Survey!

Here's a survey on how much generative AI will change important parts of human culture, inspired by a debate posted last week. In case you missed it, read the different perspectives below here:

• How Artificial Intelligence May Change Metaverse Platforms.
• How AI May Change Movies and Music. 

Roughly summarized, one argument suggests AI will totally remake our conception of movies and other culture, replacing many or most creative humans along the way. Another argument suggests AI will become just one tool, much like programs like Photoshop did in previous years -- and humans will still be central to creating culture.

Read the debate summary after the break, take the survey, and expand the conversation in Comments below!

How Artificial Intelligence May Change Metaverse Platforms -- And Challenge Creators Who Make Their Living from Them

Above: Meeting BunnyGPT, one of the first NPCs hooked up to OpenAI's LLM

Read Part 1, Will Artificial Intelligence Totally Transform Culture & Intellectual Work? 

After discussing how AI might change movies and music, Adam Frisby and I debated how it could alter the future of the Metaverse. It's topic both of us have long been thinking about, Adam as CEO of Sine Wave Entertainment (creator of metaverse platform Sinespace) and myself as the author of The Making of Second Life and Making a Metaverse That Matters.

Wagner James Au: How about metaverse platforms? How much will gen AI transform them? Roblox I know is already experimenting with it for graphics, but how much will it change the overall experience in the next decade?

Adam Frisby: Probably; the general rule of thumb is: do big datasets exist for the job being replaced. In the case of metaverse platforms, you've got primarily the following jobs done by creators:  

- 3D modelers

- Animator

- Texture artists

- Riggers

- Scripters

- Environment artists/level designers

Those are the game industry job titles, but it is analogous to virtual worlds. Of those; there are compelling high quality AI tools already for animation.

Here's a tool from 2018 [watch below]:

Will Artificial Intelligence Totally Transform Culture & Intellectual Work? An Engineer & Author Debate. Part 1, Music & Movies

Pictured: Barbenheimer meme created through generative AI programs Midjourney and Gen-2 

In the near future, will we keep wanting to watch Hollywood movies like Oppenheimer and Barbie, works of visionary highly personal art by beloved human artists -- or will much of their work be replaced by generative AI programs?

In recent months I've been having a friendly Whatsapp debate on that question with Adam Frisby, CEO of metaverse platform developer Sine Wave Entertainment, one of the smartest technologists I'm lucky to call a pal. Even though Adam takes a highly AI maximalist position on that question!

“Every intellectual job is at risk," as he recently put it to me bluntly. "The ones which can be easily outsourced already today, will be first. Even very prestigious, difficult and stressful jobs like C-suite execs are at risk honestly. Humans can only handle summaries, AI can handle huge quantities of raw data; long term, we're at a disadvantage.”

As longtime readers to my blog might guess, I'm highly skeptical of that -- and not just as a professional writer devoted to the human aspect of technology in the Metaverse and beyond. I've been working in Silicon Valley long enough to have heard grandiose predictions like the above before (and then maybe yet before that before). I tend to think AI will just become one more technology tool among many to produce movies and other art -- but not fundamentally change who makes them. (I.E., human producers, directors, writers, actors, and others talented artists and craftspeople.)

But maybe this time the grandiose predictions are different?

Recent news around OpenAI and Sam Altman recently re-joining as its CEO after a week of confusing tumult may accelerate the issues we discuss, given the org's heavy involvement with Microsoft. Disney, historically a very litigious protector of its content, recently asked Microsoft to remove some its IP out of its AI image program, and may ultimately see Microsoft as a potential competitor. 

Anyway, read our conservation and decide for yourself!

PART 1: How Artificial Intelligence May Change Movies and Music

Wagner James Au: I’m sure you saw that the Screen Actors Guild got the studios to forbid use of AI characters in movies and TV without the underlying actor’s permission. So that puts the kibosh on AI totally transforming popular culture, right?

Adam Frisby: Not at all. “Why pay for a real actor?”. Humans are expensive, notoriously difficult to work with, and the studios are investing vast sums in building someone else’s brand that they don’t even control.

(This is not to be callous, or to devalue artists' contributions to media - just that, as soon as a machine can do a human's job reliably and cost effectively, it will; no matter where or what that job is.)

WJA: But name actors are integral to funding and marketing movies. Even when it’s a CGI animated movie, studios depend heavily on having A list stars doing the voice work.

AF: Yeah; but that's just a matter of branding. AI can be tuned to reproducible results, studios could invest in their own stable of "actors" and probably will. There may be initial backlash -- it'll likely happen somewhere small first, TV maybe? Where novelty is okay.

It could happen via some VTuber initially. That's actually more likely -- low risk, already happening, has a human heavily driving it, someone gets a high profile from Twitch or YouTube, "stars" in something bigger -- low budget film or a TV series, as their character.

That goes on for a little while, the owner licenses the character more fully; precedent has been set.

WJA: But how would that supplant how Hollywood studios work with human actors, when they’re already bound not to use AI actors without permission for 3 years and SAG has no motivation whatsoever to change that in the next contract?

AF: AI versions of existing actors, and permission by the actor themselves. Hell, another route could be someone like Bruce Willis. Someone who for a tragic reason needs "AI help" to act again.

Or, it'll be someone who does something Avant Garde -- "The Film With 10,000 Endings"

However it happens, it'll start small. The studios are not fools. They're aware everyone in the industry is nervous. But that toehold will grow; and eventually the economics of brand name actors will be broken. Because the human actors -- and the supporting infrastructure (cameramen, etc) will be a huge expense for an ever decreasing proportion of the budget

VFX will be long gone by then -- that'll eventually be 5 guys running prompts over raw footage.

Look at this, this is pretty preliminary and early tech, but you can easily see how quickly it'll be able to do complex VFX shots [watch below]:

Is It Ethical for Artists to Use Generative AI in Their Work? This Creator Believes So -- See Her Images & Her Approach for Making Them

There's quite a lot of excitement about the potential to quickly create art through generative AI programs like Midjourney, but there's one small problem for starters: Most of these "artificial intelligence" programs are churning out images that are really just copies of art work by actual human artists without their consent or compensation.

Is it even possible to create AI-based, ethically sound images? One artist, who goes by the name "Milena Carbone", believes so. In real life Milena is a photographer who's shown her own work off in galleries, experimented with Second Life screenshots as a medium, and is now experimenting with generative AI.

The images you see in this post are from Milena, and by her lights, they're ethically sourced:

"It's impossible to avoid being influenced by an artist," she allows, "but this isn't specific to AI. When we create, we're always influenced by what we're attracted to, by the authors we admire, consciously or unconsciously."

To prevent outright copying, she declines to include artists in her Midjourney prompts:

UPDATE: The Verge Now Asking Meta About Quest 3 Tracking Complaints Reported in New World Notes Last Week

Appreciate this post from Jess Weatherbed at The Verge:

Meta is investigating reports of tracking issues blighting its Quest 3 VR headset following user complaints that controller tracking performance is “much worse than the Quest 2” in some games...

One comment from a game developer (known as carrotstien) for Eleven Table Tennis claims the issues stem from the LED arrangement on the controller. “Oh how I wish the developers of sports games were invited to the table when discussing controller design,” said carrotstien, adding that “there are users which are simply sticking with their quest2 headsets and returning their quest3s because of the tracking issues.” 

As reported by New World Notes, complaints have also been posted to Reddit in the last few weeks echoing similar tracking problems.

I'm skeptical that the problem resides in LED arrangement, rather than the AI models for "estimating" controller position that Meta's CTO recently boasted about. Any hardcore gamer knows that in most twitch games, you need absolutely precise. real time control over your game interface, not "estimates".

For the record, reader Liana says she hasn't personally experienced this problem playing Beat Saber, which probably the most popular twitchy Quest game:

Meta Bragged AI Would Improve Quest 3 Tracking. But Many Quest 3 Owners Are Complaining Hit Games Are Unplayable Due to Poor Tracking.

Here's what Meta's CTO Andrew Bosworth bragged about in June, regarding AI improving the soon-to-be-released Quest 3:

Our decades-long work in AI is driving progress everywhere even in places you wouldn’t expect like Quest 3’s Touch Plus controllers.

No tracking rings—AI models for estimating their position got so good we don’t need them anymore.

Result is sleeker, more versatile controllers.

But here's the kind of complaints some Quest 3 owners are making now:

Quest 3 Controller Tracking Issues

It does not seem that the tracking system was thoroughly tested with some of the biggest selling games by people very skilled at those games, nor did they test much, if at all with controller attachments for various apps, which are hugely popular across all game genres including many of the biggest selling Quest games...

If whomever tested this opened up a golf or tennis game, played Gorilla tag or any number of other games (many of which are HUGE sellers) at the level of skilled players, they would have seen in short order that the proposed tracking system would be problematic.

Veteran VR fan Tim "FlipperPA" Allen, a longtime member of the hardcore Beat Saber community, confirms this with me first-hand:

"[I] noticed almost immediately how unusable the new controllers were. I've been fighting a pinched nerve in my neck, and was hoping the new slimmer design and even weight of the Quest 3 might help. But without usable controllers with decent tracking, it is a non-starter for my exercise routine, such as Beat Saber!"

To further validate that this is not just an isolated incident, he also sent along multiple comment threads of Quest 3 owners complaining about the same problem:

• Quest 3 bad tracking 
• Is anyone else having bad tracking on quest 3
• Don’t buy the Quest 3!

Someone from Meta tech support even replied to these complaints, saying:

We'll be working to improve performance, especially for competitive players in games like Gorilla Tag, Supernatural, IB Cricket, and Eleven Table Tennis. We expect to roll some of these improvements in the upcoming releases.

But Tim (and for that matter, me) aren't convinced software fixes will address the fundamental issue. Because the problem seems to reside with VR experiences like Beat Saber that require highly precise, high speed tracking -- and apparently, the AI algorithm's inability to approximate those accurately:

LLMs May be the Trojan Horse That Modernizes Software Development -- Charity Majors

Charity  Majors, who once helped keep the utterly tangled spaghetti of Second Life's servers operational back in the day, and now puts that wisdom to practice as co-founder/CTO of observability service HoneyComb, has a pretty thought-provoking post on how large language models are changing the nature of programming:

You can make a surprising amount of improvements to an LLM based product without even touching any prompt engineering, simply by examining user interactions, scoring the quality of the response, and acting on the correctable errors (mainly data model mismatches and parsing/validation checks). You can fix or handle for these manually in the code, which will also give you a bunch of test cases that your corrections actually work! These tests will not verify that a particular input always yields a correct final output, but they will verify that a correctable LLM output can indeed be corrected.

You can go a long way in the realm of pure software, without reaching for prompt engineering. But ultimately, the only way to improve LLM-based software is by adjusting the prompt, scoring the quality of the responses (or relying on scores provided by end users), and readjusting accordingly. In other words, improving software that uses LLMs can only be done by observability and experimentation. Tweak the inputs, evaluate the outputs, and every now and again, consider your dataset for representivity drift.

Her surprising conclusion (at least I'm surprised) is she sees a near future were programmers will regularly use LLMs as part of their everyday toolchest:

Generative AI's Inevitable Rise Suddenly Stopped By Pesky Human Creators at the WGA

Hey remember reading about how AI will destroy Hollywood as we know it and change Hollywood forever? Of course you do, people have been making bold assertions like that since, well, yesterday.

And then those darn pesky humans at the Writers Guild of America organized, went on strike, and in a few months, got the world's largest film and TV companies to agree to these highly restrictive terms around AI:

We have established regulations for the use of artificial intelligence (“AI”) on Minimum Basic Agreement-covered projects in the following ways:

AI can’t write or rewrite literary material, and AI-generated material will not be considered source material under the MBA, meaning that AI-generated material can’t be used to undermine a writer’s credit or separated rights. 

A writer can choose to use AI when performing writing services, if the company consents and provided that the writer follows applicable company policies, but the company can’t require the writer to use AI software (e.g., ChatGPT) when performing writing services. 

The Company must disclose to the writer if any materials given to the writer have been generated by AI or incorporate AI-generated material.

The WGA reserves the right to assert that exploitation of writers’ material to train AI is prohibited by MBA or other law.

So I'll just call it right now: The WGA agreement signals the end of generative AI having any chance at becoming a disruptive force in the creation of intellectual property.

ChatGPT & Other AI Programs Not Conscious, Argue Experts Who Publish Checklist of Meaningful Indicators of Consciousness

Chart via Consciousness in Artificial Intelligence: Insights from the Science of Consciousness [pdf]

Here's a free gift link to a fascinating New York Times article about a group of philosophers, neuroscientists and computer scientists who published a comprehensive study on consciousness and AI:

The fuzziness of consciousness, its imprecision, has made its study anathema in the natural sciences. At least until recently, the project was largely left to philosophers, who often were only marginally better than others at clarifying their object of study. Hod Lipson, a roboticist at Columbia University, said that some people in his field referred to consciousness as “the C-word.” Grace Lindsay, a neuroscientist at New York University, said, “There was this idea that you can’t study consciousness until you have tenure.”

Nonetheless, a few weeks ago, a group of philosophers, neuroscientists and computer scientists, Dr. Lindsay among them, proposed a rubric with which to determine whether an A.I. system like ChatGPT could be considered conscious. The report, which surveys what Dr. Lindsay calls the “brand-new” science of consciousness, pulls together elements from a half-dozen nascent empirical theories and proposes a list of measurable qualities that might suggest the presence of some presence in a machine.

Here's a pdf link to the actual report, which requires some serious heavy lifting to read, but it includes this nice colored chart (above) with a checklist of consciousness indicators.

For instance: "Agency guided by a general belief-formation and action selection system, and a strong disposition to update beliefs in accordance with the outputs of metacognitive monitoring." Basically that means: When the AI knows that its conscious, and makes decisions with that knowledge in mind. Or as the Times puts it: