WEBVTT

[MUSIC PLAYING]

So generated online video has fundamentally

changed in the past couple of weeks

in a way that has a very negative impact on human creativity.

Because what we're mostly seeing with all these generated

videos is very much not creative.

And what just happened in these past two weeks

is different than a lot of the releases we've seen so far.

And it affects everyone who creates or consumes content

online.

So generating video was once kind of like this curiosity,

or it wasn't easily accessible to most people.

But now that has changed.

It's free.

It's widely available in tools that

are aimed squarely at you and your social media feeds.

So I'm going to cover the pretty significant turning

point that just happened in generative video and audio

as well, and what that means for creators, for audiences,

and really, I mean, for the whole internet itself

at this point.

I mean, we might be asking for proof of life

from the internet quite soon.

So I'll cover the new apps from OpenAI, from YouTube,

from Meta, from Google, the people

that are putting this technology in everybody's hands

on a huge scale.

And they've made it really easily accessible to everyone.

We're going to look at some of the real world consequences

of this happening.

There's questions about copyright and likeness protection.

There's a huge environmental cost

of powering these new tools.

But also what it's doing to data and the internet.

I'm Jen deHaan, and this is the human internet theory.

Today, we're going to go over some

of the latest of what's going on,

things that are relevant to the deadening of the internet.

First, I'm going to cover the specific news and updates

that you need to know about to understand this,

all what's going on.

Then I'm going to share my thoughts

on what this means for human creativity

and the things that we need to watch out for and advocate for,

I think.

Before we get into my take on all of this,

let's understand exactly what happened out there lately.

It's been a busy few weeks for the tech bros,

and these changes are substantial

and very important to the things that we do.

So the generative video sector just experienced

this huge turning point or infection point

or inflection point or whatever.

We just saw high quality generative video

become widely available with very generous limits for free,

and it's being targeted right at social media

by anyone who can access a device or a computer pretty much.

There are a few limitations in some of the tools

around region and codes, access codes and stuff,

but they aren't going to last too long at this rate,

even if they are still there.

It's like things are changing so quickly.

The big changes that have happened in these past few weeks

are support for vertical video.

That means that all these videos are really being aimed

at places like TikTok and Shorts and Reels on a large scale,

and also the ability to sync audio,

like voices and sound effects natively in these videos

when you generate them.

So those are big recent changes that have happened,

and let's break down who is doing what.

So OpenAI just launched Sora2 a few days ago

and launched a social media app to go with it.

Now, this app was initially invite only,

but this past weekend, they seem to have possibly opened

the doors to a lot more people without the codes,

like me.

I wanted to have a basic understanding

of what I was talking about.

So I went ahead and I got the Sora app.

I needed a code to access that,

but the next day when I opened up the app,

wasn't asking me for a code

and it gave me full access to everything.

So Sora is free with some pretty generous generation limits,

and OpenAI appears to also be losing a ton of cash

every time people make videos.

So they wanted to really make this available

to everyone in this kind of race.

When OpenAI launched this,

they also had a default opt out copyright policy.

So this meant that the rights holders,

the people that own the copyright

to certain characters and so on,

had to manually opt out of having their work used

in this app.

Anyways, they backtracked on the policy

that they launched with very quickly

because there was a lot of complaints

and backlash about it.

So YouTube now has its model VO3 Fast

built directly into YouTube Studio,

and that's for generating YouTube Shorts.

A VO3 Fast is the lower quality

video generation model that they have.

But this model includes both video and audio generation

in sync and in standalone content

that you can then just push into Shorts.

So their release along with that video tool

also had several other new AI tools

for creating soundtracks, audio,

and editing videos as well.

There were some region and language limitations

on their video tool,

but I'm gonna guess that that's not gonna last

a whole lot longer either.

Google also has an AI filmmaking app called Flow,

and they have their regular higher quality

VO model built in VO2.

So this is a higher quality or resolution

than what's on YouTube Shorts.

Now, there was a very recent update to this tool,

I think maybe in the past week or so,

that made Flow available also for free.

They're giving you 100 credits

to get you started on that tool.

So that is a very new change

in this whole race for video generation dominance.

There's a note on the Flow website

saying that over the next few weeks,

we'll be phasing out VO2 in Flow

as we work on having VO3 do everything that VO2 can

simplify the user experience.

So there's a big quality bump with free access

in that tool as well.

And Meta's not gonna be left out on all of this,

of course, in this race for peak social media impact,

they released a new app called Vibes,

which can generate AI video

and it can remix those AI videos.

It also is a dedicated feed of just AI generated videos

with the Meta AI app and the website,

I guess it's on both,

'cause I'll be honest,

I haven't gone and looked for these things

to look at the feed.

I just read the articles instead

because I've seen the type of AI videos

on Facebook in particular

and I felt that that was as far as I probably needed to go.

But I'll link the articles in the description

if you wanna go find Vibes and check it out.

So the four big players just put their tools out there

for everyone, for free, so wide access.

And this point, this turning point

might not be just limited to video.

So these videos have made a big leap

in audio capabilities as well.

And in audio generation on its own,

there have been recent advancements

in music generation from tools like Suno as well.

We've seen companies like Inception Point AI

releasing thousands of podcast episodes every week.

And in a recent interview on Pod News,

it was really interesting.

Their CEO admitted that they mostly don't listen

to or review this content before publishing it.

I mean, I'm really paraphrasing there.

You can go check out that interview.

I'll put the link in the description.

But I should also make a note here about Deepfakes

and all of these video generation tools

before we move on because YouTube is rolling out

likeness protections against Deepfakes in their tools.

Sora is also doing something similar

that's allowing artists to authorize their likeness

with a feature that they call cameos.

Judging by what I'm already seeing out there

from OpenAI's tool and everything,

that doesn't seem to be going too well

or might not be reliable yet.

As a related example, while writing the outline

for this video earlier today,

I got a notification that some big account

that I follow just put something new online.

You should go look at it.

And it seems like they were hacked

by somebody who posted generated videos

with fake major news events.

So between the Deepfakes that are coming out already

and these fake news stories, these fake disasters,

just basically RIP to knowing what's real

on an entirely new scale because of these.

I've heard a reaction to that criticism

of while we've had Photoshop on the internet for decades,

but I don't think that's very valid in light

of the level of access and the level of these tools

now available for free to everyone as of this month.

And on that topic of generated identities,

let's talk about Tilly Norwood for a second.

So Tilly is an AI generated character

who in the last month has been the subject of a push

for agent representation in Hollywood by its creator.

So this has led to basically a flood of clever

but angry clever posts from actors and unions

and many of my improviser friends as well.

Very memorable, clever, angry posts or sarcastic really.

The move was also really widely condemned formally

by the actors unions worldwide.

But I think that whole event that also happened this month

could be a very clear look at how Hollywood

intends to commercialize these generative video AI tools

moving forward, same but probably much more advanced

models than the things we're seeing with SOAR 2 and VO3.

Really Tilly might also be a PR stunt

to get necessary investor interest to continue

in whatever projects are going on behind the scenes.

Before we get to my opinion on all of this,

there's one more piece of news to cover.

So to power all of this massive uptick in content generation

and how big these video files are

for these short form memes,

companies are making huge deals for huge mega data centers

that are needed to store and process all of this stuff

that's now going out there.

And these facilities,

they all have a significant environmental cost.

Their energy consumption, for example,

can rival or even exceed large cities like New York.

So in some of these communities as well,

the people living there, the residents have become worried

about water safety and access to drinking water as well.

So you know where to find the links,

all of that will be in the description as well.

So while we're talking about data centers,

check out my site at humaninternettheory.com

and connect with me there.

I have a newsletter there where I'll be sending out updates

about things like this every so often.

We have these new technologies,

but let's step back for a second and ask a bigger question.

And that question is, what are we actually doing?

Right?

We're exploding the amount of content

that we're creating online

and we're needing to store it as well.

So we're really radically changing what will be available,

what is going to be the content of our feeds

and our websites and our search results as well.

And we're looking for information.

So I make light of these memes and these trends,

but as a whole, I do think that we need

to start asking these questions.

Like why are we literally creating podcasts on topics

that no human even bothers listening to before publishing?

Because I know for sure that there's a ton of garbage advice

on gardening, for example, online that doesn't even work.

And that's the stuff that LLMs are just pulling in

to create that podcast that no human listens to

before it gets served out to the masses.

And apparently that doesn't even matter.

You can read that in the interview

I'm linking in the description as well.

So RIP, your tomatoes, maybe.

We're calling this art and we're calling this creativity.

And we're saying that we've democratized the creator,

but with these tools, what we're doing instead really

is turning humans into executive producers

that are really removed from the project

or commissioners or clients.

And the creator in this case is AI

and the collective work of all the artists

the model was trained on.

The creative work is being done by those other artists

and the model itself,

not the person typing it in necessarily.

So let's talk about creativity.

Like what is creative?

What is being creative?

So ideas.

The ideas are great, but artists and educators

usually don't take the very first idea that we have.

We create or we get a bunch of ideas together in our heads

and then we pursue the ideas that really pass muster.

These are the ideas that we sort of judge

as ones that bear merit, that have merit.

And then we take those ideas

and we iterate on the good ones, the ones that we like.

And then we let those ideas marinate in our brains really.

We let them sit there in our subconscious

and then we change them, we get new ideas,

we build those ideas out.

And then the execution part happens.

And that execution part is most likely going to include

further iteration as more connections and stuff,

more marination happens.

And then we go through revision processes.

And those things all combined together

is what pushes that idea into a creation.

And all of those things are really worthy

of our time and our effort.

And they're also hopefully worthy of the power

and water consumption of a data center somewhere

if we're putting it online.

The challenge, that effort,

and most importantly that iteration process over time

is what makes something great.

That is the creativity.

Like many of my ideas, for example, for these videos,

but my audio dramas especially,

they happen, they start in a recliner

and an app where I type down my ideas as I get them.

And then I let them sit in my brain and iterate on them.

I iterate a lot when I'm on my elliptical, just walking.

I'll get an idea, I'll stop, I will be just looking

at a wall, walking on my elliptical

and thinking about the ideas and iterating them.

Or I might get an idea when I'm out in the garden, bored.

But then not because I'm thinking about ideas.

And then I revise those ideas say with another human

while I'm in the kitchen cooking something.

I might change those ideas again

while I'm actually creating the show

or writing the rundown or writing the script,

whatever it is.

So deep thinking, subconscious thinking.

Those things are so important for sustaining our creativity

over a lifetime.

Could we do this with AI though?

Collaboration for sure.

I think we can collaborate.

But right now it looks like a majority of the things

that we're doing here with these new tools

is like blasting social media and websites

in ways where there isn't maybe a lot of thinking involved.

And certainly not much iteration.

Because we also see this on a large scale

with InflationPoint AI that I discussed

that doesn't have much or any human oversight

before it gets placed online.

We have to ask ourselves these questions.

Is this what we want to watch on our feeds?

Is this what we want dished out?

When we search on a topic,

ongoing for years on search engines

or in the AI tools and the LLMs themselves.

Because it's essentially like this watering down

of creativity and accuracy,

the way that most people are doing it today it seems.

Now we could see better storytelling,

better thinking, creativity.

If more people and companies start using AI

as a more of a collaborative tool.

And some are for sure.

But many aren't as we see with this sort of prevalence

of proof of concept style output

with the low effort sounding text posts

with unedited AI written articles and podcasts

that all of this feels very divorced

from much of the human collaboration part of the process.

Like it feels like animation or CG

that kind of skipped the long

but very rewarding and fun process of drawing storyboards

and writing and editing and iteration

and using humans for motion capture and so on.

So if a lot of our content that we see online

ends up being like that

is really gonna give dead internet theory

out there pretty quickly.

So we need the human internet theory.

It's important to get ideas and let them marinate

in order to do that iterative process.

And there's science around how our unconscious mind

goes through an incubation period

in order to trigger new ideas

or the pivots from our current ideas.

And these things can be so useful.

They're critical in the implementation stage.

One interesting learning that I got from neuroscience

reading about neuroscience is called the default mode network.

And this network activates in our brains

during this stage of idea generation.

And it's great at integrating ideas,

making those connections and pattern recognition.

It's our subconscious, our unconscious brain.

All of these things are what lead to breakthroughs,

creative breakthroughs,

and we need this time cognitively.

So giving it up entirely to AI, replacing it

is a very serious issue when it comes to creativity

for humans.

And I'll put links about that neuroscience

in the description as well.

So this doesn't mean that we need to get rid of AI,

but it does mean that we need to think about

and very possibly advocate for using it differently

as a collaborative tool to help humans

rather than replace this part of the process

to make those who just replace creativity

who don't bother with human oversight.

We need to collectively say something to people

who are replacing their creativity with AI with something.

Like say, I love that for you in a way

that really implies that we don't love what they're doing.

I don't know, do something like that.

We need to make it embarrassing.

So I've heard the industry discredit this argument

that I'm making this sentiment with, for example,

lazy Luddite as the inception point CEO

loved to say about us, or that we're all dinosaurs

who don't wanna keep up with the times.

But I don't really buy this,

especially since a lot of people are not dissing

all of that AI.

Like we're not saying we need to throw it all out.

The part that I'm concerned about

and specifically talking about why I created this channel

is specifically about how this mass fire hose of content

is affecting the internet and human creativity

when it's used to just replace it for the most part

or replace most of it.

When humans are being removed from a lot most

or all of this part of the process.

So for creativity to continue,

we need to support people doing the hard work

who are continuing to do this hard work of generating

and iterating on their ideas of revising their work,

editing their work and keeping the human oversight in there.

We need places that will curate

and continue to support this kind of work

through enhanced discovery of human creation.

We need to be able to search for this work.

We need to be able to find it and run across it.

We need new human ideas

and we need that to remain accessible to other humans.

What we share as creative humans should not be lost

to this fire hose of trends and memes

made by automated process with little to no human oversight

or the industries that are pushing out

the human creativity for capitalistic interests

and deaden the whole internet or our Hollywood movies.

Like the AI generated Tilly Norwood.

That creation is not an actor

by any kind of definition of acting.

Acting requires emotions.

It requires sentience.

And just as what's being generated from a prompt is not art.

Same thing, it's not creativity.

It's not creating.

It's like producing or hiring somebody

to do something for you.

Especially if there's no human oversight

just like with those Inception Point podcasts.

People are just publishing ideas.

Often not even ideas of their own

but now on a massive scale.

So what has happened over the past month

is a pretty serious signal of what we're gonna probably see

online for a long time.

So it's a problem if humans stop using our tools

and our brains to learn and think and ruminate

and then create because an original work

is no longer supported or discoverable online.

And this is a problem if whatever we do to create

is now also quite possibly written off as generated.

For sure that's already happening.

And then that just makes people quit creating

or doing or trying or sharing or helping.

And I'll leave on this thought.

and they ranked creativity as the most crucial

leadership quality.

So again, I ask what are we doing to our human brains

with so much replacement of our creative thinking skills?

And in these past few weeks, we've sure seen a shift

to something that really prioritizes volume

and speed and investor interests over craft and creativity

and the humans on the internet.

And the risk that we already have today

is that this sheer quantity of generated content

could really overwhelm the platforms

that we use for discovery on a basically just a highway

towards an internet where it's egregiously difficult

for genuine human led creative ideas

and work and connection, all of those things

to find human eyeballs.

So the challenge moving forward is to make sure

that we don't lose that,

that we don't lose this human process of creativity

that the internet and really everything else

that touches the internet, which is a lot of things now,

that it was all really built upon,

that we don't really end up with an entirely dead internet.

So I'll be back later this week for another episode

on creating using our own brains for the internet

that this lazy Luddite is doing.

Bye for now.

Episodes are written, directed, edited and produced

by Jen at STereoForest.com

Find out more about this podcast

and join our free newsletter for additional resources

at humaninternettheory.com.

Find additional videos at the YouTube channel

called Human Internet Theory.

Links are also in the show notes.

(upbeat music)

[BLANK_AUDIO]