[English]
Our 12,026 Human Era Calendar is here. This year with something
special. Stay tuned or grab it directly in the shop!
AI slop is flooding the net. In an online world where money is earned from attention,
fake users spread crap in comment sections and poison discussions. AI has
massively accelerated this – making it much harder to spot AI slop. Today,
about half of the traffic comes from bots – mostly for destructive purposes.
Producing mediocre content was never easier: LinkedIn, the black hole
of insignificance, shallow videos for kids that are just engaging enough
to hypnotize them and fry their attention,
or soulless books on Amazon. AI music is pushing into streaming platforms. Google's AI summarizes websites
instead of sending them visitors. On YouTube, new channels constantly post videos
with AI-generated thumbnails, voices, and scripts. True crime, video essays,
science – no area is safe. We are in the golden age of soulless AI slop.
And unfortunately, genuine, creative human work is being used to train these AI models. Every
Reddit comment, every YouTube video, every drawing on DeviantArt – sold to AI companies.
Or stolen directly. Without recognition or payment to the actual creators. On a scale
that's barely preventable and already endangering the work of many creatives,
while AI companies profit from it. As sad as that is – even worse is,
that generative AI has the potential to irreversibly destroy the internet. Because it's becoming
harder and harder to tell what's true. At first, AI looked great!
A Kurzgesagt video starts with basic research that's turned into a
script and then intensively fact-checked by 2-3 people.
We try to back everything up with trustworthy scientific sources. Then
we get 1–3 experts for input and criticism. Altogether around 100 hours of work.
Of course, we make mistakes or oversimplify, and you might not always agree with our conclusions.
We are human after all. But our process is thorough, and after a
decade, we know what we're doing.
When AI arrived, we were excited. A mechanical brain that
gathers info super fast? Fantastic! It started with fact-checking. We
didn't expect perfection – but it was way worse than thought.
Overconfidently wrong – AI is SO bad at that. We'll summarize our experiences from several
projects and months in one fake project:
A video about why brown dwarfs are the worst and should be ashamed.
We got pro accounts of common AI models and started with deep research tools,
to collect info about these failed stars. And wow: dozens of pages with
summaries, unique information, and links to sources!
Then we looked closer. Over 80% of the facts were pretty solid, and we could
reconstruct where they came from. Wikipedia, scientific papers, reputable articles.
And the rest? There were cool things, like the speed of superstorms on brown
dwarfs, their interiors, or how disappointed their mothers are. But we didn't know
where the AI got this info from. Not automatically suspicious – the models
include almost the entire internet and all books. So we asked experts, and they pointed out
exactly these facts and asked where we got them from.
To satisfy us, the AI had made up facts or exaggerated to make brown dwarfs
more interesting than they actually are. Like a bad journalist inventing details
to make a story look better. Now we wanted to know more, so we checked
the seemingly solid sources. One was an article from a news site,
allegedly written by a journalist. The structure was
familiar and read, certainly by pure coincidence, as if a human had slightly modified the AI's word choice.
A tool for detecting AI text showed 72% match.
So an AI-generated article without sources, used as supposedly credible basis for AI research.
No wonder – by 2025, there are already way over 1200 confirmed AI news sites,
that massively produce AI-generated disinformation and false claims.
This mix of real, dubious, and invented knowledge leads to really sloppy
AI conclusions: They sound strong, but are often half-truths or twists.
Well then. We saved what we could and started over.
And then it happened: A few weeks later, we randomly saw a video about
a brown dwarf published by a relatively new channel. It had
hundreds of thousands of views, great visuals, a good edit! But: We recognized the structure. And
lo and behold – all the facts we had marked as unreliable or AI-made were in it.
This is where the death of the internet begins. Now there's a "solid" source
with misinformation about brown dwarfs. When the next AI does the same research,
it finds this video with lots of views, and the misinformation becomes truth.
Even before AI, it was hard to trace origins of facts. Check out our video about a 100
year old lie. With increasing use of AI, it could become impossible to know,
what's true and what's not. The most harmful of all lies
The problem is that AI is correct enough to seem super smart, and incredibly
confident while being wrong. It lies to your face casually, often very subtly.
And when caught, it admits it right away and swears improvement. And does it again.
As eloquent as current language models sound, no one is home. No higher
intelligence, no consciousness is talking to you. Current AI is a complex hammer
that doesn't know what it's doing or what nails are. Still, we're letting it put new shelves in
the library of human knowledge. AI is developing fast, maybe it'll get better. Currently, it looks bleak though. Too many people have blind trust in AI
es besser. Aktuell sieht es aber düster aus.
Zu viele Menschen haben blindes Vertrauen in KI
Studies analyzed the language of millions of scientific articles – before
and after the rise of large language models. The result: A strong increase in words
that AIs like to use. A significant portion of the works thus seem to be at least
AI-assisted – usually without disclosure. And in July 2025, it was discovered that some
researchers were embedding hidden messages in their works – in white text or tiny
font – to prompt AIs to evaluate the work positively and overlook errors. The more carelessly AI is used, the less reliable
the library of our knowledge becomes.
That brings us to the last part: How do we use AI? And will Kurzgesagt
survive the age of AI slop? Does AI slop kill Kurzgesagt and all artists?
On the net, there's only one truly valuable resource: Attention.
If this continues, cheap AI content that's just "okay" enough will grab most of the
attention. That could make us dumber and less informed, shorten attention spans further,
widen political divides, and lead to neglecting real social connections. If AI consumes most attention,
dass wir echte soziale Kontakte vernachlässigen.
Wenn KI die meiste Aufmerksamkeit frisst,
channels like ours will become unprofitable, have to shrink or use AI themselves
to survive. That's a game we don't want to play.
How will we use AI? Like the "align" tool in Adobe Illustrator.
You can align boxes manually, box by box – or select all and click
"align" and it's perfectly done in an instant. The same goes for AI programming tools
for animations or using it as a faster Google alternative. AI is a
helpful tool, but creativity and integrity still lie with us.
So dear internet, that's our offer: Kurzgesagt is made by humans for humans
and will stay that way. We'll make well-researched content and invest lots of time and
our human creativity in illustrations and animations. We'll put our creative soul into the work and continue
to fact-check and discuss with experts, to give you the most serious information
we can provide. If we make mistakes, those are our mistakes. We'd rather quit
dann sind das unsere Fehler.
Wir würden lieber aufgeben,
than produce AI slop. But to keep going,
we need your support. Kurzgesagt consists of almost 70 full-time employees – plus
many freelancers. That's many salaries, software licenses, laptops, rent, and coffee.
There's a way you can keep this human-created
project alive!
The 12,026 Human Era Calendar brings a year full of Kurzgesagt art into your home – and
is much more than that. It's an ode to humanity and its ingenuity.
Our calendar doesn't start 2,000 years ago, but 12,000 years ago – at the beginning
of human civilization. So our timeline is expanded by
10,000 years of human history and its achievements.
You can use it like a normal calendar – but WARNING:
It might change your view of history and our path as a species.
This time, we're telling 12 inspiring stories about our connection to the
stars – from the first being that ever looked up at the stars, through ancient cosmology
to the future of humanity in space. Each page is printed on high-quality natural paper
and offers lots of space to plan your months or document your experiences.
For the 10th anniversary of the calendar, we're going all out: Our first
Kurzgesagt artbook – a stuffed collection of all the calendar illustrations from the last 10 years.
Plus sketches, stories, and fun facts from the Kurzgesagt team.
Just like our videos, our products are not spit out by soulless algorithms,
but lovingly researched, illustrated, and designed by real people.
So if you value genuine, human-made content over AI slop,
join our worldwide bird community that orders the calendar every year
and supports Kurzgesagt. Together, we'll survive the flood of AI trash.
Calendar and artbook are now available – while supplies last.