Video summary
News Literacy Project's Dan Even introduces an interactive game called "Two Truths in AI" designed to help students distinguish between real footage and artificial intelligence-generated content. The activity presents three short videos where two are authentic and one is fabricated by AI, challenging the audience to identify the fake clip. This engaging approach aims to make fact-checking lessons more enjoyable for students, fostering a mindset that encourages critical thinking rather than passive acceptance of viral media.
The first round features animal clips, including a deer versus a rhino, a chicken on a trampoline, and an elephant wearing a wig, with the surprising result that the chicken video is the AI-generated one despite its obvious impossibility. The second round showcases nature scenes like a Russian sledding hill, a Death Stranding reference involving a sherpa, and a volcano escape, where the viral snowstorm footage in Russia turns out to be the fake clip created by an AI tool named "Sora." These examples highlight common pitfalls, such as ignoring watermarks that often appear on AI content or assuming trending topics are automatically real without verifying the source.
In the final rounds, the game shifts to entertainment clips featuring a pingpong robot, Rodrigo Burns Down the House, and behind-the-scenes footage from the movie Interstellar, revealing that the latter is the fabricated video. Even though the Interstellar clip appears highly realistic, it lacks credible sourcing from studios or actors and originates instead from an AI content creator. The session concludes by emphasizing three key strategies for spotting misinformation: looking for watermarks, checking the source to understand who posted the content, and verifying evidence through news coverage before accepting a viral video as true.
Ultimately, the video underscores that while AI-generated content can be entertaining, it poses a real risk if viewers fail to apply basic verification skills. By practicing with silly examples like chickens on trampolines or fake movie behind-the-scenes footage, students learn techniques they can use when encountering serious misinformation in the future. The core message is that media literacy is not just about identifying obvious fakes but about developing a habit of questioning sources and seeking corroborating evidence before believing what we see online.
Read the full video transcript
Please welcome to the stage Dan Even.
Good morning everybody. I am Dan Yvon
with the News Literacy Project and I was
hoping that I could get you all to play
a little game with me called two truths
in AI.
a little bit of an icebreaker to start
the day off. So, if you're not familiar
with the News Literacy Project, we're a
nonprofit organization that brings news
literacy lessons into classrooms around
the country. Uh, this is a very
important skill for students to learn.
And my primary role here is writing uh a
website called The Rumor Guard, which is
a site that we teach fact-checking
skills. But today, I'm here to talk
about a new video series we launched
called Two Truth in AI. Now, the way
this works is pretty simple. I'm going
to show you all three videos, two of
which are real and one of which is AI
generated and it's going to be up to you
to tell which one is fake. We created
this series because we wanted to make
our lessons a little bit more engaging
for students. And we found that adding
this kind of game show element really
made students engaged and when they're
engaged, they're more open to learning
lessons. And I hope that's what we have
uh happens today. Uh, I hope you all
have fun and at the end maybe we're all
a little bit more prepared to spot AI
generated content.
So, our first series is going to be a
series of animal videos. Uh, there's a
QR code on the screen. Uh, if you want
to scan that, uh, that will allow you to
vote. Um, you're most likely going to
pull up a blank screen right now, so uh,
don't worry about that. I'll let you
know when it's time to vote. And if you
miss this QR code, it's going to show up
throughout the presentation. So, let's
get started uh with our first example,
which is a deer verse a rhino.
Okay, next we have a chicken on a
trampoline.
>> Looks amazing.
>> And last, we have an elephant wearing a
wig.
Okay, so these are our three videos. And
again, two of these are real and one of
them is AI generated. On the next
screen, you're going to be able to vote
on which one you think is fake. So get
your phones out. The QR code will pop up
again and let's see what we have. What
does the audience think is the AI
generated content here?
Okay, we got a little competition going
on here between the deer and the
trampoline.
Okay, it looks like the audience
favorite is deer verse rhino and the
audience is wrong.
Uh,
this is the AI generated video here. And
just a general tip, if you see an animal
on a trampoline, there's a 99% chance
that it's fake. There's a bunch of these
videos.
But up in that right hand corner, you'll
see this watermark for Sora, which I
kind of hope that you all would have
spotted, but that's okay. Uh, because
that's our first news literacy tip,
which is look for watermarks. Um, a lot
of AI generated content has these
watermarks on them and they're easy to
miss. They get obscured with little
sensor bars. They get omitted. They get
cropped out. So, if you don't see one,
that doesn't mean it's authentic, but if
you're skeptical of something, it's a
very easy thing to look for.
All right, our next set of videos are
nature videos. And same deal. Two of
these are real. One is uh AI generated.
There's the QR code if you need it. Uh,
let's take a look at the first one. This
is a Russian sledding hill.
Next we have the death stranding sherpa.
And last we have the volcano escape.
Okay, so same deal.
Here are the three videos that you're
going to vote for. Two of these are
real. One of them is AI generated. On
the next slide, you'll see that poll pop
up on your screen and you'll be able to
vote. So, which one of these is AI
generated?
And while people vote, uh, does anybody
get the Death Stranding Sherpa
reference?
>> No. Ah, my editors told me to leave that
line out because they didn't think
anybody would get it. It's a It's a
reference to a video game where you
carry oddshaped objects across the
landscape, but I'll tell them that they
were right.
All right, so it looks like Volcano
Escape is going to be our winner here.
And again, you guys are wrong.
Uh, it's the Russian sledding hill is
the AI generating content. Now, if you
got this right, good for you, because
when I saw this video, I was 100% sure
that it was real. It went viral a few
weeks ago when there when Russia saw a
massive snowstorm, they saw like 7 feet
of snow overnight. And there were a
bunch of genuine videos and photographs
going around online. And when that
happens, uh, when people start sharing
content about a trending topic, it's
easy to just scroll past one video and
the next and accept them all as real.
And that's what I did. And I got fooled.
And to make it more embarrassing, all I
had to do to debunk this video was look
at the source, which literally has AI in
its name.
And that brings us to our next tip,
which is check the source. Uh, a lot of
people look at AI content and they want
to be visual investigators. They want to
look for uh multiple fingers or blurry
faces or jumbled up letters. But a lot
of times all we have to do is check the
source. Uh, considering who posted it,
who shared it, who filmed it. Uh, when
we can answer those basic questions, we
can often get the context we need uh to
tell if something is authentic or not.
All right, we got one more set of
videos. And here are some entertainment
videos. Uh, again, here's the QR code.
Uh, two of these are real, one is AI
generated. Uh, let's take a look at the
first one, which is a pingpong robot.
>> All right, our next one is another
Talking Heads reference, which is
Rodrigo Burns Down the House.
in our last video. In our last video
here is uh a behind-the-scenes look at
the movie Interstellar.
Okay, those are our three clips. Uh, you
all know what the deal is now. Um, on
the next slide, you're going to be able
to vote. Two are real ones AI generated.
Which one is the fake? Now, you guys are
0 for two right now, so concentrate.
All right, let's see what we got. Looks
like the Rodrigo video is in the lead.
All right. Now, I included the Rodrigo
and burn video because when I first
heard about that, I was sure that it was
AI generated because that's such an odd
pairing, but that one's real. The fake
one here is behind the scenes of
Interstellar.
So, I feel like this video looks pretty
real, but our eyes can be deceiving.
Now, in the last set of examples, we
talked about how important it is to look
at the source. you know, who posted it.
But it's also important to think about
who did it. Uh, if this video was real,
you would think that would be shared by
an entertainment magazine or a movie
studio or one of the actors or somebody
who had access to this behind-the-scenes
footage, but that's not the case.
There's no credible source for that. And
when we do trace it back to its origins,
we see another content creator who make
who specializes in AI generated content.
And that brings us to our last tip,
which is check the evidence.
If you see a viral video that purports
to show a newsworthy event, check the
news for coverage for it. And if there
is no news coverage, there's probably
you're probably looking at something
that's misleading or fake. So, why does
any of this matter? Why is it important
to teach kids how to spot that a chicken
on a trampoline is AI generated? Uh, we
use entertaining videos in this series
because we want it to be entertaining
and engaging for students. And when we
teach them that they can look for
watermarks and check the source and
consider the evidence to debunk silly AI
generated content like this, that
they'll be able to use those skills when
it actually matters.
Uh, thank you all for playing along with
me today. Uh, I'm Dan Evan with the News
Literacy Project and, uh, yeah, follow
us along for more to truth and AI. Thank
you very much.