Tag Archives: deep fake

An AI advertisement

I scrolled past this add a few times before paying any attention to it. But then it gave off an uncanny valley feeling that made me look a little closer. I think it was the very staged first question that bothered me most, and yesterday I finally took the time to watch it through a critical lens. It’s an ad for a Tai Chi app, but I cropped the video to hide the brand because I don’t want to amplify it, I want to critique it.

Here is the ad:

And here are a list of telltale things that suggest it is AI.

1. Look at the opening image. The woman is talking at a 90° angle to the stage, and there is no one at the podium below her.

2. The ‘expert’ is a perfectly chiseled man who is never named. No recognition of him as an expert in the field… because he’s fictitious.

3. Obviously fake audience members. The first image shows a blurred bearded man who doesn’t seem real to me. The second image has a man wearing a partial microphone like the expert.

4. The painfully fake script.

“Isn’t a gym better?”

“Gym doesn’t work after 40.”

This isn’t necessarily evidence of AI, it could just be bad writing, but it comes off feeling very wrong and unnatural. It’s like there was an intent in the text to make the expert sound like English is his second language but his voice doesn’t carry that same suggestion.

5. Comments are turned off. There is no benefit in having viewers outing the ad as fake. It’s better to allow the ad to fool more people without being called out.

The reality is that I could pick this ad out as fake, but that’s only because it was done poorly. We are going to see a lot more ads done this way and they are going to be good enough to fool us completely. It’s just a matter of time, and that time is approaching very quickly.

Do not amplify if you can not verify

This is a simple, but potent message. Before hitting the ‘Like’ or ‘Share’ button, before telling someone about the interesting fact you heard online, verify it in some way. Is it true?

Do some Ground Truthing. Can you verify the claim? Is it real or AI? Is it worthy of your amplification or are you just contributing to the spread of something unworthy to be shared.

How much better would the internet be if everyone paused and verified what they were sharing before amplifying misinformation, disinformation, fake news, and AI deep fakes?

It’s a worthy and effective mantra:

Do not amplify if you can not verify.

Isn’t this amazing?

I saw this video on TikTok last night and thought it was amazing!

I was wrong, but also right… let me explain.

I was initially amazed looking at a phenomenally muscled 78 year old man. Wow, what a body he has, it’s amazing what’s possible!

Then I went to the account, is it the account of this fit, old man or was this just a video on the account highlighting him? The next few videos on the account didn’t only show that it wasn’t his account, but they showed men way too muscular and disproportionately sized to be real.

So I went back to the original video and then noticed the audience members applauding… totally fake. I was fooled.

The body is fake, the video is fake. The AI rendering of the bodybuilder is amazing.

Sure the rendering of the audience members applauding was faulty, but wow, I didn’t think twice about the validity of the video on my first viewing. And in another 6 months even the audience rendering will be perfect too.

This video was indeed amazing, just not for the reasons I initially thought.