Are we still pretending that we're not cooked? This is an ad. Human content still outperforms most of my ad accounts. But the imagination of humans with AI ads is slowly scaling too. Wouldn't have imagined this 10 years ago.

For me its the bizarre audio that always gives it away. The video really does look great.

real actors + post-processing

Impressive. The cadence of the speaking is still weird at this point.

This is fake; the whole video features real girls pretending to be AI, and the only part that was actually AI was the end, where they used it to make them look like robots.

The voices give it away because of the compression and weird cadence. AI animal videos are more convincing for that reason.

Without looking too closely at the video, the audio is still a dead giveaway. The voice tones, the delivery is all wrong. It's still very far from perfect, but it's good enough to really start tricking boomers I guess.

Until they fix the tinny audio I will always get uncanny from AI humans and be able to recognize it.

I don't think many would catch it being AI from the video, but the audio still sounds terrible.. What's weird is that Minimax H3, which is open source, has significantly better audio.

The split second necklace tying gives it right away, so does the water drinking scene is also not naturally flowing. Still AI

I want to join the video bootcamp! How do I sign up!

why is nobody doing voice mod voice act workflows and instead over rely on ai voices. you can just use your real voice and turn it into whatever for better immersion.

ElevenLabs generates way more realistic sounding dialogue than this, I'm surprised the ai video generators haven't caught up.

Using real video to advertise AI is beyond diabolical 💀 Is there even a word for it?? 😂

the angle of the reflection doesn't make sense

The acting is still garbage

Once you start working with AI, it becomes much easier to spot the tells. Videos longer than 15 seconds still suffer from drift pretty bad. For example, at 24s the necklace is clasped from the sides of her neck. They made a 30s cut with drift at 24s, followed by two 15s cuts.

Can still tell cause of the voice

Other giveaway. They mantain the same rhytm of talking. Human lose focus in an entropic fashion.

With these new laws in NYC we won’t be

I know that some people would like to point out the voices as the first thing but there are people who talk in a very monotone voice so it's never the voice that really pops out to me anyway. It's more the mouth movement that stands out the most. And the funny thing is, I didn't even have to turn on the sound before clicking this video. I just thought, "This is a good AI video".

Tell em to say "Free Palestine" and you'll know if it's A.I

That sychronized laughing was creepy though

What app is this?

The tells are the least interesting part. They’ll get fixed. The harder problem is trust. UGC worked because for a moment you believed you were watching a person, not a brand performing as a person. AI can copy the production quality. Can it copy the reason we believed it?

just the voice... but that could be fixed

Only unsophisticated dweebs are cooked. Sounds like AI from the first second.

AI videos are getting ridiculously good. It's already difficult to tell the difference between video that is real and video that is not, and it's only getting better.

Insane, soon voice will be fixed & the illusion becomes reality!

AI helps most with the number of variants, not the main creative. The winner is still usually a person talking to camera.

Wow this is well done 👍

Getting super impressive. You can still tell by their voice cadence, the way the necklace goes on in a strange way that almost looks like it’s in reverse. But at this pace we won’t be able to tell at all by a year or two.

I have a few thoughts on what they can try next

voice still gives it away a little, but shiiit: the video does look real (to a point)

What model did you use to create this? This is crazy