How to spot what’s real, fake with AI: expert

Mike McCready, the associate dean at NAIT's school of Media Innovation Sciences, shares tips on how to spot what's real and what's fake online.

The rapid emergence of artificial intelligence (AI) in our daily lives has been useful for some and confusing for others. People have been exposed to AI-generated content, including pictures and videos. Mike McCready, the associate dean at NAIT’s school of Media Innovation Sciences, joined CTV Your Morning Edmonton’s Kent Morrison on how to spot what’s real and what’s fake online.

This transcript has been edited for length and clarity.

Kent Morrison: Do you have some telltale signs for what people should be looking for when they’re examining if something is AI or not?

Mike McCready: These signs are getting hard to use because AI is getting more powerful. With photos, look at things that seem a little bit too perfect. Glossed over skin, maybe the textures are inconsistent, fingers are a telltale sign, those are some of the ways with photos. With videos, it could be that the lip sync is not quite matching up with the audio, some of the textures are not quite correct, also the physics, such as objects occupying the same space. Those are some ways to tell if videos are created with AI.

Kent: Talk us through this photo and what we should be looking for.

Mike: This one is one I created to celebrate my 50th birthday using technology. Some of the ways they can look for AI is that it seems a little bit too perfect, and some of the skin tones are a bit blurred over, although this one did take a lot of time to create and it’s pretty hard to tell. One of the other ways to look is to do reverse image searches on Google. You can look online and see where that image might have been sourced from. Ultimately, we have to be vigilant and not too trusting when looking at content.

Kent: Walk us through this first video example.

Teddy bear in water video

Mike: This is a video from one of Google’s early research projects for text-to-video prompts that they created in 2022. You can see the frame rate is a little bit slower, it looks more animated. This is an early one from 2022, which is not as powerful as the models that are today.

Kent: Walk us through the second video.

Google’s Veo 3.1 video

Mike: This one is actually from Google’s latest model Veo 3.1, and it just came out this year. It’s a lot more challenging to tell. The reflections are consistent, the audio is actually present, it’s harder to tell if the lip-syncing is incorrect. AI is becoming more challenging to identify, and it’s up to us as citizens to be able to use our common sense to challenge everything and not take things at face value.

Kent: Is this all bad at this point? It seems to have a negative connotation.

Mike: No, there are a lot of great things. For me, it allows us to empower the creator to go beyond their original ability. It allows people to brainstorm, to ideate and create new content that doesn’t exist. There are a lot of positive outcomes of using generative AI.

Kent: What would you say to people who are looking at this and would rather just run and hide from it all than try to deal with it?

Mike: The reality is this tool is not going away, similar to the internet. The more prepared you are to identify and use these tools, the better you will be in business and in life.