Deepfakes & The Turing Test

• Deepfakes

• Turing Test

• Erosion of Trust

• Liar's Dividend

Deepfakes are media (videos or audio) that mimic reality so well your eyes/ears can't tell the difference. The official definition focuses on mimicking real people or real scenarios, such as creating a voice clone of your friend's voice, or making fake videos about a disaster that occurred and spreading misinformation about the actual event. These deepfakes seem so real and convincing that our natural eyes can't distinguish fact from fiction. Though they can be entertaining (Have you seen any of the Will Smith eating spaghetti stuff?), they can also be problematic. The focus is typically trickery and misinformation. You're not just trying to make something fake seem real, you're trying to make someone believe that something happened that didn't.

• Actors playing roles they never actually played and Performers at concerts that never happened ("Ghost" celebrity endorsement, identity theft).

• Politicians saying/doing things they never said/did. (political sabotage)

• Political riots that paint an inaccurate portrayal of that party (fabricated civil unrest).

• Footage of historical events that never occurred (historical revisionism)

• Students/peers/colleagues "caught on tape" doing things they never did (nudification, slander, academic performance spoofing)

• Medical professionals encouraging you to buy products - but they're not actually medical professionals (Deepfake doctors)

• Phone calls about family in trouble and in need of money, but the family member's voice is a voice clone (Ransom scams, voice clones)

I tend to expand my definition of deepfake to include non-real people and places as these are still meant to fool you into believing something that never happened or something that isn't real - and the media tends to agree with my definition. If the item passes the Turing Test (See Turing Test defined below) and has no disclosure to help the user know it's not real, I'd say that's a deepfake.

You may have heard of Alan Turing - he's a famous Brittish mathemetician that helped crack the Nazi Enigma Machine during World War II. Computers as we know them today did not exist back then, but some amount of machine computing had been something that was being developed for a while, and these computers had people starting to question whether computers could think. To address this, Alan Turing created a thought experiment called "The Imitation Game". In this thought experiment, you would have someone monitoring communications through a terminal (screen) with two hidden participants, one would be a human, and the other a machine. The person monitoring these communications would need to determine which was a person and which the machine. For decades this was a trivial matter as computer technology grew rather slowly over the remainder of the 20th century, and therefore computers weren't all that great at fooling anyone into thinking it was human - All AI at that time was rule-based, meaning that every choice it could make was pre-programmed by a computer programmer and therefore there was a limited number of potential responses that could be made - not like the AI we have today that use machine learning and neural networks, and can generate a near endless number of content; no longer just text, but images, audio, and video. 

Now, especially in education, everything seems like a Turing test - we are constantly trying to figure out what is real and what is made by a computer. Think about this for a second. If you had a friends texting you but they used AI to generate all their responses (with nothing but your texts as input), would you be able to tell that it was AI and not their responses? If you were shown a picture of your friend, how would you determine whether the picture was an actual photo of them or AI generated? Could you tell if a person on the phone with you was AI generated?

• In 2021 a cheerleader's mother generating fake images and photos depicting her daughter's teammates doing inappropriate things in order to get them kicked off the team and get her own daughter promoted in their place. (ABC News - "Cheerleader's mom created deepfake videos to allegedly harass her daughter's rivals", March 2021)

• An ongoing series of deepfakes of Elon Musk endorsing crypto schemes since 2022, some going viral, causing billions of fraud losses, such as one reported in 2024 (CBS Texas - "Deepfake of Elon Musk Contributes to Billions in Fraud Losses in the US",  Nov 2024).

• A political consultant used AI to clone President Biden's voice during the 2024 New Hampshire Primaries (Adaptive Security - "5 Deepfake Examples of AI Videos and Images That Blur Reality"June 2025)

Deepfakes create an Erosion of Trust. When you can no longer trust what you see, what do you trust in? If you can't tell fact from fiction, how do you determine whether something actually happened or not? Fake videos are questioned as real, and real videos are questioned as fake. - and that's what makes this twice as hard. The "liar's dividend" is a benefit to dishonest people - they can claim that their misconduct isn't real - it's AI generated (even though it really wasn't). They now escape accountability until proof of the reality of the video is made. 

With deepfakes, you run the risk believing a lie and of doubting the truth.

Can you tell the difference between AI & Reality?

• staysafeonline.org - Safeword

• The NY Times - This Photo Is an A.I. Fake. Can You Tell?

• The NY Times - Who's a Better Writer: A.I. or Humans? Take Our Quiz.

• The NY Times - Test Yourself: Which Faces Were Made by A.I.?

• The NY Times - A.I. Videos Have Never Been Better. Can You Tell What's Real?

Would these have fooled you? Check out Tavus

• NY Times - Cascade of A.I. Fakes About War With Iran Causes Chaos Online (Mar, 2026)