AI: Fantasy vs. Reality

How do we know if an image, a video, an audio file is real? And when should we care?

Generative AI: Gen AI

Generative AI refers to images, audio, and other media and data generated by the use of AI tools. “Gen AI” is an abbreviation, not the label for a generation growing up with AI.

This picture is not AI-generated. I took it. Think about how that changes the story. A person has a white horse on their roof in Tribeca. What if I prompted an AI tool to generate an image of a “white stallion on a city rooftop”? Who’s crazier? The real person who put the real statue on the real roof? Or the person who imagined such a crazy scenario? Or an AI tool that is hallucinating, aka confabulating?

@ connie connors

DEEP FAKES

Almost Real

Almost every day, there is news of a “deep fake,” specifically someone/something pretending to be someone else through audio or imagery, coupled with specific actions, to maliciously manipulate the recipient. Bad. A few weeks ago, it was someone impersonating Marco Rubio. Last week, it was a Wells Fargo scam that was so real in terms of “knowing” the recipient that it was hard to tell. What you should know is that recreating someone’s image, including their movements and the sound of their voice, is very easy to do. How? For one, the average American is captured on camera more than 75 times per day, according to Safety.com.

There are TED talks and many other sources that show you how to spot a fake. If “it happens” to you, the first thing to do is not overreact. When you finally notice the very tell-tale signs, you’ll laugh at your momentary hysteria. Why do her hands look like feet? Why is my friend suddenly speaking without a single “umm?” And then, if it didn’t “happen” to you, why should you care? Suffice to say, learning the difference between fantasy and reality started way before the advent of AI.

Almost Profound

Here’s a fake, but relatively benign ad for Puppramin, a pill for puppies, made with the app Veo 3 from Google DeepMind (What’s with all the deeps? Naming is a struggle for many new industries in general, but that’s a whole volume of newsletters.) Neither the product, the pharmaceutical company, nor the people in the ad are real. Okay, re-watch it. Those are not real people talking about a not-real antidepressant for puppies. Who cares? If I put down hard cash for a product that I would never see, I’d be pissed. But it’s no different than a really good sales pitch, is it?

From an advertising industry perspective, this is profound. This commercial, by all good guesses, would have taken more than three months and $500,000 to produce (and more than 50 people from idea to storyboards, sourcing vendors to production and testing). This version took less than a day and cost less than $500 to produce.

EVERDAY AI TO TRY: Better than a fresh box of crayons

This is why I end my newsletters with “stay curious.” I get so giddy playing with the simplest of AI tools. I’m going to stay tool/app agnostic, so for this one, I tried Google’s Gemini, another free (to a certain level) Chatbot or LLM (urban dictionary is coming soon), and asked it to create an image of this, and then further refined my query. This is where time savings can become a time-suck, especially if you love that box of crayons and asking, “What if?”

TRY IT ON: Facetune

A few years ago, my youngest son posted a picture on his LinkedIn profile that looked like a professionally produced (and touched-up) headshot. He didn’t divulge his secret. I searched through dozens of apps — there are easily 100 — and settled on FaceTune (free for the trial, but note that “free” for these apps is changing as popularity increases). With any image-altering tools, you can decide which attributes to change or enhance with many, many features — too many to become truly proficient. So, take it easy. Luckily, it’s not plastic surgery, and “reset” is an option. Again, “almost” is good enough. Here’s my example:

I asked it to “smooth lines”

NEWSLETTER FROM PROTEST TO PROPOSAL

If you haven’t had a chance to read the last newsletter, take a skim. Asking ChatGPT or any other Chatbot (Gemini, Grok, Copilot, Claude, etc.) to reframe an argument is a great stress reducer. I am not a proponent of using AI as a therapist, but for me, blowing off steam by giving me a sane argument or counterargument to any situation, large or small, can really diffuse emotional energy that is perhaps best used doing something good for you. 😄 

Until Next Time

Stay curious, calm but cautious, always reach for a fresh box of crayons, and remember, almost intelligent is more than enough.

Connie

UP NEXT: AI can save time in planning travel. Did you know that the average person spends 16-18 hours planning a vacation? (Source: Priceline)