AI Myths Exposed: Unraveling the Truth About Deepfake Tech

In an earlier blog, Chris Roberts discussed Artificial Intelligence (AI) at a high level, providing general information about what it is, how it’s being used today and how it impacts our daily lives. This blog will address some questions the ThirtyFiveSixtyFour team received from listeners after our first episode discussing AI.

What’s a deep fake?

A deep fake is generated on the dark web and is the result of AI being used to create a fake image, video, audio, or even a robocall script and connected or attributed to a public figure. The most egregious example from recent events is the explicit Taylor Swift deep fake image. There was also a dental ad video using an AI-generated image and voice of Tom Hanks to promote a dental service. While it was not explicit, it was just as insidious. Because of its seemingly innocuous purpose, people are more likely to accept it as approved by the public figure represented in the image.

To generate the deep fake images or audio, a bad actor can use an open-source AI tool that doesn’t impose guardrails that limit such nefarious outputs. That bad actor has the ability to run the model on their machine and train the open-source AI tool to create the deep fake outputs that the Taylor Swift image and Tom Hanks ad represent.  

Closed source AI tools, such as Claude from Anthropic, Open AI, and Google Gemini, have guardrails in place to prevent such false images and content.  As AI technology gets better, regulations to help protect users and the public will become more rigorous as well. But even with that, deep fakes and fake imagery, will continue to exist because the possibilities are really endless with what one can ask of and create from AI tools.

The bottom line: If you hear or see something that seems far-fetched, take it with a grain of salt. Do your homework before you simply believe it and run with it. 

Can I trust Dr. Google? AKA: Should I be worried that my doctor is using AI?

Not necessarily. Used responsibly, AI can help your doctor more quickly and accurately diagnose your illness or injury by scraping the web for the most recent articles and studies related to your complaint. A recent study by the American Academy of Family Physicians revealed that physicians using ChatGPT to help their diagnosis reported 98% accuracy in the answers they received. AI can help a physician enhance their knowledge within their specialty and keep pace with the latest techniques so that they can spend more quality time with their patients. It can fill in some details from areas outside their specialty to prompt the physician to consider additional diagnoses and recommend treatment plans.

While AI outputs can help a busy doctor to more rapidly evaluate the patient’s treatment needs, it can also return erroneous information. In the aforementioned study, the error rate in the study was low – only 4 in 140 responses per the cited study. This just underscores the physician’s responsibility to carefully consider the results of an AI-generated query because the incorrect information could negatively impact the patient’s recovery or the physician’s ability to care for their patient.

Is AI really bad for the environment?

One might think that since AI is virtual it shouldn’t impact the environment. This is not necessarily true. While computers are much smaller than when they were first introduced there are many, many more of them. Computers used to take up entire rooms, entire buildings and now they are phones and laptops as well as large data center computers. Whenever compute power is needed, the computers in the data center use electricity and generate heat. Then, the computers generating heat require cooling. All this consumes energy and expends heat.

When someone makes an AI query, that AI tool generates a lot of compute power because that query is accessing not just one processor but multiple processors to hunt down and grab the data you’ve requested. Therefore, one AI query can consume a lot of compute cycles. Multiply that by one hundred, one thousand, ten thousand queries. Processing increases. Heat gets generated. Processors must be cooled.  It’s a vicious cycle. So it’s important that AI providers and data center providers discover ways to lower power needs.

What’s the big deal about Nvidia?

Nvidia was one of the first, and is still one of the biggest, processor manufacturing companies in the world. It was started to do parallel processing which is critical to graphics processing. Any gamer knows that Nvidia has pretty much cornered the market in gaming processors. As a pioneer in computing, it’s no surprise that it’s also a major player in AI processors. The company is the number one provider of Graphic Processing Units (GPUs), the processors that are responsible for most of the AI compute being done today.

A GPU chip is an important piece of the AI food chain and is incredibly expensive, in the tens or hundreds of thousands of dollars. To build an AI-processing data center you need many multiples of GPUs.  Which is why companies like Intel, Nvidia, AMD, Amazon, and even Google, are entering the GPU game.

What’s the open letter against AI in the creative industry about?

On the heels of the 2023 Hollywood writers’ strike a group of prominent musical artists and actors signed an open letter to industry leaders demanding limits be put on the predatory use of AI in the creative industry – music, movies, TV, etc. AI cannot do what it does without the input of human content, but it is very good at mimicking a certain artist’s musical style or a scriptwriter’s writing styles. AI can:

  • generate a song that very closely matches the artist’s style
  • compose a complete symphony
  • score the soundtrack for a movie
  • write scripts for movies and TV shows

The artists want to ensure their creative rights and privileges are protected under their contracts. To this end, two screenwriters unions, the Federation of Screenwriters in Europe (FSE) and the International Affiliation of Writers Guild, have passed a resolution calling for the ethical use of AI. Among other clauses, the resolution demands that only human writers be granted copyright for original works.

Listen to Chris Roberts and Karen Stones in Episode 8 Artificial Intelligence Q&A: Deep Fake Nudes, Music, Dr. Office Visits, Jobs and more as they talk about these topics and a few more in detail.  Then, stay tuned for more AI knowledge drops from thirtyfivesixtyfour.com. and Chris Roberts.

ThirtyFiveSixtyFour
ThirtyFiveSixtyFour
Artificial Intelligence Q&A: Deep Fake Nudes, Music, Dr. Office Visits, Jobs and more...
Loading
/
  • Mary Cook

    Mary Cook, also known as “MC” and “Mother Mary,” is heralded as one of the world’s few content whisperers. She is the creative force and Marcom Director at ThirtyFiveSixtyFour. Armed with a degree in English from UCLA, Mary is not just your average wordsmith—she’s a grammar nerd with a penchant for storytelling that captivates and resonates.

    Born into a big, close-knit family with seven siblings, Mary is committed to keeping family connections and gatherings alive with boisterous fun and games. Mary brings a lot of energy to everything she does. She’s as dedicated to her role as Marcom Director as she is to her role as favorite auntie to her 22 crazy, loving nieces and nephews.

    A life-long athlete, Mary’s passion for sports has transformed into a love for the adrenaline rush. When she’s not weaving words for our podcasts, you’ll find her carving waves on a jet ski or navigating desert trails in her RZR. Mary’s adventurous spirit is as diverse as her ability to craft compelling narratives for our audience.

    In a world that often craves attention, Mary thrives behind the scenes. Her meticulous attention to detail and commitment to excellence are the driving forces that elevate our Marcom strategy. As the wordsmith-in-chief, Mary ensures that every piece of communication reflects the essence of ThirtyFiveSixtyFour.

    View all posts
About the Author

Mary Cook, also known as “MC” and “Mother Mary,” is heralded as one of the world’s few content whisperers. She is the creative force and Marcom Director at ThirtyFiveSixtyFour. Armed with a degree in English from UCLA, Mary is not just your average wordsmith—she’s a grammar nerd with a penchant for storytelling that captivates and resonates.

Born into a big, close-knit family with seven siblings, Mary is committed to keeping family connections and gatherings alive with boisterous fun and games. Mary brings a lot of energy to everything she does. She’s as dedicated to her role as Marcom Director as she is to her role as favorite auntie to her 22 crazy, loving nieces and nephews.

A life-long athlete, Mary’s passion for sports has transformed into a love for the adrenaline rush. When she’s not weaving words for our podcasts, you’ll find her carving waves on a jet ski or navigating desert trails in her RZR. Mary’s adventurous spirit is as diverse as her ability to craft compelling narratives for our audience.

In a world that often craves attention, Mary thrives behind the scenes. Her meticulous attention to detail and commitment to excellence are the driving forces that elevate our Marcom strategy. As the wordsmith-in-chief, Mary ensures that every piece of communication reflects the essence of ThirtyFiveSixtyFour.