Can’t access the audio? Click here. Subscribe to the City Times Media podcast.
City Times Sound presents Thoughts, a podcast hosted by Joel Garcia where he gives his personal opinion on a variety of relevant topics. In this debut episode, he discusses the topic of Artificial Intelligence. He highlights the positive and negative uses of music, images, videos, and text created by AI. A full transcript is available below.
Thoughts and CT Sound are part of City Times Media, award-winning, student-produced digital news, TV news, magazine, radio/podcast and indie film, all on one platform at San Diego City College. CTM offers hands-on experience for students using new and emerging technologies in state-of-the-art facilities.
Transcript
Joel Garcia 0:00
In the world of today, people have opinions. For me, I have thoughts. Hi, I’m Joe Garcia. Welcome to thoughts.
Joel Garcia 0:11
Thoughts is essentially a show about well, my opinion, whether it be technological base, sports base or anything. On today’s show, I’ll talk about something that has become very prevalent, Artificial Intelligence.
Joel Garcia 0:41
There’s a possibility you might have stumbled onto a video either on Tiktok, Facebook or YouTube, which has someone talk about a general topic, such as a comic book or the news. While listening to the video, you might notice that the speaker talks in a way where they do not pause to take a breath. They also don’t appear at all and said video.
Joel Garcia 1:03
Similarly, you might have spotted a picture of video which seems authentic until you look a bit closer, and now does either gibberish text or a six fingered hand. To a lesser extent, there’s also the use of text that might sound robotic.
Joel Garcia 1:18
I am, of course, talking about artificial intelligence, or AI, for short,
Joel Garcia 1:24
AI artwork, video and audio have become prevalent in the past few years. For better or worse,
Joel Garcia 1:32
it has become all but impossible to avoid them. Several Tiktok videos, for example, start with a robotic voice declaring, check this out.
Joel Garcia 1:42
This is not limited to robotic voices or other tech to speech programs. It also includes the gray area of AI voices, which try to mimic actors, singers, influencers and even politicians.
Joel Garcia 1:57
Over on YouTube, you can type in a name, add AI to the search and get swamped by a wide range of results. In general, they are pretty flat.
Joel Garcia 2:08
The thing about AI voices is they are basically text to speech software which does not tend to emote. I recall years ago using Microsoft Sam and finding it funny how it said everything the exact same way, regardless of it being a positive or negative statement.
Joel Garcia 2:27
When it comes to songs composed with AI, they are limited to a single inflection with no way to alter their tone. The issue is less noticeable with someone who might have sung with the same general inflection, such as Frank Sinatra or John Lennon, it is much more blatant with other artists such as Michael Jackson and Eminem.
Joel Garcia 2:49
I have heard AI covers created by users which ponder how someone, such as one of those artists, could sing a song they never performed for Frank Sinatra, his AI voice tends to get integrated either with similar artists like Michael Buble or covers of modern pop songs made to sound old fashioned. It makes for a decent hypothetical of wondering, What if Sinatra was still around and covered modern music. But that can only go so far when someone tries to make Sinatra’s AI boy sing an original song like, say, a parody of Eminem’s Benham. It made him sound intoxicated.
Joel Garcia 3:30
I have also heard Genesis, land of confusion, covered by an AI Michael Jackson. It sounded off due to the AI trying to make Phil Collins that band’s lead vocalist sound like Michael Jackson, with the former’s distinct accent, making the latter sound British.
Joel Garcia 3:50
A similar issue happens with AI voices based on living singers. There are plenty of AI covers which use Eminem, all of which make him sound like he did in a song the real slim, shady after hearing more than one AI song using that inflection, it rapidly gets banal to listen. It also feels like a rejection of his newer music. But that’s a whole other story.
Joel Garcia 4:15
I want to point out. I have nothing against text to speech software. It’s an amazing tool used by those with the inability to speak. I just find it wrong to use AI to recreate other people’s voices without their consent and have them either sing songs they never performed or say something they never said.
Joel Garcia 4:35
If the artist did consent to such usage, that is fine. Prior to his death, the late James Earl Jones signed over to write to his voice to Disney so he could use it for Darth Vader in Star Wars. The next time Vader shows up in that franchise, there’s a chance he will still be voiced by Jones with the help of AI.
Joel Garcia 4:56
Otherwise, I imagine this can be used for things like having the. US, President, take your pick, discuss crypto, interviewing a deceased celebrity or something far more nefarious.
Joel Garcia 5:10
Moving on to a different medium, another use of AI would be through pictures. You might have noticed them while scrolling on either Instagram or Tiktok or even while walking around the mall,
Joel Garcia 5:22
the quickest way to detect an image composed by AI would be by how many fingers a character has on their hands. One of AI’s biggest flaws is how the software is unable to accurately create a hand with the right number of fingers. Suddenly, a human being can have anywhere between four to eight fingers on one of their three hands. Ai artists will sometimes try to hide this flaw by having their hands conveniently out of frame or tucked in.
Joel Garcia 5:53
Another obvious method is through wording. By that I’m referring to the gibberish text seen in these images, they might have a logo that looks fine, such as a McDonald’s M but then there are those which have signage or some other text that either tries to spell out a word, like Popeyes, only to wind up with poop eyes.
Joel Garcia 6:16
Then there are images which try to pass off as real photos. A great example of this happened back in January during the Los Angeles fires, when several people posted AI images as their so called proof of the Hollywood sign being on fire. This was despite the fact actual TV footage on the fires showed a noticeable lack of fire on the sign itself.
Joel Garcia 6:41
This leads me to another form of AI video.
Joel Garcia 6:46
There are plenty of them online, from well SD eating spaghetti to a certain controversial holiday ad from cocatola.
Joel Garcia 6:55
Some might look decent or even fool you into thinking it’s real, but there are obvious signs of fakery. In addition to the aforementioned flaws with AI photos, AI videos struggle to keep a character moving around naturally, and the backgrounds constantly change within every single frame,
Joel Garcia 7:14
One can easily look up videos labeled “ruined by AI” which use footage from either a film show, or even a public event, and have aI keep the sequence going. One good example available on YouTube is a video in which AI is used on a trailer for the first home alone. Some of the many odd changes AI makes to the trailer include Freddy Krueger from the night round Elm Street series chasing this film’s kid protagonist, Kevin, former President Joe Biden, walking into a scene with ice cream and lots of random, nonsensical explosions.
Joel Garcia 7:52
Last but not least, there’s AI text. Unlike the other forms of AI, which can show their flaws in some form or another, text can be a bit trickier to spot. It might be obvious by some misspelling or bad formatting, but that can also be caused by a poor writer. I have personally used AI text for the occasional eBay listing, but I can say it is not perfect. Sometimes it can badly describe a listing or come up with something completely different.
Joel Garcia 8:26
Ai text does have its uses, most especially for transcribing audio or video recordings of Film TV or interviews. In that case, it is easier to use such software to produce a transcript rather than go back listen and type out every single word in a time consuming ordeal. Now, the software is not perfect and can misconstrue certain words based on what it hears. For example, it might confuse the word odds may terminology used to refer to the 2000s for either odds or odds, because they all sound similar to one another.
Joel Garcia 9:07
While AI does have several positive uses, it is sadly used as a tool for lazy people.
Joel Garcia 9:16
Whenever I look at something really produced by AI, I always wonder, why did the person or company who created that chose to use it rather than go through the process of hiring someone to compose either a song, image, video or even a text message? They just told a computer to do it for them. Sure someone hired to compose something may make mistakes, but at least they would know how many fingers someone has how to create legible text and make an end product with more effort than just typing it into some program.
Joel Garcia 9:49
For those who think that might be an overstatement, it leads me to the following question, How could some AI software produce something that is equally as creative as. Something produced by either someone or a group of talented people.
Joel Garcia 10:05
Think about that.
Joel Garcia 10:06
And on that note, that is all the time we have for today’s show.
Joel Garcia 10:19
Thoughts is produced by Joe Garcia of city time sound. Thoughts and CT sound are part of City Times media, award winning student produced digital news, TV, news magazine, radio, podcast and indie film all on one platform at San Diego City College, CTM offers hands on experience for students using new and emerging technologies and state of the art facilities.
Joel Garcia 10:47
To read more, visit SDCityTimes.com.
Joel Garcia 10:50
You can also follow me. Joel Garcia on social media, apps. MrJoelGarcia9 on Instagram, threads, Twitter, slash x as well as at MrJoelGarcia on TikTok.
Joel Garcia 11:11
But until next time, thank you for listening. You
