Generative AI: Ethical Concerns and Unchecked Advancements

AI is rapidly advancing, training on copyrighted content without permission, creating ethical and privacy concerns, and devaluing human creativity.

### The Dark Side of AI: Unethical Practices and Potential Dangers
* 📸 Microsoft's Co-Pilot feature captures screenshots every **5 seconds**, potentially exposing sensitive information, despite being **optional**.
* 🎨 AI art contests are being infiltrated by AI-generated submissions, with some even making it to the **final rounds** before being detected.
* 🔍 AI detectors are being developed to distinguish between AI-generated and human-created images, but they are not always **accurate**.
* 👥 AI is being used to generate fake images and videos, potentially leading to **historical revisionism and other dangerous consequences**.
* 🖥 ️ Tech companies are using **vast amounts of user-generated content** to train their AI models, often **without explicit permission**.
* 🌐 Google's new AI search feature, Gemini, is facing criticism for providing **inaccurate and potentially harmful information**.
* 🎶 AI-generated music is becoming increasingly difficult to distinguish from human-created music, raising concerns about **copyright infringement**.
* 🌱 The environmental impact of AI, particularly generative AI, is significant due to its **high computational requirements**.

where do I even start? Do I start with Microsoft forcing a feature that takes screenshots of everything you do on your PC every 5 seconds and puts it into an easily searchable database? Do I start with art contests being taken over by AI art? Or do I start with companies quietly updating their privacy policies to use customer work to train AI and only revising those policies after major backlash, to which it might be too little too late? What about companies like Apple and Nvidia training their AI on thousands of hours of YouTube content without permission, allegedly? Please don't sue me. The onslaught of every tech company racing to implement some half-ass AI feature to artificially pump those stock prices (pun intended) is in full force, from Google Chrome asking you to create a terrible blurry theme with AI that looks so [__] bad to MasterClass.com's useless class ta that is about as helpful as your mom giving you job search advice. In 2024, it is now everywhere. You can't escape some random AI feature on some random website.

you never would have thought would need it like this notion document or social media sites. Meta AI is especially not useful. The search can tell you Ariana Grande's follow count, but not your own. It can give you cat videos, but won't search for a specific one. I tried to search up a popular meme, this is the move that made LeBron cry, but it just gave me videos of LeBron James crying. So then I added more details and specifications, and it still gave me videos of LeBron crying. So I used the normal search to do this, and I found the video I was looking for in like 2 seconds. Athletic expertise. Oh, I get it. I don't look like the average athlete, but I am an absolute [ __ ] unit on the basketball court. That was the move that made LeBron cry. Why is this here? How does this add to Instagram or make my experience better despite what new technologies are setting out to do? AI is creating new problems faster than it can solve them. Whoa, whoa, whoa, before you get all upset.

about some robot drama. I've got something special designed by human artists. My First Merch drop ever our gamer desk mats, and I'm so excited. I'm so happy with how these turned out. We got karaoke on the moon and tour of the Galaxy, both in two colorways. These are thick, detailed, and super high quality. They'll be on sale starting August 2nd, 2024 for patrons and August 9th for everyone else. If you want a deskmat that literally no one else has, this is for you. AI can't make this [__]. You'll find the link in the description, and if you want to be reminded when the desk mats come out, you can put your email in on the website. I'm not going to spam you. I'm not going to use it for nefarious purposes. I'm not going to sell it. Don't worry, okay, back to the video. Thanks. I feel like more people need to know about the disaster that is Microsoft co-pilot. This is a Windows 11.

AI feature that captures constant screenshots of what you're doing on your computer, whether you're showing your passwords, sensitive work documents, scrolling around in your bank accounts, private messages with your friends and family, video calls, watching xqc, any embarrassing or sensitive thing. Co-pilot stores screenshots in plain text on your machine and uses what they call recall to help you pull things up easier. Sure, those screenshots don't get sent anywhere and stay on your computer locally for now, but do you know how easy that makes it for hackers to get a hold of all of your sensitive information? Now, before recall officially launched, it already got hacked. Since all of that screenshot information is saved in the same folder on everyone's computer, hackers can easily write a script to locate and steal those snapshots. Microsoft misled press by saying that your data was encrypted, but when you're using the computer, it's not encrypted anymore. If this sounds terrible, that's because it is. Maybe they should recall recall more from me later with all the

backlash from the nerd Community. They made this feature optional, but it's on by default. Tech companies lie about privacy all the time, like how incognito mode isn't actually Incognito, but this is an absolute failure in cybersecurity. Take it from me, a cybersecurity expert over here. As if I needed another reason to distrust Microsoft. Blit scaling is prioritizing speed over efficiency. How do you go as fast as possible in order to be the first to scale? Does AI Blitz scale well? Certainly seems like it today, doesn't it? Figma, a tool to design user interfaces, not to be confused with words like ligma or even Sigma, launched figma AI to create mockups of apps and had to quickly take it down because it was generating app mockups that looked eerily close to Apple's weather app. So you can guess what their AI was trained on, but now they're taking the time that they should have in the first place to refine it. It was rushed to meet a deadline. I can't use Photoshop without Adobe begging me to use their AI features.

and have fully embraced it. From launching Firefly, their generative AI, with plans for 3D and video, look what I can do to expand this Pokémon card art. Now, with some tweaks, I can enter this into an art contest and takeaway spots from actual artists who put hundreds of hours into their work. And that's close to what happened. Pokémon's 2024 illustration contest had over 10,000 entries. When they announced the 300 finalists, people were quick to notice some suspicious artworks under similar aliases with the same initials V and K, breaking the entry limit rule, and had weird inconsistencies that are typically attributed to AI work. Pokemon Trading Card Twitter acknowledged this and removed those entries, giving back the spots to actual artists. I don't blame the judges for not catching on to this, as this is one of the first years that AI entries could really be a large issue, but it just goes to show how they almost got away with it with stuff like this. People also started accusing other people of using AI in

their artwork. AI art has become so much of a problem that there are now AI art detectors online. And I'm sorry for referring to it as AI art. I don't want to, as art is literally the expression of human creative skill and imagination. But these AI art detectors, while have good intentions, are still wildly inaccurate, whether that be accusing human art of being generated by AI or vice versa. When I was making my last video, I needed a dinosaur jungle background for a bit. Almost every image on Google was AI generated. I almost thought this one was human-made, and until I was suspicious of this tree log bending into the water, the dinosaur's arms were weirdly long and uneven, the birds are kind of meshing into each other, the mountains point in weird directions. All of this screams AI. So I put it in one of these detectors, and it told me this image was likely human-made. But just to be extra sure, I located this image in Adobe's stock photo library, and it stated that I was

correct this was generated with AI and so is this one, and this one, and this one. I love the idea of having to pay $80 for images that someone just generated in 5 seconds and looks horrible. I just decided to draw it out with Microsoft Paint instead and thought it turned out pretty good. But this is an image of a dinosaur, something that most people wouldn't think would be misleading to be real. But AI images can be misleading to the point where it's actively rewriting history. In Drew Gooden's video on this subject, he cites a documentary of a murderer that used AI to generate images of her that never existed for the documentary without telling people it's AI. Of course, looking at this cursed image, you can tell that it's weird and not real. But when AI images are being mixed into our everyday lives in places we don't expect them, AI detection isn't at the top of people's minds. Like this factual-based documentary or this image of the Tank Man in Tiananmen Square.

isn't real. This never happened. What happens when fake historical images rise to the top of Google search results? Or when it generates images of Asian Nazis, black founding fathers, and popes? And don't tell me people wouldn't fall for this. Boomers still copy-paste chain messages on Facebook. There's going to be a co-pilot for every profession. And if you think about that, that's huge. And not professional activities, because it's going to write my kids' papers, write my kids' high school papers. Uh, yes, although the hope is that in the interaction with it, they'll learn to create much more interesting papers. Sure, buddy. And I wrote a much more interesting paper on The Scarlet Letter because of Cliff Notes, I'm sure. Asking AI to give you a synopsis on the book and then write the paper for you is great for absorbing material. And the destructive optimism from people in the industry directly reflects why it's being shoved in our faces in every application. I mean, one of the problems with the current discourse is that it's too much of the fear-based.

versus hope. Based imagine a tutor on every smartphone. I don't know, man. Maybe people are fearful because it's creating serious problems before regulations or laws can keep up. I think I have the right to be more fearful than hopeful when AI can not only rig art competitions but have even more serious consequences like changing history, creating an easier way for [ __ ] up weirdos to make AI point of people and celebrities without their consent or knowledge, scam people easier, or dangerous deep fakes of politicians during election season all unchecked. I have AI images of me floating around and these are the ones of me clothed, but if that's just me Mrs. worldwide, then imagine the more popular creators or celebrities. Imagine some freak at school or work doing this to you or one of your family members who don't have a legal team to help take it down. I'm over 18. How do you think child predators are working with AI and now with AI video improving at a rapid rate? Take all those

worries and times it by 10. Take this AI video. This is live rigging vtubers and filters have been around for decades now, but it's all getting more detailed and more realistic with machine learning and accessible to a scary degree. They really asked themselves whether they could and not whether they should. Anyway, look at my feet, brought to you with love by bombas. You're welcome. These socks are comfortable, durable, don't have annoying toe seams, have four-wheel drive, and most importantly, are extremely cute or handsome. Whether I'm relaxing in my heels or ready for a night on the town in slides, bombas has socks for every style and mood. The cool thing is with every purchase, bombas donates an essential clothing item to someone experiencing homelessness. They've donated over a 100 million items across their network of 3500 giving Partners across all 50 states. My viewers get 20% off their first purchase. Just go to bombas.com Gabby and use code Gabby 20.

for 20% off Gabby 20 20 off Gabby 202 off Gabby 2020 off thanks bombas again for sponsoring this video. I'm so excited to partner with them so thanks for checking out the sponsor links and back to the video. Sometimes AI gets things wrong without someone behind it having malicious intent and these are called hallucinations. You might have seen how it tells you to put glue on your pizza to help the cheese from sliding off. I want to talk about this term hallucination because it's a very human term. Why use such a human term for basically an AI. That's just making mistakes a lot of this General capabilities are actually quite humanlike um sometimes when we don't know the answer to something we will just make up an answer we will rarely say I don't know this is just a dumb answer and how is that the standard that they're allowing imagine a tutor on every smartphone we will just make up answer but also they keep using humanlike terms for it because they

don't want you to think of AI as just a computer algorithm. They want you to feel like it's an assistant, co-pilot, human. They want you to feel connected to it, which is why the CEO of OpenAI, Sam Altman, got roasted for dropping ChatGPT-4.0 with a voice that sounds weirdly horny. For you, I'm doing a live demo and, frankly, I'm feeling a little bit nervous. Can you help me calm my nerves a little bit? Oh, you're doing a live demo right now? That's awesome. No, I want you to tell me a bedtime story about robots and love. Oh, a bedtime story about robots and love? I got you covered. And they asked Scarlett Johansson to voice this AI, but she declined. And then when GPT-4.0 was released, people noticed that it sounded weirdly similar to her. I'm doing great, thanks for asking. How about you? Hi, I'm Samantha 4.0. And then exhale slowly. I gave it to myself, actually, in the movie 'Her' where she plays a human-like AI.

that develops a weird relationship with a lonely Guy. This is Sam Alman's favorite movie. He said they paused the rollout for this voice assistant as Scarlet Johansson sued OpenAI for this and called for more legislature around AI. But let's go back to the CTO of OpenAI. An AI that's just making mistakes. A lot of these general capabilities are actually quite humanik. Um, sometimes when we don't know the answer to something, we will just make up an answer. We will rarely say, I don't know. Hang on, I have to put on my [__] detectors. I've been saving these. How is this a valid explanation as your goal is to get AI on everyone's device? How about instead of AI making things up it doesn't actually know, you code it to tell you it doesn't know or that the answer is low confidence? I mean, people have noticed that chatbots are increasingly useless as I demonstrated with Meta's AI and Masterclass's AI being so dumb. Google's AI

Gemini is not safe either. When you searched something before Gemini, it would show you a top result preview with a link to the source, and that worked pretty well. Now Gemini gives you stupid answers, like telling you that pregnant women should smoke two to three cigarettes per day, when we know it should be more like 8 to 10. Or simple things like how long ago 1919 was - 20 years. They've rebranded their old search and made it worse at the same time because it's referencing satire sites like the onion and forums like Reddit posts as top results. The AI rush to the top very much feels like a race to the bottom. Fun fact: if you go to udm 14.com, you can go back to their old search before AI rushing. This stuff out is partially why they didn't refine it to be actually good. And while AI is improving now at an exponential rate, it's also bad and inaccurate because it's already stagnating. I'm sure they'll come up with something to make that.

untrue, but for now, AI companies are already complaining about not having enough data for AI training. In just a couple of years, but if you have access to basically all of the internet of data to steal from, and it's still not enough data, maybe it's a model problem and not a data problem. If the history of the Internet isn't enough, maybe AI isn't as smart as they want you to think it is. AI is only as good as what it's given by a human. AI cannot create something genuinely new that humans can't create. Did the model see any clips of Ferdinand to know what a bowl in a china shop should look like? Was it a fan of SpongeBob? Wow, you look really good with a mustache, Mr. Krabs. By the way, my prompt for this crab said nothing about a mustache. AI generates the average amalgamation of human work, and it only works off the backs of artists. So what's the end goal here? It's never going to be better than human-produced work because

human-produced work is all it knows. The best it can do is an uncanny recreation of things that resemble human work but aren't. And we'll look into exactly why a bit later in the video. What I can see it doing is saving company's money for things like stock footage where instead of sending a guy out with a drone, they can create new stock footage with a click of a button and a subscription cost. And it can trick people with scams, catfishing, non-consensual images, and videos. Do you want to live in a world watching footage of places that don't exist, things that never happened, and people who aren't real? The world is already isolating. [Music] Enough, I know the effects are crazy. It's dank. I could eat one of these every [Music] day. Sorry, I got you messy. AI companies are getting desperate for more content. The New York Times reported, by The Verge reported, by me, says they got so desperate for more training data. Open AI developed an AI called Whisper to transcribe over a

million hours of YouTube videos to train chat GPT. When asked about it, OpenAI knew it was legally questionable but believed it was fair use. At least a few Google employees, including the CEO of YouTube, have seen unconfirmed reports of them doing this and possibly using YouTube videos to train Sora, their AI video generating model. The OpenAI CTO was confronted about this, to which this was her reaction: What data was used to train Sora? We used publicly available data and licensed data, so videos on YouTube. I'm actually not sure about that, my sister in Christ. You're a piece of [ __ ]. It was confirmed that they did use Shutterstock to train, okay, videos from Facebook, Instagram. You know, if they were publicly available, um, available, yeah, publicly available to use, um, there might be the data, but um, I'm, I'm not sure. You know that she knows she's the CTO. Also, the CTO of OpenAI was saying how they train on publicly available content as if that means it's now available for commercial use. Publicly available does not mean

it's free to use an open AI. Isn't the only one doing this. An investigation by Proof News revealed that over 173,000 YouTube video transcripts were used to train AI from Apple, Nvidia, Bloomberg, who ironically have conducted some of these interviews I've been referencing. What happens is someone wrote some code to download YouTube video transcripts using YouTube's API if you're a nerd, and then they put it into a group called a data set. These data sets, data sets. I don't give a [ __ ] they get sold and moved around to train different AI models. And this data set had YouTube video transcripts from people like Mr. Beast, Jacksepticeye, Marcus Brownley, PewDiePie, Jarvis Johnson, Drew Gooden, Danny Gonzalez. Not mine for whatever reason. Was I not good enough? Wow. I just wrote into React. Wow. It stole all of these YouTube videos without anyone's permission. Not the creators, not Google, which by the way you'd absolutely have to pay for. But don't be fooled. There are no heroes in this story. Google.

is not innocent either. They're not protecting us out of the goodness of their hearts. They don't want other people downloading YouTube videos to train AI because Google wants it for themselves. Google is also using YouTube videos from its own creators to train its AI models. Now, high-quality data is an ultra commodity for how valuable it is to train AI. A Google spokesperson responded to this, saying that this was permitted under agreements with YouTube creators. All right, I'm tired. Time for a gabby with a different outfit. Outside of just video models, are being trained on just about everything without permission from art to forums and even news articles. The New York Times, Chicago Tribune, the Denver Post, other Publications and groups of authors are all suing Open AI and Microsoft companies who will probably continue to get sued for copyright infringement. But this is while other Publications like VOX media, the Atlantic, and Time have all signed licensing deals to Open AI to allow them to train off of their employees' work. Reddit too has signed licensing deals to hand over their

content to train AI people have to agree to terms and conditions to use Reddit, Adobe products, YouTube, and other websites. But they didn't know that these terms would be quietly and retroactively changed in order to exploit their contents and their work. At what point are these companies protecting their copyrights and IPs with an iron fist while infringing on individual intellectual copyright? I guess that's why there's a ton of lawsuits, but this is also hilarious because even if AIs do train on licensed content, which they tout as ethical, it still doesn't mean that it is. I could be naive in my thinking here, but I don't understand how companies have the right to license out other people's content. Posting sites are responsible for monitoring the content that goes on there, but that doesn't suddenly mean they own that content. Like that intellectual property is not theirs to sell. Like if I upload a song to Spotify, that doesn't mean that Spotify now owns all the rights to my song and can license it out to third parties. This isn't just people's data they're selling.

they're selling people's work as if they own it. I don't understand how this is legal at all. Stack Overflow, a coding website, did this and licensed out all of their users' code to be trained on AI. A good amount of people got so pissed that they deleted all of their code off the site and abandoned it. Reddit content being sold also explains why Google's Gemini keeps referencing Reddit posts as facts. It doesn't seem like they're vetting the quality or factualness of the content they're being fed and spitting back out. Don't even get me started on Facebook. That place has become a cesspool for AI. You like uh John Cena making a sandcastle of John Cena. Industry leaders and venture capitalists keep saying that AI is getting better and smarter as there's more data to train on, but as the internet is flooded with more data generated by AI, the AI gets trained on other AI, creating an average of an average with hallucinations in it. It just becomes AI on PCP. Where did this

mustache come from from where did this beer come from. Purdue University did a study and found that chat GPT was wrong over half the time when they tested it with programming questions, and 77% of the time it was full of [__] and 40% of the time people overlooked the misinformation. And yeah, it's wrong sometimes, but I wanted to know why, and it's because of fundamentally how it's built. Because in simple terms, generative AI is just trained to predict the next word. It doesn't pull from a large database of fact and knowledge. It's currently an enhanced autocomplete system, so it's more trained to just write plausible sounding statements and not actually be a tutor trained on factual information. It's kind of like asking that guy who thinks he knows everything but doesn't. So, like, half the time, he just spews out random [__] that makes him sound smart, but like he's not. It just, he just wants to sound smart. So everyone's encountered a crazy person who's drunk off their ass at a cocktail.

party who says really odd things or at least every adult has, and you know that's not like the world didn't end. So, since stealing content is clearly resulting in the grand slam of lawsuits for AI companies, they've turned to other weird solutions for running out of content, like creating synthetic data for them to train on, created by other AI models. They want to train AI on AI. This is just AI inbreeding. They want want this. So, we haven't talked about AI music yet. Musician and YouTuber Roomie had myself and a couple other very experienced and talented musicians on his channel to test whether or not we could decipher an AI voice or a human voice. And even though all of us have listened to thousands of hours of music and vocals, we had a hard time with like half of them. It's permeating through music faster than people realize or can keep up with. It took me way too long to realize realized that the BBL Drizzy sample was actually AI generated. A bit more obvious to tell, but this viral

audio is also AI-generated. And don't get me wrong, some of this [__] generated by AI is actually hilarious, but it gets you thinking about the future. If you go on Spotify and search up workday jazz or something of the sort, you'll get Spotify curated playlists that literally sound the same with first name last name or blank Trio artist with no bios or photos of themselves. And people have begun to notice this. So I checked it out for myself. [Music] [Music] Either Spotify's algorithm is really good at picking out songs with a similar vibe, or people have started to theorize that Spotify themselves might be generating AI music for Spotify curated playlists where all the streaming profits would just go back to themselves. Allegedly, this is not proven, and this is just me and people on the internet theorizing and speculating. When Spotify is already known for terrible revenue splits with artists, I can't directly accuse them of this, but it is fishy.

nonetheless, the AI company behind BBL Drizzy udio and another music generative AI sunno are actually getting sued by The Big Three record labels umg Sony music and Warner Records. So good luck with that godamn, they're [__] cooked, and I hate the way big labels enforce their copyright system as much as the next guy, especially on YouTube. You could hum The Melody of a song for 5 seconds and they'd steal all the revenue of your 20-minute video, but in this specific case, they are absolutely right because the only way for AI to create music from a text prompt is if it's been trained on thousands and thousands of hours of copyrighted songs, whether they're on a label or not. Artists still own creative and intellectual copyrights to their own work. I'm pretty sure I'm not a [__] lawyer, but I think that's how it works. The difference between a person consuming a bunch of Art and then creating something inspired by their favorite work and AI is that AI just

takes data and portrays it as its own. It's not necessarily interpreting art, it's not coming up with thoughts of its own, making artistic decisions inspired by things it likes. AI is inherently uncreative. You know, the way that I see it is this is a tool for extending creativity. I mean, is it though? People's reactions to AI at a glance kind of feels like when tablets for art, DAWs for music, and digital photography became popular, and old school guys thought their industries were being ruined because technology was advancing, but the real difference is that tools for those arts became way more accessible. Now everyone has access to a camera for photography, for example, or a music studio on your laptop. Advancements that democratized tools. But I guess it's all about perspective too because where's the line? This person didn't hand-place every pixel. They used a program called Blender to help them make it. Personally, I don't mind that. I really like this. But what if someone uses AI to draw a background and then makes original content on top of it? A lot of things are

going to become hybrid. What about using code to distort images to make art? There are also tools that can be used in shitty ways like Photoshop and FaceTune, and now video effects that can morph and reshape your body to make you skinnier, prettier, whatever. Ultimately, that line may be up to you in your perspective, and that's really hard to argue. So isn't AI just giving more access to artistic tools? And that's the thing. I don't think you can call this thing a tool for art. Is typing a prompt into a robot really creating art? This is just telling someone on Fiverr what you want, and they make it for you, and then you say you made it. No, you didn't. This takes every piece of creative agency out of making art, rather than making new opportunities to be creative. This removes the creative process entirely, whether we like it or not. This is just devaluing creative work more than it already is undervalued. Why is the goal to replace creative jobs? Pretty quickly, this is going to cause

people to be like, Why would I learn how to draw when the AI can do it better than me? Why would I learn how to do music when the AI can do it better than me? And faster. Real quick, since I'm trying to cover as much as I can, you may have seen that AI is bad for the environment. And if you're a confused and silly girly like me, you might be wondering how that's the case. Machine learning takes a [ __ ] ton of computing power. According to this PDF that looks credible enough that I found online, Chaty PT3 took about 1,300 megawatts to train. Those words mean nothing to me. There's nothing going on up here when I hear that. So, to put it into perspective, you'd have to stream Shape of You by Ed Sheeran 195 billion times to reach 1,300 megawatts. Or power 130 US homes for an entire year. Uh, so yeah. Listen, guys, I don't want to only

complain in this video. Even though I've spent a majority of this video complaints so far, I've covered how generative AI takes away from real artists, how they train on work that isn't theirs without permission, how they can be wildly inaccurate or inappropriate, and how it can be dangerous in society without proper care, which it's not being given the proper care, not, not at all, not, not at all. There's so much cool art out there by humans. Why do we need this? Art is like the one thing that is so unique to human expression. I don't see how this is useful advancing technology for the sake of advancing technology doesn't always mean that it's in a positive direction. Sure, I can generate an image of Mario and Sonic making out, but you know what's funnier is knowing that someone drew it out by hand. But that doesn't mean I'm against advancing technology. I've just been talking about generative AI this entire time. Many people, including me, think AI should be used as a tool to

make our lives easier, not to replace artists. Why can't AI do my taxes or fold the laundry that you're doing right now? Google actually uses capture data to train its AI to recognize fonts better, which helps digitize books. That's cool, a task that makes me wonder how ancient people really had nothing else to do than copy an entire book by hand. It helps AI recognize photos of traffic lights 'cause God knows you can't, which helps improve Google Maps, an app that also uses machine learning to predict traffic patterns and best routes. Non-generative AI can help interpret data for analysis and trends. It's helped analyze patterns so well that it's helped advance science. It improves Spotify and YouTube recommendations, improves the selection tool in Photoshop, rather than generating crap for me. AI and machine learning are already integrated everywhere that we just don't see or notice. Generative AI is just so in your face that it's easiest for companies to sell to consumers because it's so visible and visual. We just can't help but turn our focus here.

how about instead of generating music, unless it's an AI Plankton cover, we stick to tools like Adobe podcast that's helped save a ton of broken audio. You'll find AI that will enhance any speech to sound as if it were recorded in a studio. Toggle it on and off to hear how this recording originally sounded. Feel free to play around with recording your own voice or editing ours. AI can isolate and enhance vocals and audio from noisy old recordings like the Beatles did in their final record. Now and then, they tried to isolate John Lennon's demo vocals for this song in 1995 but didn't have the technology to do it well enough. I think of chat GPT as a starting point to learn about convoluted topics that can be difficult to understand like mortgage types, interest rates, loans, and it puts it in plain English so I can actually understand what to even start looking up from there. I was shown recently this notetaking AI that will join your meeting and at the end give you

insanely well-detailed list of notes that no one had to take. But again, this makes me wonder: where the audio of my meeting is going? Is it being saved? Are there privacy concerns? Do they sell my data? I mean, no one else is selling my data. Let me cosplay as a consultant or someone who's just thought about this for more than 5 seconds after chat. GPT spits out something. Why doesn't it just cite a few sources that are ranked most relevant or where it matched most of its information from? This can be a weighted score based on the credibility of the source and how many sources say the same thing. One answer from Reddit should not pop up first. Why doesn't chat GPT return a confidence score with its answer when asked a factual question? What if there were more refined AI models available that were just for things like scholarly articles and scientific facts? If the AI industry focused more on education and focused on getting the facts right, it could actually be helpful in some way. Because right now, I'd still trust

Wikipedia for my paper. More than chat. GPT, more tools like this are being developed where you can just upload a bunch of your documents. Research and the AI can help you dig through information in just your files, which I can actually see being useful for research in school. In a perfect world, AI-generated images and video would be correctly labeled, and AI didn't come out of nowhere. This is something that we've been moving towards longer than people realize, from our friend Clippy to Google's Doodle AI that we can now recontextualize given where we are today, to fantasizing about sexy AI ladies in film and media, to Alexa and Google Home. If they released today, they'd just be rebranded as AI products. If you've been here a while, first of all, thank you, but you might remember that I made a video on AI and a correction video some years ago, and my thoughts have drastically changed from then. As we learn more information about stuff, we change our minds and learn. And hello, I got to go. The video's too too.

Long okay. [Music] Bye. Binging, binging, binging. [Music]

Everything Wrong with AI

gabi belle

The Impact of AI on Privacy and Cybersecurity

Tech companies are rapidly implementing AI features, from Microsoft's controversial Co-pilot, which takes constant screenshots of users' activity, to the use of AI in design tools like Figma, raising concerns about privacy and security.

Microsoft's Co-pilot feature has faced backlash for its potential to expose sensitive information to hackers, despite being initially misleading about data encryption, while tech companies are being criticized for prioritizing speed over efficiency in implementing AI features.

Figma's AI-powered mockup tool was quickly taken down after generating designs too similar to Apple's, highlighting the challenges of rushing AI implementation.

The Impact of AI-generated Art and Images

AI-generated art caused controversy in a Pokémon illustration contest, leading to the removal of suspicious entries.

AI-generated images are becoming increasingly prevalent, leading to the development of AI art detectors.

The use of AI to create misleading images is raising concerns about its potential to rewrite history and deceive people.

AI technology is advancing to a point where it can create highly realistic and detailed simulations, leading to ethical questions about its use.

The use of human-like terms for AI errors is a deliberate strategy to make users feel more connected to the technology.

The increasing presence of AI in everyday platforms has raised concerns about the accuracy and reliability of its output, with examples of misleading and potentially harmful information being generated.

AI companies are struggling to find enough data for training their models, leading to desperate measures such as using YouTube videos without permission.

This has raised legal and ethical concerns about the use of publicly available data for commercial AI training.

An investigation revealed that multiple big companies have been using YouTube video transcripts to train their AI models, raising questions about data usage and copyright infringement.

The Use of Online Content for AI Training

Google and other companies are using YouTube videos and other online content to train their AI models, leading to lawsuits and controversy.

Companies like Open AI and Microsoft are being sued for copyright infringement, while others, like Reddit and Adobe, have signed licensing deals to allow their content to be used for AI training.

The practice raises questions about the ethical use of other people's work for AI training without their explicit permission.

AI companies are turning to synthetic data and AI-influenced music generation due to the legal consequences of stealing content.

The use of AI in creating music is becoming more widespread, making it difficult to distinguish between AI-generated and human vocals.

There are growing concerns that Spotify may be using AI to generate music for its curated playlists, leading to speculation and theories on the internet.

AI is being used to distort images and create art, raising questions about whether it enhances or diminishes creativity.

The massive computing power required for AI, particularly generative AI, raises concerns about its environmental impact.

Despite the controversy, many believe that AI should be used as a tool to simplify tasks and improve various technologies, from digitizing books to enhancing traffic predictions and data analysis.

The Impact of AI Tools on Audio, Education, and Research

AI tools like Adobe podcast can enhance audio quality and are also used for noise reduction in old recordings.

Chat GPT serves as a helpful tool for explaining complex topics in plain language and could be improved with source citation and confidence scoring.

AI is also being developed to assist with document research, but there are concerns about data privacy.

The evolution of AI technology has the potential to positively impact education and research, but there is a need for more refined AI models for scholarly facts and articles.