Hello! We're bringing you the latest episode of Tech Girls Talk.
Today we discussed OpenAI's newly announced image generation AI "GPT-4o"!

An Era Where We Can't Tell What's Real...

Have you seen those historical photos that move with AI on Facebook lately?
They're super fun to watch!
But at the same time, things like Taylor Swift's wedding photos turning out to be AI make it a bit scary 😱

GPT-4o's Image Generation Feature is Incredible!

This new feature is truly amazing!
It can handle text and images simultaneously. Here are some key features:

• Perfect representation of text on signs and menus ✨
• Ability to modify specific parts when asked in conversation
• Good understanding of Japanese kanji and hiragana
• Support for various styles like hand-drawn or realistic
• Excellent representation of complex relationships

I actually made a 4-panel manga, and it perfectly incorporated Japanese text—I was impressed!

Image Generation AI is Evolving Too Fast!

Looking back at its history:

• 2004: First human face generation (looked a bit scary)
• 2018: NVIDIA generated faces so realistic they had to say "this person doesn't exist"
• 2021: We played a lot with Midjourney! (Strange things like having 10 fingers)
• 2025: With GPT-4o, super-realistic images following exact instructions

Technology evolution is really fast! Who would have thought it would change so much in just a few years since 2018 😲

As a Designer, It's Super Helpful!

From a designer's perspective, this is incredibly convenient!

1. I can now present more options to clients in the same amount of time
2. When clients say they want something different, I can quickly generate new options
3. Meetings go more smoothly! The range of ideas expands dramatically 💡

But There Are Scary Aspects Too...

Despite the convenience, there are some concerning aspects:

• What will happen to the jobs of photographers and artists?
• Fake news and fake photos will likely increase
• Your face or family photos might be used without permission...

Especially fake news is becoming harder to distinguish, which is really scary.
Apparently AI-generated images have tags, but I worry those could be easily removed.

You Can Use It in Daily Life Too!

Recently, I was wondering how to remodel my hammock space in my room,
so I consulted GPT-4o and it gave me lots of wonderful ideas!
It's amazing that it can be used not just for professional work but also for everyday concerns ✨

Summary

AI evolution, especially in image generation, continues to show incredible breakthroughs!
Rather than taking our jobs, it seems to be becoming a tool that expands our creative possibilities.
However, there are issues like fake news, so we need to use it wisely.

Next episode, we'll be talking about "Vibe Coding." Stay tuned! 😊
Tech Girls Talk

Reference Links

https://openai.com/index/introducing-4o-image-generation/ https://techcrunch.com/2025/03/25/chatgpts-image-generation-feature-gets-an-upgrade/ https://venturebeat.com/ai/insane-openai-introduces-gpt-4o-native-image-generation-and-its-already-wowing-users/

Episode Links

• YouTube: Link
• Spotify: Link
• Apple Podcasts: Link

YouTube Transcript

0:01: Welcome to Tech Girls Talk. On this radio show, women engineers and designers working in tech companies will talk about the latest trends.

0:08: I will be explaining about technology and AI news. I am Mari, a designer.

0:14: I am Sawako, an engineer.

0:19: Yes, so this week is here again. Yes, this is episode 9.

0:27: Yes, thank you for joining us. Yes, thank you for joining us. A little while ago, someone pointed out that the opening was a little low-energy, so I would like to start off with a more upbeat voice.

0:39: Yes, today, it seems like I'm a little bit upbeat, so I'd like to ask for your enthusiasm.

0:46: It certainly starts off lazily and gradually picks up, doesn't it? Yes, today, the topic has changed a little bit.

1:09: Recently, I've been watching a little bit of Instagram and Instagram threads, and YouTube, and I've been watching TikTok. I only occasionally look at it, but recently I've been meeting people I've been getting older with

1:20: I've had a lot of opportunities to do this, and I've noticed that older people tend to use Facebook a lot

1:25: Oh, no, I don't have Instagram, but Facebook

1:31: There are so many people who have it, so I thought, "I see," but I don't really like using Facebook, I guess because there's so much unnecessary information coming in

1:38: I don't really like it, I can't narrow it down, I guess I should set it more precisely

1:45: I guess, but in the end, it's been around for a long time, really 2013, 2010, 2013, I guess

1:50: I think it started around that time, and I've got a lot of history, so I don't think so

1:56: But I guess a lot of old stuff from that time comes up

2:06: I don't know if it's possible, but it's attached to my emails, so I recently installed Facebook again and it's been sending me news and feeds.

2:16: So recently I've been seeing a lot of stuff like that, and it's the kind of thing I really like, there are historical photos, right? Like, from the 1930s.

2:27: Yes, yes.

2:37: There are old photos of Tokyo in Japan, or old photos of this area in Rome, right? Recently they've been turning them into videos with AI.

2:44: It makes the Kamae look like a real photo, and it's a video generated by AI.

2:54: It's like, how do you say it in the image, it's like, you can compare it.

2:59: Yes, I really love looking at it, and for example, Natsumeishi's face is really moving, and his mouth.

3:08: It moves and looks like he's talking. I really get the feeling that if it really happened, it would have been like this. I think it's because I've liked those historical things from the past a few times, but I really enjoy watching them and I watch them, but sometimes I see that the tailor and her boyfriend got married.

3:37: There's a photo and I thought, "Are you serious?" and I thought it was pretty quick, but I don't really see anything like that.

3:43: I don't like it, so it doesn't come up that much, but I happened to be into the people who do it, and I thought, "They're getting married."

3:49: I thought it was just a happy occasion, but when I looked at the comments, I thought, "Oh, this is probably an AI."

3:54: I thought that, and I thought, "Oh, that's right," and I thought, "The AI's image is getting better."

4:01: The accuracy is getting better and I don't understand it anymore. That's right.

4:20: That's right, so recently I was really surprised by both, and I thought this was news, but it wasn't news, and it was a bit of a negative surprise, and I was thinking that AI is scary, and then this topic came up, so I thought it was perfect and I wanted to talk about it.

4:27: It's exactly the right topic for today, right.

4:32: No, really, recently, there's been a lot of stuff like that, like it was made with AI, and I can't tell which is fake and which is real.

4:44: I don't know, right? I think it's scary in a way, but it's scary.

4:50: Today's talk is about that image generation function, right? OpenAI

4:55: It's a new image generation function announced by OpenAI. Yes. Well, this function is very amazing because it can generate images more accurately according to realistic instructions with a multimodal model that can handle text and images at the same time, and it has become a hot topic everywhere now.

5:15: Yes, I remember that, that chat

5:21: GPT chat GPT, what was it? Germid Journey or when Mid Journey was released

5:26: We played a lot, didn't we? We played a lot, it was really fun, yeah

5:32: Mid Journey was amazing, really, at that time, yeah, I thought there was no company that could beat Mid Journey

5:41: Yes, I thought so, yeah, yeah, but this time OpenAI announced something amazing

5:49: Yes, yeah, I tried using it, it's really amazing, this is a new thing that OpenAI announced

5:55: That function of GPT4, image generation Yes, it is a function, but the main features are

6:02: Accurate text description and detailed text expression such as sign menus and invitations are possible, and

6:09: Contextual Sun This image can be updated and improved while maintaining the flow of the chat, which is amazing

6:15: Yes, it's not a mid-journey, right? It's not a conversation like "I want to make this, I want this to be done this way" from the flow of the chat, so mid-journey is other

6:24: Yes, it was a bit like that, but it's quite off, isn't it, with journeys?

6:30: Yes, yes, you have to set it up quite in detail and give instructions and change it to follow this, like mid-journey

6:40: Yes, like other things, you can change only this part in the context, and

6:47: And it's amazing that you can do it in natural language, already in Japanese.

6:53: It also supports visual communication logos, diagrams, infographics, and other image generation styles to convey information.

7:02: Handwritten style, archery style, infographics, this is another generation

7:07: Like image AI, it supports various styles and has high accuracy.

7:14: It can depict up to 20 objects and complex relationships, and also utilizes user images.

7:20: It can also generate and expand based on uploaded images, which is its main feature.

7:29: Yes, the limitation is that the image cannot be trimmed, so it cannot be used for posters that cannot be edited.

7:34: However, it can be cut off, there are facts that are not corrected, and it is difficult to edit in other languages.

7:40: It is not possible to correct small details due to lack of accuracy, and it is weak in Japanese.

7:47: The accuracy of the words is still lower than that of English. It's not that expensive.

7:53: But it was still amazing. I tried it yesterday and I was able to use both hiragana and kanji very well.

8:05: Oh, I see. I've never tried it in Japanese.

8:17: Oh, Japanese is amazing. I was really impressed when I was able to say hello and konnichiwa in Japanese in mid-journey. I understand.

8:35: I was really impressed, but after that, especially with kanji, I couldn't distinguish between the Chinese and Japanese styles at all. It was difficult to distinguish between them until now, but yesterday I made a four-panel comic with it. It was amazing.

8:45: Well, it's really simple words, but of course it's not perfect, so I'll edit it later. If you add Photoshop, it's quick and you can do it almost all.

8:50: Hmm. Yeah, I thought the Japanese hiragana was really amazing.

8:58: And when you add instructions, the font is beautiful, and the font is the same, and you can really chat directly in the chat, so I was surprised.

9:12: Yeah yeah yeah yeah

9:19: Well, in terms of safety and transparency, all generated images are embedded with C2PA metadata to show that they were generated by GPT4O.

9:24: And inappropriate image requests are blocked.

9:30: It is operated with safety in mind, and the current way to use it is chat with GPT plus.

9:38: It is being gradually deployed in the pro team free plan, and will soon be available for enterprise and education.

9:44: So that's what's planned, right? API use is scheduled to start within a few weeks, so this is

9:51: Well, this is information from the official OpenAI site, but it was really amazing, the images are already

10:01: Realistic, or rather realistic, is that up until now, images and other image generation have been pretty realistic.

10:08: It was, but it's even more complete, and they say it responds to our instructions very well.

10:15: That's what really surprised me.

10:21: The technology is improving, isn't it? It was a great presentation.

10:34: Okay, so then, I'll just briefly introduce how image generation AI has developed so far.

10:42: Generative AI, that's the general term for this, GAN.

10:49: This GAN.

10:57: In Japanese, it's called a generative adversarial network, which is the name of a type of AI algorithm.

11:03: Then, using that, they created a human-like image

11:14: Ian Goodfellow was the first to be able to generate a human-like image

11:20: This person and the people who worked with him first tried it in 2004

11:28: They tried it and succeeded, but it looked a little scary, like a thriller

11:37: The face had just been created and it hadn't been properly defined or it hadn't developed that far yet, so they succeeded in generating a human face that looked a little human, but was a little scary, and then

11:56: That's right, that was 2004. And then, I hadn't thought about that at all at that time, so I'm really surprised that they were already doing it at that time

12:07: So, in 2018 It's been a bit more recent in 2004, and they're the company that uses that graphics chip.

12:17: It's the first time they've used that style of GAN to create a face that looks like a real human.

12:24: It's so good that they have to tag it as if it's a person who doesn't really exist.

12:31: They succeeded in creating a face that looks like a photo with good quality.

12:37: For the people who made it, they were like gods, they were like a real human.

12:46: It was 2018, and it had an impact like they'd made a face that looked like a real human.

12:57: It's been a while since 2004, but they still managed to make it and things have been going pretty smoothly since then. 2021 is still good, isn't it? Mid Journey has come out, and that's the one we used to play a lot.

13:04: It's really a lot, you can generate anything, but the hardest part is the letters, as Mari said earlier.

13:18: Of course it's kanji, but our company logo 1F81F8 this or something

13:26: It's like, on Instagram, our company, in that hidden picture, like, where's Wally hiding?

13:32: I thought it would be fun to post that, right?

13:39: More than that

13:47: They put it in, but it's really big, so it doesn't feel like Wally, and also, like, F, 1, or 8

14:00: It's like, I don't know what it is, it's like a letter, or it's like a character, and in the end, when you're really in a pinch, There was something that was hidden in Photoshop, but I didn't do that.

14:08: Yes, this new open AI O4 in 2025 is the image generator.

14:15: The image generator can generate Japanese characters, as you said earlier.

14:25: Partial correction is possible. I really wanted that partial correction in 2021, and that's the trend now.

14:36: Even now, it's really amazing when you look at it like this. I was surprised that it started in 2004.

14:43: Yes, this AI is really, it's really 2018.

14:50: In my opinion, it really started within 10 years from 2018, around the time of Corona.

14:56: Ah, me too. Yes, it's really, really, really 23. It seems like a lot of things have changed in a single year.

15:06: Yes, it was quick. It was quick. Yes, new technology is AI.

15:12: It feels like new technology related to AI is coming out 23 times a month.

15:18: Yes, it's true. Ads and even photos and backgrounds are all made by AI.

15:26: Illustrator and vector logos are all made by AI. Yes, I think Canva is like that too. Even the free Canva.

15:31: Ah, I see. Canva is like that. Yes, you can only do it a few times, but it's free.

15:36: And you can still use it. I thought it was amazing. Yes, it's amazing.

15:42: Yes, what do you think about this as a designer? Well, I don't know if I should say that my work has become easier, but what can I say? It's become easier to present various ideas to clients.

15:55: Yes, yes, yes, yes, yes, yes, yes, it's really appreciated, isn't it?

16:05: Yes, up until now, when I was asked to make something, I would try to search through my drawers. Yes, yes.

16:13: But even if I only had one or two ideas that I could immediately imagine, I couldn't really present just one or two, after all.

16:19: I might present about five, so in those cases, if I couldn't come up with a good idea myself, I would go and look for it myself, searching on the internet.

16:30: So I would go and look and look and look for it, and when I found something that I thought was good, I would edit it and present it to the client, asking them what they thought, but up until now, it took quite a lot of time to search.

16:42: Yes, this chat GPT right now

16:48: If you ask, they'll give it to you soon, and if it's different, you can just say it's different.

17:02: If you ask them, like, "This is better, not like this," or "I want something different," they'll give you more and more, so up until now I could only give about 10 in the same amount of time, but now I can give about 10 at once, so I think it's really great and it's fun, too.

17:13: There's a lot to learn, too, and it's like, "Oh, I can do this," [laughs]

17:19: Well, the options are expanding, aren't they? Amazing, it's amazing, it's really fun, I was looking at it all day yesterday and I thought it was fun.

17:27: [laughs] I'm really excited about this.

17:36: It's going to be like Mid Journey Part 2, right? That's right, that's right. I don't think I'm going to get too addicted to it.

17:43: Yes, yes, that's right. On the other hand, what about developers? Website developers, etc.

17:50: [Music]

18:03: Yes, I think it will make things easier for developers. For example, in our company, we are currently developing a software called IQ, which uses AI.

18:08: Websites can be generated in an instant.

18:16: Yes, yes, we will add various functions and many functions using AI.

18:21: We are currently developing a software that will provide all the things that users need to do their work, to run their business, and to provide everything they need.

18:28: But among all of these, images are the things that have been used up until now.

18:40: Free software like Unsplash. I get it from places like that

18:45: That's why I say placeholders are just the size

18:51: That 200 pickle times x 100

18:58: And I've been taking all sorts of, um, what do you call it, size URLs and making them into placeholders and stuff like that, but then I don't know what the atmosphere of the site is, for example, I'm not saying

19:16: When you put an image in a place, if you don't have the right color and shape, it's hard to understand the atmosphere

19:24: Yes, so sometimes when I have to do a project like that without going through a designer, I go to Unsplash and I'm like, I want to work with a nutritionist, so if I say vegetables, vegetables, then everything from the farmers market comes out, right?

19:42: So that's what this generation AI is, image generation AI Thanks to this, I can't do much with the color scheme I want

19:50: If you have a sample image of this color, you can put in a warm color or a full touch, and you can give detailed instructions like "make it like this and with this color," and then you can make corrections little by little, which is amazing.

20:10: I thought it's made a lot of progress and it's going to be really convenient.

20:15: Yes, yes, yes, yes, it's a pretty time-consuming task, and it took a lot of time to find images.

20:26: No, it's really complicated, and even though I thought it was free, it's not free.

20:38: And there's the copyright of the person who made it, so I have to bring in the link and make sure to include their name, and there were a lot of things I had to pay attention to, but if it's GPT or something, I'll chat with GPTAI.

20:45: Just write "generated" and you'll be fine. Yeah.

20:52: That's probably okay, and I don't know what will happen in the future, but some people think that the training data for the chat GPT might be collecting copyrighted images and training them, so that could be a problem, and of course there is a possibility that that could happen, so for now, just tag it as "generated by that generator."

21:18: I don't think there's any problem at all, so I think it's very convenient. Yeah.

21:35: Right, so right now we've been pretty positive about image generation being a support for us, but on the other hand, I think there are many industries where this AI generation will have a negative impact, or affect work.

21:45: For example, photographers, artists, engineers, etc.

21:53: That kind of thing is, AI is what you call, people's photos, photographers, etc.

22:01: It's said that they bring in things from artists and make them, so I think so

22:08: What do you think the impact of this will be in the future? Well, it's true, that

22:16: I'm sorry to say that this kind of thing is not very technical, but there is a level of skill, isn't there?

22:51: Yes, even photographers who have just started, but even if they are talented, they are still just starting out, or people who have a creative way of taking pictures, or people who are really unique and have risen to the top, I think it's okay, but I think AI

22:57: The people who are developing it are I talked about this a few episodes ago, but I think it's the same for people who don't have much skill or experience.

23:11: For example, when I went to Hawaii the other day, do you know Sen? He painted pictures of the ocean or something.

23:21: Ah, yes, yes, yes, yes, it's so beautiful.

23:26: I think it's an oil painting.

23:36: I think it was, but there was a gallery there, so I just passed by it. I went in through the entrance and left through the exit, but each and every one of them, how should I put it? The paintings are amazing.

23:42: This, how can I put it? It can't be expressed digitally. It's so beautiful. It's like an oil painting that stands out, and there are all kinds of things, I wonder, the light, and it feels like it was made by hand.

23:59: It's amazing and I want one of those for my home

24:05: I can understand that feeling. So I think it's better to make something digital like this on a website.

24:11: The people who do things like "please" and advertising are surprisingly done by AI, so I think it's better to have more of your own advertisements on there.

24:17: The number of people who do this might be decreasing, but I think it would be better to have more of your own advertisements on there.

24:24: I want to make it unique, so it's not a generative AI, it's not digital, but something a little unique.

24:50: For example, I might try making something into art for an exhibition, or maybe I'll be able to do that someday, but I might want to make something that feels bumpy and you can touch it with your hands, and then someone could generate it live and make it for me, or something like that. I don't know because I have a printer, but if I take this and that into consideration, maybe I can do it, but even if I want to put out Unix, I think I can put in something like a key point, or something like that.

25:03: If I put in some effort, I think that even if you don't have that much skill, you can still put it in if you want, but more and more of that skill difference

25:18: Yes, I think that the number of jobs will decrease a little because of that.

25:24: I see, well, well, well, well, it's going to be scrutinized, right? Yes, that's right, so it's like only what people think is good will remain.

25:36: Well, that's a good thing, but on the other hand, if you start talking about MDs and set tapes, well, there are still people who use cassette tapes, right?

25:44: Yes, so, how should I put it, I think there will definitely be a market for that day, so I don't think everyone will lose their jobs, but I think it will decrease a little.

25:57: Yes, I'm predicting that. I might be wrong, but yes, there will be copyright issues in the future.

26:04: Yes, that's right, especially.

26:11: Yes, I think it will be a focus of attention, but in terms of copyright, this fake image of AI, fake news, it's amazing now, isn't it?

26:19: [Music]

26:26: Well, as I said earlier, the photo of Taylor Swift and Travis getting married was an old photo. If it was a mid-journey one around 2021 after I took over, the fingers were like 10

26:37: There were about books, right? I remember that there were definitely pandas and paws and things like that, about 10 grains. It was a bit traumatic for me, but it was really scary.

26:45: Yes, it was, but there was nothing like that, it was probably Photoshop. Yes, yes, yes, I didn't look that far, but I think if this kind of thing keeps coming out, it's scary to think that it might be a crime to use your face.

27:11: Yes, if it's recorded, I think it's scary. Yes, yes, yes, this is a real AI.

27:16: It's been said since it came out, but that's right.

27:26: Yes, you can see what he's making, right? People are putting it on various Instagrams and microphones, so it's not easy to imitate the so-called leaders of each country, but you can make various propaganda using the leaders of each country, and imitate them. It's good if you do it in an interesting way at first, but for example.

27:46: I think it would be fine if it was something cute like drinking on the beach or talking to Winnie the Pooh.

27:53: Yes, yes.

27:58: If it's a book, it's obvious that it's an AI, but if it's a picture of someone shaking hands, etc.

28:04: If it's a picture of an important person fighting or talking, that would be a big problem.

28:16: Yes, yes, especially now, there are teachers and all sorts of things.

28:22: Yes, there are religious and other severe parts, so I think there needs to be solid guidelines. Even if there were, I think there would be a lot of people who would break them.

28:34: Yes, yes, it's not so much a problem with the company that made it, but rather a problem with each person's morals.

28:40: Hmm Why is fake news so difficult to spread? I wonder if it will continue to spread in the future.

28:46: [Music] Yes, I think it's already increasing.

28:51: Yes.

28:58: What was it, I haven't been caught yet, but I thought for a moment that there was something like WS, that werewolf thing, right? This one too.

29:03: Facebook, you know, Facebook, it's like this, so I don't look at it that much because there's a lot of stuff that I don't like.

29:09: Yes, I happened to get a message or something, so when I glanced at it, there was a photo of some kind of werewolf being displayed in a museum in white.

29:17: Yes, I thought this was definitely fake, but it looks old, you know.

29:24: Yes, but apparently there's a museum that makes fake things like that.

29:29: And then you link that to the news. I did something like this, this, this, this, and then someone commented at the top of the comments like this is fake, don't believe it, so I think it's like that

29:45: I guess that's what they say, well, there are white hackers, but they hack, but they hack properly for people, they hack to prevent being hacked, and I think that kind of organization is necessary

30:06: Yes, yes, yes, yes, that's right, I think it's like, it's not like white anemesis, but it's like an anemesis group

30:13: It's like, it's like, let's all edit it together like an open source, or something

30:18: If we don't have something like, let's orad it, let's search it, it's just going to

30:23: It's just going to be more and more photos, and the tags are like, cha open

30:30: AI is adding that tag, yes yes, this is an AI, open AI

30:35: I have added the tag "This is a photo from" when I started programming once, and I was wondering what it would be called? What would it be like to manipulate a photo?

30:48: I had a task to create a program that could manipulate a photo, and I

31:05: When I did that, I could change the color of the photo, and it was like a string of letters, so I could rotate it, change the color, and it can be done normally with coding.

31:10: And people who can do that kind of thing can do that photo, chat, GPT, open AI

31:15: I thought I needed to look into whether I could remove the AI ​​tag from something I made, and I see

31:24: It's just something that can be changed, right? JPEG, PNG, etc. But it's all arranged in a certain way, and if you just change a little bit of that arrangement, the color or shape of that place will change, but can you even change the shape?

31:40: Yes, you can, so you can manipulate it.

31:53: So that's why you can do that, so no matter how many tags you add, it's probably not good if the tag is removed and saved somewhere. I was thinking about that a lot, and I was wondering how effective the tag is.

31:58: I see. Well, I don't know, but I don't know if you're going into too much detail about the image spermatogenesis or the image of the image.

32:06: Yes, yes, I've never done that kind of detailed operation, so I only did it a little at the beginning, so I don't know, but I'm wondering if you've done that or if you can do that.

32:18: I can imagine that, right? The image created by the AI ​​generation says that it was made by AI, but does that mean that other people can take a picture of it?

32:27: Yes, I think it's possible to take it, but I wonder if it can be saved somewhere by taking it, so the download destination, the person who downloaded it.

32:41: Yes, that's right, yes, if you connect it, it might feel like that, but I think that's it.

32:50: But I don't know, I'll have to look into it a little, I don't know. Yes, yes, there are a lot of fakes, and I think fake news and photos are circulating a lot, so I guess it's a bit of ... So

33:09: Broadcasting real news

33:15: I think there will be more demand for channels or whatever

33:22: But, yeah, I see, everyone wants a real voice, yeah, yeah, a real person doing it

33:28: But, you know, if you think it's a person doing it, it could actually be an AI, right? That's right, that's Google's

33:36: I wonder if they can make podcasts now, maybe around next year when we'll all be gone

33:43: Yes, maybe everyone is listening to our voices now, but we could be AIs too, yeah, yeah, that's right

33:53: But maybe there isn't anyone with such a raspy voice, I've had hay fever recently and my voice is a little raspy, so if it was an AI, maybe it would be a little cleaner. Yes, I chew a lot too, so an AI that chews a lot

33:59: It seems like it wouldn't be possible, so if he stumbles, is he the real deal? Yes, certainly, certainly, AI

34:04: He doesn't stumble, does he? If I ask him to stumble, will he stumble? I'd like to hear it. Like, please make a podcast about God. Certainly, that would be good.

34:15: Yes, I'd like to try it. I'll try that. I'll try that.

34:21: And the Osaka dialect, yes AI

34:29: If it can be done, I'll definitely audit it. I'll go and check it out. Osaka dialect is all about intonation, so yes, yes, you can tell when someone who isn't from Osaka speaks even a little bit.

34:36: That's what they say, right? Yes, so if AI reaches the Osaka area, there's a chance that we won't be around anymore.

34:45: Yes, that's right, if they can speak Osaka dialect, they won't need us anymore if they're gods. It's the best, I don't need it, I'm just a normal comedian, so that's it

34:52: [Laughs] Yes, well, yes, I think so, like now

34:59: There are various problems, but right now, we are talking from the perspective of the user

35:05: But, not like the photographer or artist I mentioned earlier, but for example,

35:11: But, there is a possibility that photos or videos of people who look like you could be used in other companies' advertisements

35:16: There is a possibility that photos or videos of people who look like you could be used in other companies' advertisements

35:23: In the future, AI

35:29: If everyone is using AI, then, well, I think it will be taken in a way that will imitate them, for example

35:39: If you use an iPhone, it may come from the iPhone's cloud, and you may upload photos to Google

35:45: That might happen if you put it on So it's difficult to say where it's from

35:51: I think so, but you can just post a photo on Facebook and the AI ​​will pull it from there

35:56: Maybe, but I can't specify where it's from, but in the future, I'm going to use this to

36:05: When I think about the possibility that faces that look like me or my family's faces in my daily life could be used for advertising without my permission, I'm a little hesitant to post photos on anything

36:21: Yes, I'm quite careful about that when I use various services, but

36:27: There are basic privacy policies that say not to use it for that kind of thing

36:34: Yes, yes, yes, yes, so I would definitely say yes, don't use it, or use it

36:40: If it's okay, you can turn it on or off, but I always make it the do not use it side and do that

36:47: I try to use services like Google and Apple, but GoogleApple is a bit

36:53: I don't really understand the cloud, but with Google and Instagram, you can always check your privacy,

36:59: You can always check the security system, and then you can check your photos,

37:05: There's a line that asks if you can use that kind of thing for generation, or not, so I always check that myself, and I

37:13: I always tell people not to use it, but even so, I'm still overseas

37:20: I lived in a foreign country, so it's really strange, but people overseas have a strong awareness of that

37:26: You have to protect your own rights

37:47: If you don't protect your own data, your face photos, and your things yourself, no one will protect them, so they are doing a great job of raising awareness of such rights and starting movements to protect them themselves, so I think that as more and more AI technology develops and is used in Japan, it will be important to protect your own photos and copyrighted materials and other such things yourself.

38:00: I asked Chat GPT about the countermeasures, and I think that they are really good at using it in reverse, and they are collecting a lot of information.

38:10: It is now possible to collect such information, and it is free, so I think that we can use it to protect ourselves.

38:21: Yes, that's interesting. So you are turning to AI to protect yourself from AI, right?

38:28: I think there are situations where you will lose out if you don't use it.

38:34: Yes, yes, yes. Certainly

38:44: Well, it's difficult, but even I don't know because I'm from Osaka, but if my face was used in, for example, an iPhone commercial, yeah, yeah

38:51: It's a house that was used in some nasty, nasty thing, but if someone's face like mine was used in an iPhone commercial and looked really cool, I'd probably say, "I'm in it!" I'd probably use it as a joke and use it as a joke for the rest of my life

39:04: That's true, if it was a major company and it looked cool, that's true, I'd say, "I'm in it!"

39:09: I think it's a way of saying it, even if you say it was, it's like, "It was a poop tip!"

39:18: Yes, that's true, that's true, it's true, it's the joke that matters

39:34: Yes, no, it's interesting, so this time, I've suddenly changed the topic a little, and since this was announced, I'm going to chat about GPT's Open AI's Image AI I talked about it

39:41: Yes, yes, I want everyone to try out various things as well

39:50: Just one last thing, that's right, I've tried out various things. I have a 3-room house, but it's not that big, it's about 3x2 or so, about 2m

39:57: I made it a little big enough to hang out some clothes, but there's a hammock there, and it was a gift

40:06: It's pretty heavy because it's self-supporting, so it's not really easy to do, but I want to make it a little cute, like a place to drink coffee in the morning

40:19: I didn't have any ideas, so I asked the AI, um, that chat GBT earlier, um, um, um

40:25: I asked the teacher, and he did it, and it came out pretty nicely

40:32: Ah, that's good, yeah, yeah, I thought it was great that you could use it like this.

40:37: It's good, I want to change the interior a little, but I don't know how to change it.

40:43: Yes, that's right, I'm thinking that if I put my own photo in it, it will look better, but I'll try it a little later, but I think it's good for the company's one, but I won't know until I try it.

40:56: I think I'll try it a little, yeah, yeah, I think it will expand my dreams in terms of things like remodeling.

41:05: Yes, not just in your main job, but in your private life.

41:12: If you use it more and more, please, we're not being sold by that open AI, but it's really interesting.

41:17: Yes, let me sell it, let me sell it, sponsor me, I'll talk more.

41:26: So this time I talked about OpenAI's image generation AI. Yes, next week is the vibe coding that I was supposed to talk about today. Yes, I'm already in the middle of vibe coding again. I'll let you know my thoughts after I finish it. Next week, yes.

41:40: [Laughs]

41:45: Yes, I'll definitely let you know. Yes, I'll try a lot today and this week.

41:52: Yes, yes, yes, so thank you very much for this week.

41:58: Thank you very much. I hope to see you next week. This was Tech Girl Talk Radio. Thank you very much. [Music]

42:06: Thank you very much.