In our first episode of "GIRLS TECH TALK," our duo of Engineer Sawako and Designer Mari discuss the latest topics in technology and AI. This time, they focus on the highly debated topic: "Will ChatGPT take away engineers' jobs?" As this is our first episode, we try to go over some topics that you should know from the Tech world without assuming you know too much.

1. What is ChatGPT?

• ChatGPT: A conversational AI developed by OpenAI that uses LLM (Large Language Model) to generate natural human-like text.
• Along with Google's Gemini and other brand names, these are collectively referred to as "LLMs."
• Popular among engineers as an "AI that writes code."

2. Trends in Major Global Companies

• Meta (formerly Facebook): CEO Mark Zuckerberg declared plans to automate mid-level engineering tasks with AI by 2025.
• Alibaba: Developed an "AI programmer" that can generate apps in minutes, sparking debates in China's developer community.
• Salesforce and others are also reducing engineer hiring due to AI implementation.

3. Sawako's Perspective: Is Relying Solely on AI Dangerous?

Engineer Sawako shares her thoughts after regularly using ChatGPT and other LLMs:

1. Insufficient Error Checking

   • Copy-pasted code often doesn't work properly.
   • Library versions and configurations can be outdated.

2. Difficulty in Adapting to User Requirements

   • Real projects require understanding clients' detailed specification changes and design requests.
   • AI alone isn't sufficient; human experience is ultimately needed.

3. Entry-Level Tasks Are More Likely to Be Replaced by AI

   • Simple front-end tasks like HTML/CSS could be adequately supplemented by AI.

4. Conclusion: The Engineer Profile Needed in the AI Era

• Programmers won't become "completely unnecessary"; human experience and verification skills remain crucial.
• While "non-skilled" layer work might decrease, creative work remains firmly in the human domain.

Please listen and subscribe to the full episode!

📱 Available Platforms:

• Spotifyで聴く
• Apple Podcastで聴く
• YouTubeで観る

YouTube Transcript (Translated from Japanese)

0:03: Welcome to Tech Girls Talk

0:06: On the radio, women working in tech companies

0:09: Engineers and designers are trending

0:11: Technology and AI news

0:13: I will explain. I am Mari, a designer.

0:17: I am Sawako, who is involved in development.

0:21: Yes, so it started. Yes, the first time.

0:27: First time.

0:30: Nervous.

0:32: Yes, so I hope you will be happy this year.

0:36: Thank you very much. Thank you very much.

0:38: Congratulations. Congratulations.

0:40: Yes, it's been a month. Yes, it's already the end of the month.

0:43: February is almost here. Yes, I'm already happy.

0:46: Valentine's Day is coming soon. It's so fast.

0:50: Right, we're all working remotely.

0:54: I only go once a year because I'm working,

0:57: I don't actually face it, but this month last year

1:01: There was a year-end party, yes, all the way to Tokyo

1:04: I went, um, it was my first time

1:07: We gathered in Chiba, Yokohama, and Yamahayama

1:12: We looked at the sea and watched Seto

1:16: It was great, it was great, it was delicious

1:18: The various foods were also delicious

1:20: It was delicious, it was delicious, it was delicious

1:23: We worked there while relaxing

1:27: Yes, yes, by the way, on an old Facebook

1:31: Well-known meta is replacing great engineers with AI

1:34: I heard that, right?

1:37: I saw that in an article, I saw that too

1:41: It's better

1:42: So today, I'm going to talk about chat.

1:46: Will GPT take over developers' positions?

1:49: I want to talk about that.

1:51: It's very deep, the first episode.

1:54: It's certainly deep, it's deep.

1:57: I wonder if I can do it.

2:00: It's a bit strong, but no, no.

2:04: It's a hot topic. Before that, how about it?

2:07: Chat GPT Chat GPT is

2:10: I want to ask you a bit about it.

2:12: Thank you, but what will you chat with?

2:16: I think there are a lot of people who already know about jpt.

2:19: I think there are a lot of people who know about jpt.

2:21: It generates natural sentences like questions written by humans.

2:23: You can answer questions and have conversations.

2:26: It's called AI.

2:28: Artificial

2:29: Intelligence sounds a bit off.

2:31: It's a word that feels like it's artificial intelligence.

2:35: It refers to that. It's a bit in GPT.

2:39: There's ChatGPT and jem.

2:42: Jem is a Google chat.

2:43: GPT was made by OpenAI.

2:46: Is that an AI function?

2:49: I asked them to think of all the brand names.

2:51: I think it would be good if they had them, but what's on top of that?

2:54: There's a large run MOD called Garumu.

2:57: There's a large hope language model.

3:00: It's collectively called by names like that.

3:03: Below is the chat GPT and jiga

3:06: It's like that.

3:08: I'll add a little more.

3:11: It's a lot of text data, and it's the llm

3:15: That means it's learning and learning

3:18: Humans feed it and then use it to make natural sentences

3:21: It's the main way that mallm works.

3:30: I see, thank you very much.

3:34: That's good, so this time, the chat GPT is

3:37: So, are they going to take over the developer's seat?

3:40: I'd like to introduce two of my articles

3:44:

3:46: The first article is about South China

3:49: It was an article in the Morning Post

3:53: First China Morning Post

3:55: and Alibaba Group have developed an AI programmer

3:58: I saw that programmer

4:01: If you use it, you can create applications in minutes

4:04: and there is a debate going on among developers in mainland China

4:08: The next article is an article by The Byte

4:15: It's an old Facebook meta

4:18: CEO Mark Zuckerberg announced plans to introduce AI for mid-level engineers by 2025

4:24: The company has announced plans to introduce AI for mid-level engineers by 2021.

4:27: This AI will have behavioral capabilities and will be able to perform many of the tasks currently performed by human engineers.

4:31: Zuckerberg said that this AI technology will be

4:42: expensive at first, but as efficiency improves,

4:45: most of Meta's apps and AI models will be built by AI engineers.

4:48: However, he avoided a direct answer to the question of whether AI will take over human jobs, claiming that it will instead amplify the imagination of workers.

5:02: This movement is not only seen in Meta but also in other large companies.

5:08: It can also be seen in companies like Salesforce and

5:10: Companies like Klarna have stopped hiring engineers and cut staff due to the introduction of AI

5:14: As automation by AI advances

5:17: It will have a major impact on Silicon Valley as a whole

5:21: It is disguised as such, yes, yes, yes

5:26: So the two articles were about how AI will change the

5:30: but there are also articles with opposing views

5:33: That's right, Mr. Sa, well, the article I read was

5:39: makeuseof.com

5:42: It's an article from that place

5:44: Well, according to this article, NVIDIA

5:47: is a company that makes GPUs called image processing equipment and

5:51: It's a famous company, but that

5:54: The company will be held in Dubai in 2024

5:58: At the World Government Summit,

6:01: CEO Jin Sen Han said this

6:05: In the future, it will be necessary to teach people programming

6:07: It was predicted that there will be no need

6:10: But this makeuseof.

6:12: The reporter at COM said that

6:15: I don't think it will be

6:17: That article was written and the contents

6:20: There are exactly 4 points and error checking

6:23: It is necessary

6:24: I wonder why AI won't be like that

6:27: If you want to say, you need error checking and

6:30: One is that programmers can think of better ways to solve problems than AI based on their experience

6:33: The other two are that programmers can improve and refine the code to meet the needs of the user

6:41: The third is that they can improve and refine the code to meet the needs of the user

6:44: This

6:46: It was a little hard to understand when I read it for a moment

6:48: But when building an app or website

6:50: This is the interaction that takes place between the client and the business

6:53: This button

6:56: This color is good or not

6:58: Or features and things like that

7:00: When consulting, it's about consulting and improving.

7:05: Humans are better at making improvements.

7:07: In that respect, humans are better than AAI.

7:12: Humans change so rapidly.

7:14: They can respond quickly to tech trends.

7:17: These are the four things that can be done.

7:19: From the point of view of the reporter, Meta has asked AI to

7:26: It seems like developers and programmers are taking jobs

7:28: There's no reason to say that

7:30: The article is written like that

7:33: There are various opinions, but actually, I'm currently active

7:37: As a working engineer, what do you think, Sako-san?

7:41: What do you think? Um, in that MUD article

7:45: I think there was also that, but personally,

7:50: Error checking and the needs of the user

7:54: I really agree that it will improve according to

7:57: Well, when I say why, I think it's because

8:01: To be honest, I

8:02: I sometimes use that chat GPT or Jim

8:05: I don't have it, so I do my work

8:07: But I try not to rely on it 100% as much as possible

8:10: I try not to do that, so why?

8:12: To be honest, I tried it a few times last year or last year, when I started

8:16: If there is such a useful tool

8:18: I thought I should use it, so I looked up various things and wrote the code

8:22: I thought it would write the code, so I wrote the code

8:24: I tried copying and pasting it, but

8:26: I tried pasting it, but it still didn't work

8:28: There are so many things that don't work

8:31: There are also a lot of things, and the llm is a human

8:37: It understands what you feed it and breaks it down

8:42: It breaks it down and outputs it itself

8:45: It outputs it, so why

8:48: Or maybe the version was a bit old.

8:51: That's what happened, so I copied and pasted it.

8:55: The code was actually from a year or two years ago.

8:58: It was really

9:00: It happened many times, so I had to check it myself.

9:04: But when I wrote the code, I had to check it myself.

9:08: When I had it written properly in GPT or something.

9:10: I also had to go to the documentation myself.

9:12: And it was a plugin or library or something.

9:14: It was absolutely necessary to go to the documentation and check it.

9:16: That's what happened.

9:19: That's why.

9:23: Nowadays, it's more like that.

9:27: Plugins and libraries like that

9:30: It's amazing how fast

9:34: The version is really getting higher and higher

9:35: so the settings and the way to use them

9:40: Settings and configuration

9:41: What they say in chat GP

9:44: How

9:50: Look at the configuration and see this

9:54: You can still use this kind of pattern

9:56: I do use things like that

9:57: That's why you can't rely on it completely

10:01: Well, there were other articles too, but

10:03: is the one I read and that was

10:07: 56 questions, stack overflow?

10:10: Chat with questions taken from somewhere

10:12: Feed it to GPT Turbo 3.5 and eat it

10:16: Let it go and ask it a question and get the answer

10:19: When I asked it, only about 52 were correct

10:21: Really, one out of two were wrong

10:23: It seemed like they were making mistakes at about that rate

10:27: So in that respect

10:30: I still don't feel like all of my work will be taken over by AI

10:35: I see, so some of the work can be done by AI

10:43: But not most of it

10:46: So there's not a lot of AI

10:48: I think there are blood vessels, but it's something like academic

10:51: SBTC is a term used in academic papers

10:56: There is something called skill vice

10:58: There is something called technology leche

11:00: That's skill tendency change or skill change

11:05: It's like a god of technological progress

11:07: But it's like I'm going to get stuck, but in short

11:09: It's difficult, right, right, right, that technology, those who have that technology

11:12: Those who are skilled and have that

11:14: It's amazing when technology comes out

11:17: It's advantageous, but it's a little

11:19: Unskilled non

11:21: For skilled workers, that technology

11:25: That is, those who do simple tasks

11:29: There is a word for that, but it is

11:31: There is a word for that, but it is

11:33: I really feel that it is unskilled

11:39: What kind of people are workers?

11:42: And

11:43: When making websites

11:46: HTMLCSS and such

11:49: Creating the content that users see

11:53: Those who are now using Figma and probably designers

11:56: I think it is used by Mr., but normally

11:58: That's right

12:01: I think that person will come out

12:05: But yes, now the design

12:08: AI is also playing an active role in tools, so

12:11: If you ask them to make something like this

12:14: They will make it and you can actually use it as a website

12:17: There are even some places where you can actually do it

12:19: Yes, it's easy, isn't it?

12:21: Even among the programmers who do the work

12:24: The content that is considered to be easy is

12:29: The HTMLCSS that I mentioned earlier

12:32: Make the parts that the people who do it look at

12:34: The front-end and unit testing

12:38: I think that the number of jobs that are said to be entry-level

12:40: will decrease more and more

12:44: That's what I think, but overall, it's still

12:48: I think there is a need for programmers.

12:51: I see, thank you. Yes, AI too.

12:58: You know, every day,

12:59: new software and companies are coming out.

13:03: so it's really hard to keep up.

13:06: it's hard, but that's all for today.

13:10: yes, thank you, thank you.

13:12: thank you. Next week is the current topic of deep

13:16: is it a seek? Yes, yes, next week.

13:19: no, well, it's learning, so our

13:22: the channel is learning, so next Wednesday.

13:25: I guess so, yes, yes, deep seek.

13:28: I'd like to talk a bit about it.

13:31: Yes, thank you very much. Yes

13:34: Thank you very much, it's my first time today

13:36: There were some parts that were lacking because it was a radio show

13:40: I think there were some, but yes, I'm not satisfied

13:43: I can only think of some parts

13:45: Radio is a bit difficult, it's embarrassing

13:48: I'm embarrassed, or rather I'm nervous

13:50: Yes, yes, love

13:53: I'm going to get through it with my cuteness

13:56: That's right, I look forward to your continued support

13:59: Yes, thank you very much. Yes, okay

14:01: Thank you very much