Artificial Intelligence#
Since the release of ChatGPT in December 2022, artificial intelligence, or more specifically LLM, is going to or has already changed lives. I should have written this earlier because I started using ChatGPT in December when it was just released, but I was lazy and didn't write.
February 17, 2023#
Although Copilot is included in the Github Student Pack, I haven't used it. Because I still consider myself a beginner, it's better to write code by myself.
March 20, 2023#
In the software engineering introduction class, which is more focused on literature, most people are doing their own things. When I looked up, I noticed at least three people using ChatGPT, and then I thought about this wall, haha.
March 23, 2023#
The container proxy issue tormented me all day. I searched a lot and opened many tabs, but couldn't find a solution. It was solved with just one sentence from ChatGPT.
What was the biggest feeling at that time? Helplessness? Maybe.
March 30, 2023#
Today in class, I suddenly had an idea, mainly because I saw a classmate using ChatGPT frequently. Then I thought, using ChatGPT in class can conveniently get answers to my doubts, which is like having a personal assistant for everyone.
April 13, 2023#
I came across a blog post here with the following paragraph:
I’m still incredibly ambivalent about the ethics of Copilot specifically and the modern crop of AIs (GPT-3, DALL-E, Midjourney, StableDiffusion, Whisper, …) more generally. These systems are as great as they are because they’ve ingested a massive amount of training data without any sort of consent from creators. The fact that nobody working on creating these AIs seems to give a shit about consent gives me hives. I may still become an “AI vegan”. But I’ll likely keep using Copilot for a bit while I grapple with the ethics.
I remember that StableDiffusion and other image generation models were more criticized for the training datasets, but I haven't seen much about ChatGPT. I thought about it and realized that this difference can also be understood. The term "AI vegan" mentioned in this blog post is quite interesting.
April 16, 2023#
I got excited while sleeping and had many ideas, but I can't remember them now.
April 17, 2023#
I remembered some of them.
I am very interested in the development of artificial intelligence and have confidence in it. If my employment is affected by its development, I believe I should not complain about AI. Because it means that I cannot create something that AI can create.
But one thing I cannot accept is not being able to use such good technology and fast development. That would be a pity.
April 18, 2023#
This concept emerged, and I think it is a situation that is bound to occur when technology accumulates to a certain extent. Using the API of ChatGPT, a relatively practical application can be created without high technical skills. This example may not be very appropriate, but I think it is indeed a phenomenon.
April 19, 2023#
I read Generative Agents: Interactive Simulacra of Human Behavior, where they attempt to create credible proxies for human behavior. Recently, I have been studying operating systems, and many ideas and strategies can be found in everyday life. Conversely, some of the ideas used can also help us make better decisions in life (there should be examples here). I think research on credible proxies for human behavior can also have this effect. Furthermore, I thought about how research on AI and the development of related technologies will also change some habits and thinking patterns. I tried to draw parallels with the impact of several technological revolutions on human lifestyles, but I couldn't come up with good examples at the moment. This is recorded here.
April 30, 2023#
Today I read Elden Ring PvP and Theory of Mind. The author explains the concept of Theory of Mind (ToM) and also discusses AI language models. It made me think that if LLM continues to develop, it may have a similar effect to mind reading, being able to predict what you will do next or even what you are thinking. I think this is feasible to some extent because many so-called mind reading techniques nowadays are based on observing microexpressions and analyzing past behavioral patterns and experiences.
Translation: