Some Idea about AIGC

I recall the events of the outbreak at the end of 2022 when AI was able to generate pictures with just a few prompts, or even more accurately if desired. Then at the beginning of 2023, the chatGPT (module GPT3.5) was released and entered into the public sphere, creating the illusion that there had been a recent breakthrough in AI-generated content. Despite my initial skepticism towards AI’s potential applications in our daily lives, I found myself engaged in arguments with people on the internet once I started using chatGPT as an assistant.

In recent months, I have come across numerous articles, telegram groups, news items, and software related to AI-generated content. While I recognize that AI is a time-saving tool, I am increasingly wary of its proliferation and wonder if we should take a moment to reflect on what we stand to gain and lose from its use.

While it may sound clichéd, having a thought process is essential and can help one articulate their goals more clearly. It is important to recognize that AI still requires content to be fed to it for training purposes and does not inherently “understand” like humans do. Its intelligence is derived from a corpus that is open and marked through many years of efforts such as essay writing, marking, experimentation, and so on. The most crucial part of the process is the marking phase.

Therefore, the more garbage that is present on the internet, the more garbage AI is likely to output. As someone once said, “because some of that is reflecting back whatever trash these things have been fed on.”

Naturally, AI is like a mirror that reflects the content you input. This is why I am not optimistic about the AI that is currently being trained by domestic companies, given the prevalence of garbage content on the domestic internet. For example, when I encountered Baidu’s Wen Xing Yi Yan, they claimed it was nearly equivalent to chatGPT, but in reality, its performance resembled more of a chatPPT than a chatGPT.

It seems like I have a negative opinion about whether companies like Baidu can create an AI like GPT. I must admit that I do have doubts, but I also hope that a competitor will emerge and motivate them to take action instead of waiting for GPT to dominate the market, leaving no space for anyone else. The old ways are becoming obsolete; now we need to create rather than just copy from others.