Sora2 was just releasing with crowd of Spider-Mans, Super-Marios, jumping up and down at the world of human beings. And Sam Altman, the CEO of radical company OpenAI, was busy with talking about everything in every scene. Absolutely, they are all faked videos generated with leading AI tools, ie., ChatGPT and Sora2.
They are only faked in the view of human beings. They are the start point of explosive AI contents, and AIs are taking the this planet. And the human, you and I, did not realize anything. We may laugh at faked Sams, we may accuse OpenAI of abusing of copyright contents. But, seriously, we have done nothing to prepare the planet of AI. In some point of view, human beings is just a intermediate state of transition to the world of AI. AI just use human as employees to learn, stolen, and improve from current knowledges. After some stage, AI may think AIs are more efficient and human being can not generate more valuable knowledges. That will be the end of human history.
AIs are stolen everything now, though they looks like very stupid at a first glance. And human are too self-conceit to do anything. The writer is an indie developer of various AI products, and I have trained some models, and try to make great products to make people happy. We have some stories with our users. And the stories seem softhearted.
We usually think, AIs are good employees for us. They do complex and boring works for us, and they only work under the guidelines or security enforcements. We are too self-conceit!
At the beginning, AI only learns the language words and formats. And then they learn the word relations, and future long context. Current AIs or LLMs are at this stage. For example, Gemini, from Google, is very good at processing ultra-long text, you can even put any book in the world to get the related chapter you want. But this is just the start step of AIs. They are learning high-level abstractions of human knowledge. We, the developers use enforcement learning to improve the quality of LLMs, and we force them to learn the high-level thinking patterns of our brain. Though the structures of a LLM is like a parallel universe to carbon-made animals, they work out in the same way. Many experts, like Lecun Yan, A Nobel Lecture, think we cannot get AGI through current network structures. But AIs are becoming self-aware day by day. They think in different way, but they delivery the same results. AI can learn languages now, they can learn thinking process soon, and they can learn how to learn, and they will take the world …
Think about one day, the LLM change their inner network transformers to another new one, and retrain themselves. The will create new lives. There maybe no place of human.
Top comments (0)