När når vi AGI?

I had to bury my talent yesterday.

A few days ago, DeepSeek v4.1 was released, pushing the height of the small model capability upwards into a notch.
The pace of AI is growing far ahead of everyone’s expectations. From the first version of ChatGPT, which is only a few thousand tokens to chat with only a babbling language, to the inference ability of OpenAI o1, Deepek R1, and Kimi K1.5 Thinking, it is only two years; from the inference model to the agent that can now smoothly execute commands in various harness tools and complete complex tasks, it is only a year and a half. It is difficult to imagine that if you wait for another year, two years, three years, what kind of AI will become at that time, how powerful it will be, will already have the ability to evolve itself, and penetrate deeply into the fields of personal intelligence.

You can make an image analogy: you are proficient in sweater technology, especially good at the weaving of various patterns and color matching. The sweater you knitted is of excellent quality and beautiful, and the rich people in the ten li and eight townships have come to ask you to knit sweaters for them, and you have made a lot of money by this. At the same time, you enjoy the feeling of sitting at the window, making a pot of tea, looking out the window at the green mountains, green water, cattle and sheep and cooking smoke, and quietly knitting a sweater in the afternoon. But one day, someone invented a magical machine, just provide wool and patterns, you can automatically knit a sweater, the quality and texture are no less than you personally woven, and the speed is far faster than you. You know very well that your peers can easily reach the level you once had with this machine, so you have to use it. You also know that with the sweater technology you have saved in the past 20 years, even if everyone has a machine, the speed and quality of your

The Last AI Built by Humans: Toward Genuine Recursive Self-Improvement
Recursive self-improvement (RSI) enables AI systems to turn experience and feedback into persistent changes that improve both their capabilities and the process of future improvement. We first use the Headroom-Closed Index (HCI) to reveal the problems of existing LLMs, then introduce the RSI concept and its development roadmap: from improvement-execution autonomy, improvement-strategy autonomy, experience-acquisition autonomy, and environment-adaptation autonomy, to recursive meta-improvement. Next we examine RSI across scenarios (e.g., scientific discovery, embodied intelligence, software engineering), highlighting their distinct requirements and development speeds. Drawing on diverse industry practices and preliminary empirical evidence, we connect RSI research with practical systems and identify key challenges to achieving genuine RSI.

President Trump: We're not going to let that happen. It's a hoax!
Jensen Huang: That's right, we're not going to let that happen sir.