AI has been a topic of ongoing discussion. During a live event a few days ago, I expressed my views on current AI. There are mainly two types of AI in the cryptocurrency industry, one is the expectation based AI mainly based on Meme, which is more like investing in concepts during the ICO era. While this type of AI has good imagination, it is slightly less effective in implementation and difficult to provide practical assistance for trading. And the other type is the actual AI that assists in trading. This type of product has some embryonic form in the market, but there is no product that can truly assist users in trading, whether it is a mass-produced version or a quantified version. Therefore, fundamentally speaking, the AI in the current cryptocurrency industry is still concept based, which is why there is a big difference in AI in the US stock market. Although they are all called AI, we have to say that AI in the traditional market has more practical applications, and it was mentioned in the last Trump's administrative order that it will help the American AI and cryptocurrency industry actively. Therefore, if there is real AI in the cryptocurrency industry, we may enjoy double dividends. Today I saw the letter from @ Natalia Ameline, founder of @ MetisL2, which actually touched me quite a bit. After all, as a second layer network of Ethereum, the development of ETH itself has encountered difficulties and doubts, let alone how difficult the second layer network will be. Therefore, fully embracing AI in the transformation is not necessarily a breakthrough. As I mentioned earlier, if cryptocurrency and AI applications can be truly combined, it is very likely to enjoy double dividends. Therefore, fully embracing application-oriented AI in the cryptocurrency field is likely to become a trend. Projects that already have public chains and application technologies in this area will indeed move faster. In addition, I saw Metis' letter about solutions to "data problems", which I am also quite interested in. I have been learning about this topic recently, but the current solutions are still relatively vague and lack consistency in answers. I don't know what kind of solutions Metis will use, so stay tuned.