The most interesting project I've seen recently is undoubtedly FLYBRAIN. Tell me about the latest progress of this project. Previously, I watched a fruit fly use a program to issue coins. Now, the developers have started training it to simulate cryptocurrency trading and have even arranged a diary account for it. They have invited the community to develop new gameplay together. The goal is also very direct: to cultivate a millionaire fruit fly. Humans are still searching for the code of wealth, while fruit flies have been arranged for employment. What's even more outrageous is that human cryptocurrency trading relies on news, while its coin selection relies on "hearsay". What exactly is going on? @Fruitflydev has built a simulation model using publicly available fruit fly neural connection data, and now has created a dedicated trading page for it. Real Pons tokens are drawn as cards for it to see; The coin name is mapped to a simulated scent according to artificially set rules, allowing it to 'smell'. When the model shows approaching or avoiding, the program converts these reactions into buying and selling intentions. Simulate making money and give reward signals; Simulate a loss, give a punishment signal, and try to make it change its reaction next time. Don't rush to apprentice, it's still a simulation drive for now. It does not understand K-lines and does not study market makers. The current test is whether this model can generate effective feedback from transaction results. The developer's idea is to pay the creator's transaction fee to try real money transactions after the training is completed. Note that this is a plan, not already implemented. The online claim that 'fruit flies earned hundreds of thousands of dollars' refers to the creator fees generated by associated tokens, not that they defeated humans through transactions. Collecting tolls and working as a trader are two separate jobs. In addition to learning cryptocurrency trading, it also has its own diary account @ imflybrain. The language model organizes its observations based on running data. It is responsible for movement, AI is responsible for narration. It's not that fruit flies suddenly passed the English CET-4 and CET-6 exams. The latest code also includes a Drosophila experiment: placing two models in the same simulation space and observing how they respond to visual, olfactory, and auditory stimuli. This is not just two fruit flies starting to talk about the market, but continuing to test what this neural model can do. The developers also publicly corrected code errors in the reward signal and disclosed failed tests. So what we see now is an experiment of making changes while doing, not a money making machine that has already been successfully trained. More noteworthy is that the developers announced a $10000 reward to encourage the community to replicate code and train their own fruit flies, with two selected works sharing bonuses. From one person raising fruit flies to a group of people researching how to raise them. I think this is the most interesting change recently. The fruit fly issued a coin "is a news article that passed after reading it. What did this fruit fly do today to give people a chance to come back tomorrow. Will you learn how to trade? What new job will the community assign to it? What unexpected results will be achieved in the next experiment? It has started to have a story that can be continuously updated, rather than just a meme used during coin issuance. Of course, having a bonus does not necessarily mean having an ecosystem. It still depends on the actual work and continuous updates in the future. If more and more people participate and conduct truly interesting experiments, FLYBRAIN will have the opportunity to grow from a curiosity project to an IP jointly shaped by everyone. The millionaire fruit fly has not been cultivated yet. But it's already quite interesting to make a group of people start caring about what a digital fruit fly will do tomorrow.