🧠 Codatta @ (codatta. io) AI learns to respect knowledge—— Following Jason @ jason_chen998's article, it actually sparked an interesting discussion: AI is still the current hot topic and the hot commodity of attention; Since OKX has already stolen Sahara from Binance, is it possible for Binance to seize Codatta, which OKX has invested in? AI is constantly evolving, and no one knows who will be the king in the future, especially if you have noticed that big models are becoming more and more like exam masters. As long as the "question bank" is used frequently, enough emoticons are added, and the answers are written long, it can score high on the leaderboard and reap a wave of "smart illusions". But do we really need such a model brushing game? What we want is not for AI to be better at exams, but for it to truly understand our problems, know how to solve them, and know what we really want. At this point, the problem arises: Who should AI learn from? How to make human expertise a trustworthy, traceable, and incentivized infrastructure for AI? one ️⃣ What truly makes AI smarter is never more parameters, but human knowledge itself—— In the past three years, we have seen the transition of models from GPT-2 to Claude-3, with people crazily increasing parameters and piling up computing power, trying to squeeze out the limits of intelligence. But a consensus is emerging: the ceiling of AI is no longer computing power or models, but knowledge. And true knowledge is in the hands of millions of human experts. Doctors, financial analysts, traders, pathologists, and on chain security researchers; These knowledge can never be "generated" out of thin air by GPT. The problem is that these experts have no motivation to appear in OpenAI's training set. three ️⃣ What Codatta wants to solve is the problem of 'experts being silent in the AI world'. Many people may ask what the relationship between Codatta and Sahara is: If Sahara makes data 'usable', then Codatta makes data 'understandable'. Codatta is a decentralized knowledge layer built on the Web3 infrastructure, By using on chain authentication, DID identity system, pledge endorsement mechanism, and usage based revenue distribution, professional knowledge can truly become a "tradable asset" and contributors will be continuously rewarded. One sentence summary of project positioning: "Let AI subscribe to human knowledge and let knowledge producers own data assets. ” It is not rebuilding a crowdsourcing platform, but restructuring the supply and demand relationship of knowledge assets: 1) Each data/evaluation will be recorded on the chain, auditable and traceable; 2) Professionals with DID reputation and data ownership can receive continuous rewards for contributing valuable data; 3) You can stamp to support data labels that you approve of, and even crack down on cheating; 4) Data assets can be traded like NFTs, and even packaged into "knowledge funds"; 5) Model evaluation is no longer a 'manufacturer's black box operation', but a 'user random question, on chain voting', with no way to cheat. One sentence: Let AI learn to pay for knowledge, learn from humans, not from rankings. four ️⃣ The problem solved by Codatta: Trust crisis in model evaluation: Existing evaluation platforms such as Chatbot Arena have serious "brushing" space, for example, model manufacturers can test multiple times, select the best version for release, and lack transparency; The incentive gap in data supply: high-quality knowledge (such as medical image annotation and legal text annotation) still relies on expert manual labor, while traditional platforms (such as MTurk) cannot provide sustained value distribution; The ambiguity of intellectual data property rights: Professional data is privatized by platforms, and original contributors cannot enjoy its derivative value; The rupture of professional identity: Experts are unable to migrate their reputation and achievements across different platforms, resulting in duplicate authentication and "information silos". five ️⃣ This is not a new platform, but a new order—— As a pioneer in the integration of AI and Web3, Codatta not only creates fair revenue opportunities for data contributors, but also provides high-quality and reliable data solutions for enterprises and institutions. When AI no longer belongs only to a few large companies, When every contributor can be recorded, rewarded, and respected, and when knowledge can participate in collaboration, circulation, and distribution like assets, we truly begin to build a 'future where humans tame AI'. six ️⃣ Basic activities and entrance Participation entrance: https://app. (codatta.io)/referral/i696622t You can find related tasks on the main page; I feel like we can start with it first. TGE is approaching, and it's a wave of ups and downs!