Milan, Italy, 30 Dec 2021, An executive summary of a very ambitious project had catched our eyes, and it reads “ZENFI AI provisions an Intelligent Metaverse that allows homo sapiens to create emulations of their mental state (i.e., mind and soul) that are encoded into intelligent NFTs and placed on a decentralised and everlasting storage medium.” If you find this is science-fiction, we are sure you are not alone. However, as a number of seemingly science-fiction stories we watched on Hollywood movies would eventually come true, let us spend some brain effort on analyzing if there is any realistic basis to ZENFI AI’s ambition.
Our primary questions are on how one could create AI-Agents with goals and personalities on-chain, and how now-lively on-chain inhabitants preserve the integrity of their creations and memories .
From the whitepaper of ZENFI AI, they indicate that “… whole brain emulation (WBE) is studying the theoretical futuristic process of scanning a physical structure of the brain so accurately that the result can be deemed an emulation of the mental state. This emulation can then be transferred or copied into a digital form. Vocal supporters agree that tools and ideas needed to achieve mind uploading already exist or are currently under active development; they are in the realm of engineering possibility… It is discussed in artificial intelligence (AI) research publications as an approach to strong AI and to at least weak superintelligence. ” and then “The digital transformation of one’s self and intelligence needs to be stored on an everlasting storage medium so as to ensure its long-lasting existence. To this end, blockchain technology and more specifically, Non-Fungible Token (NFT) is a prime and most natural candidate. The end result would be an intelligent NFT.” These arguments at least suggest that they know a way to bring a representation of humans’ intelligence on-chain.
The project also raises the notion of a Train-to-Earn model, which basically attempts to marry financial incentives with AI’s well-studied training pipeline. Their website hints at a mobile application that engages users in answering logical, IQ questions. It then collects the users’ answers and then uses such a data set as ground truth data to train its AI models. The users are rewarded with ZENFI, which is the ecosystem’s native token, for their efforts. In this spirit, if the app is sufficiently well-designed and the questions are adequately engaging, there will be an additive element to it. Gamification would be a nice approach to explore.
So, our verdict is that while ZENFI AI’s attempt may appear bold, it is still within the realm of engineering possibility.
• Website: https://zenfi.ai
• Email: firstname.lastname@example.org
• Twitter: https://twitter.com/ZenFiAi
• Telegram: https://t.me/ZenFiAI