Albert Castellana, co-founder and CEO of Yeager.ai, believes that artificial intelligence (AI) can push blockchain-based applications beyond simple fixed contracts. AI integration opens up “more objective decision-making,” and opens the door to entirely new use cases, he said in written responses.
Risks Associated with Centralized AI Models
Castellana said he envisions developers building applications that make autonomous, data-driven decisions without human intervention, potentially enabling AI-powered oracles without relying on central intermediaries. He says the skill will revolutionize everything from insurance contracts to supply chain management.
While Castellana, a serial entrepreneur and crypto industry veteran, recognizes the benefits of centralized AI, he highlights the lack of transparency that he says undermines trust in its results. Decentralized AI, in contrast, allows for multiple models and validators to be involved, thereby reducing the risk of bias and manipulation.
Castellana also explores the future of AI and blockchain technology, emphasizing the importance of staying informed about these evolving technologies. Below are the Yeager.ai co-founder’s responses to questions submitted.
Most AI models today rely on data stored on centralized servers. Now, thanks to the advent of blockchain technology, we are witnessing a wave of decentralized AI gaining momentum. For the benefit of our readers, can you explain the difference between centralized and decentralized AI?
Albert Castellana (AC): The separation between centralized and decentralized AI is determined by three main aspects: who trains the models, who performs the inference, and the nature of the models themselves. When thinking about decentralized AI: training models is a very complex and expensive task, which makes it difficult for small teams to do, but some companies like Meta have decided to open it up; running models is really difficult in a distributed way, so a lot of work is being done on inference testbeds that allow you to verify the success of the implementation.
However, to create decentralized applications, not only the computing but also the decision-making process needs to be decentralized. Each model is centralized in its own right; with its own biases, characteristics, and understanding of the world. It is opaque and difficult to audit. So when making decisions, relying on a single model means having a centralized model.
At Genlayer, our goal is to decentralize decision-making by engaging multiple models and leveraging blockchain to reach consensus that allows them to agree on the outcome of personal tasks. Think of it as moving from a single-judge system to a multi-judge or even jury system, where multiple perspectives come together to produce a more accurate and fair outcome. This democratizes AI and mitigates the risk of a single point of failure or bias.
BCN: The emergence of decentralized technologies has highlighted the vulnerabilities of centralized systems, particularly in areas such as privacy and data security. Given these issues, should internet users who interact with AI solutions, knowingly or unknowingly, be concerned about the potential risks associated with using such solutions?
A.K.: Over the years, we have seen most users trade utility for privacy, often without a clear understanding of what they are giving up. Many have accepted the trade-off of handing over personal data in exchange for the ease of using services that simplify their daily lives.
With AI, this problem will get worse at first and then better. The better the AI gets, the more information we’ll be willing to share with it. However, the better decentralized technology gets, thanks in part to AI, the more we’ll be able to take back control of our data.
If we want to speed up the process, we need to educate the public, encourage open source tools, and promote decentralized AI.
#BinanceTurns7 #MarketDownturn #Ai #elaouzi #IntroToCopytrading
