I sat down with Michael Casey, author, journalist and founder of DAIS, the decentralized AI society, to explore the profound questions emerging from our accelerating relationship with artificial intelligence.
Casey opened with a simple but urgent question: how do we keep humans valuable in the AI economy? The conversation quickly moved into deeper territory.
“Our creativity, our capacity for imagination, our ability to recognise patterns that are unpredictable,” he says, are all uniquely human traits. “I think a lot of all of that defines the human condition.”
He warns against viewing intelligence through a mechanistic lens.
“Mechanistic approaches tend to be linear. This causes this because of that,” he says. “It’s a very deterministic way of defining action.”
In contrast, human thought is shaped by randomness, serendipity and a rich layering of experience that machines cannot replicate.
In this new AI-driven economy, Casey argues, data is more than a by-product. It is an extension of our personhood.
“Our lives, our personhood, is now defined in digital terms,” he says. Protecting and controlling that digital personhood is essential if humans are to retain agency in the future economy. He speaks about the emergence of the “intention economy,” where people would actively manage and communicate their wants rather than being passively tracked by invisible cookies.
“I absolutely want some product, and I can prove it by the fact that I control the data around that message,” he says.
To Casey, data ownership is more than a technical matter. It is a moral and political imperative. Without it, he suggests, humans risk becoming devalued participants in their own society.
“Making human data scarce and controlled raises its value,” he says. “In the old model, we diminished the value of the human being because it was just this big wave of aggregated social media data.”
He also addresses the argument that data ownership might exacerbate inequality, given that wealthier individuals’ data could be seen as more valuable. Casey acknowledges the challenge but rejects the idea that this should prevent people from owning their data.
“The answer is not to say, don’t have control of your data,” he says. “The data could actually be an equality thing if everyone’s data is equally valid.”
Interestingly, he pointed to examples where data from marginalised groups is in fact more valuable. The Human Genome Project, he notes, lacked significant contributions from the African Caribbean diaspora, resulting in important gaps.
“Science may well find that there’s more value in the marginalised people,” he says, a shift that could have profound implications for research and healthcare.
Protecting data is not just an individual concern. Businesses too face challenges in managing and unlocking their own valuable information. Casey speaks about the reluctance of companies to expose sensitive data, noting that corporate data is often more closely guarded than personal data.
“If I’m Goldman Sachs, I’m not going to give my proprietary trading data to OpenAI,” he says. “You don’t know what’s behind it. It is a black box.”
The problem is structural. While large language models have been trained on vast pools of public data, “the most valuable data of all”, ie data locked inside companies, remains out of reach.
“We’ve got these silos of data,” he says, “and that’s where we’re going to find and solve the problems.”
The key to unlocking this data without compromising privacy or intellectual property, he suggested, lies in emerging technologies like confidential computing. Using cryptographic techniques such as zero-knowledge proofs and trusted execution environments, it is possible to run computations on encrypted data without ever exposing the underlying information.
“You don’t need to know all the underlying details of a data set to be able to run computation on it and come up with conclusions,” he says.
Confidential computing could allow businesses to contribute their data to AI models while maintaining complete control over its use. This, Casey argued, could shift AI from being a “rapacious” collector of information toward being a tool that solves real-world problems like climate change or disease.
“The goal should be: can we cure cancer? Can we solve climate change?”
The final theme Casey explored was perhaps the most subtle: the way language shapes our understanding of AI itself. In particular, he highlights the risks of anthropomorphising machines.
“Anthropomorphic means that we’ve attributed human characteristics to something that is not human,” he says.
Words like “learning” mislead us into thinking that AI systems think and grow like humans do. In reality, most AI systems operate through reinforcement learning, a system of programmed incentives rather than genuine understanding.
“I’m not sure that’s exactly how we learn,” he says. “Learning is something that we take in from the outside world.”
Casey shared an exchange he had on LinkedIn where a debate arose around whether we should be polite to AI. This was prompted by a comment that politeness wastes tokens and processing power in language models. When someone asked ChatGPT for its opinion, it responded that politeness was important for humans themselves, helping preserve constructive norms in society.
Casey initially defended the idea that we should be polite, arguing that human behaviour even toward machines reflects and reinforces how we treat each other.
“Our own survival as a species depends upon us being nice to each other,” he says. He worried that normalising rudeness to AI might corrode human civility in broader society.
However, an interlocutor challenged him, arguing that such an approach risks anthropomorphising AI and conferring it a moral status it does not deserve.
“They all should be treated as tools,” Casey concludes. Ultimately, he found himself persuaded by this argument, at least in part. “We should be creating that distance, that objective view of the technology. AI is just a tool.”
Michael Casey’s vision for the future is neither naïvely optimistic nor apocalyptic. It is grounded in a belief that with careful thought, clear boundaries and new technologies, we can build an AI economy that enhances rather than diminishes the human experience. His words are a reminder that the future is still, very much, in our hands.
