From Passive to Purpose: The Rise of Intentional AI

From Passive to Purpose: The Rise of Intentional AI

Last week I wrote about our growing dependence on ever-smarter AI, and how this risks creating a 'great intelligence divide.' To make AI work for human wisdom rather than against it, we need to actively build it around human agency.

That path opened up this week with DeepSeek's release of their V3 model - an open source AI system that matches the performance of leading commercial models, vastly lowering the barriers for technical and non-technical people with vision to exercise that human agency. DeepSeek says it achieved this performance with training costs under $6M by using a method that activates only the parts of the model needed for each task, and by making smart use of less powerful chips.

As foundation models become more accessible, the real value emerges not from the models themselves, but in how humans use them to create Intentional AI - systems designed to create real value for other humans. When the barriers to building AI fall, what matters isn't the underlying technology, but the human vision that shapes it into solutions that matter.

Consider one such builder, Piers, who is tracking corporate activities through AI. "What DeepSeek allows AI builders like myself to do is boil the ocean, or at least warm it," he explains. "Because of token costs, processing everything published by a major news platform each day has been prohibitive. Now it's 27 times cheaper - I can track everything mentioned about 600 companies daily without worrying about fees." This opens up entirely new possibilities for what individuals can build.

Samir, founder and CEO of Wizly, sees even broader implications for vertical AI companies: "With DeepSeek being an open source LLM, companies can use it to build custom small language models and vertical applications. You don't need massive funding to build application layers, you can innovate faster because of access to open source, and the quality of the LLMs makes this a groundbreaking opportunity."

In the social media era, our online activity - every like, share, and comment - became a passive data trail to be mined and monetized by platforms. Today's AI models follow a similar pattern—much of their apparent creativity stems from vast amounts of internet data, almost none of which was intentionally provided for this purpose. But there's a crucial difference this time: as high-quality AI models become increasingly accessible at lower cost and easier to use, we are entering an era where individuals can be builders, not just users. The small language models Samir describes represent a shift from passive use to active creation—enabling builders to refine AI systems for specific domains and missions, shaping AI to serve their unique needs.

This shift from harvested to intentional data, from general-purpose to mission-specific models, represents more than technical progress. When Piers processes corporate information at scale or when Samir builds vertical solutions for experts, they're creating systems that amplify human capability and purpose. What we're seeing is the emergence of Intentional AI. And this is just the start.

Get in Touch

I’d love to hear from you! Reach me at [email protected] or by responding to this email.