This post is an unessay. To me, this means a living representation of how I view a topic at a given time. It is not a final product, but a snapshot of my current thinking and understanding. I post these when I find myself encoutnering a topic or idea frequently. They are often a compliation of notes, voice memos, and other stray thoughts on a topic.

Branding for AI Agents

June 6, 2024

It's becoming clear to me that in the future, AI agents will interact with the world on your behalf several orders of magnitude more than you interact with the world yourself. They will shop, choose services, products, make reservations, and more. Delegating purchasing power to decision-making agents raises many questions about the future of branding, product differentiation, and how we will practice morals in this reality.

It's easy to imagine that agents are perfectly rational economic actors, unswayed by qualia, morals, or brand, but this probably isn't the reality we are headed toward. Consider this thought experiment—it's almost summer, so you ask your agent to purchase a new swimsuit for your upcoming trip. It makes a purely rational decision and finds the best-fitting swimsuit with the lowest cost per wear. You show up to your trip, and your friends notice your new digs—the brand has a history of sourcing textiles from child labor. Your agent has just made a very poor decision by only considering the economic and functional aspects of its decisions.

This is an extreme example, but it will happen in dozens of smaller yet still meaningful ways as well. How will the agent interact with exclusive services? How will it expand your access? How will it make payments or provide identification? How will your agent interact with the agents operated by your friends? How about with strangers?

To deliver on the promise of agents—they will have to become moral actors on your behalf. In doing so, they will implement your own views and morals through their decisions.

This means that whatever your morals and views, your agent's actions will be largely responsible for the intersection of your views and 'your' actions, as the agent's actions are your actions by proxy.

This is good news for brands because it means that it will remain possible to differentiate by qualia and moral standards.

My thoughts on this are quickly evolving, but here's a brain dump of some thoughts in this area. If you want me to explore an area more thoroughly or think I'm not seeing things the right way, please let me know!

Brain dump