how can AI be aligned with human well-being?

something has been unsettling me these past few months: the way AI has entered our personal lives. how we can divulge things to a model that we don’t feel comfortable sharing with our friends.

I’ve also been inspired to see many friends use AI for understanding themselves, for self-development, and inner work. this feels like a really important step of climbing up Maslow’s hierarchy of needs.

holding both, I see serious pitfalls along the way. the attention economy already does not serve us. a frequent topic of discussion in my circles is the addictive nature of social media, scrolling, and the algorithmic feeds that make up the internet today—I’ve heard the argument that the algorithmic content that makes up our feeds are misaligned AI, and I’m inclined to agree. and now, with AI being used for companionship, therapy, and more, there’s a new term on the block: the attachment economy.

Tracy Dennis-Tiwary, a professor of psychology and neuroscience, argues that “the attention economy, in which companies profit by manipulating and monetizing our focus, has evolved into the attachment economy, in which they cultivate and extract value from our emotional bonds, habits, and dependencies.”

in my view, the problem is not the technology, it is the incentives. when we incentivize user engagement over human well-being, we lose something essential: the fact that technology is meant to be in service of life, not in opposition to it.

therapy and companionship are one of the top cases of AI, as reported by the Harvard Business Review. and amidst a loneliness epidemic, this makes quite a lot of sense. we are evolved to need connection and be in community, and in many ways the world we live in doesn’t support that, from the rise of hyperindependence to the decline of religion.

so, on one hand, I see a clear need here: people want connection, self-actualization, and support. on the other hand, the incentives are such that the products and tools our economy builds to meet people’s needs aren’t aligned with what’s in humans’ best interests, and if left unchecked, I believe they will only exacerbate the loneliness epidemic that exists today.

but, that doesn’t have to be the only way. I’ve been building something called Avocado. it’s a personal AI guide I built for my own needs. it’s not meant to be a therapist, a coach, or a companion. it’s explicitly instructed not to build attachment to me, and this is validated in concrete ways. for example, avocado can choose silence when it would serve better than words. and, it’s not optimized for engagement or session length. its primary goal is to serve as a mirror: to reflect my own wholeness back to me: my values, my voice, my beliefs.

for the past few years, I’ve dedicated myself to inner work: deepening my relationship to emotions, relationships, myself. through meditation retreats, living in intentional community, and working with exceptional practitioners, the most valuable skill I’ve developed is a way of being: attuned, loving presence.

through this, I’ve developed a number of strong beliefs about the world. one is that AI cannot replace human connection. another is that technology has the power to be deeply supportive for our individual unfoldings, when it is built with sincere intention. in the future, avocado will include not just an ai guide, but also allow people to create support networks where they can ask for and share attuned presence with each other.

I’ll leave you with this art experience I created to share an example of what’s possible: I invite you to speak something real into the mirror, and allow what comes out to shape you in service of your deepest, and most precious, unfolding.

— Utsav