Machine Learning, Random Thoughts

Developing Artificial Intelligence while Developing Human Intelligence

This one is going to be more personal than usual, so buckle up, friends. Content warning – I talk a bit about pregnancy in this post.

Developing Artificial Intelligence

I’ve spent most of 2023 focused on AI after a decade of metaverse technology being my primary focus. I’ve actually been working with artificial intelligence since 2015, but I was primarily using it as a means to an end in my VR and AR projects, rather than as a standalone technology. When I moved from my role as Director for the Mozilla Hubs team to the Innovation Ecosystem Development lead earlier in the year, I put my metaverse projects on hold in order to better understand the landscape of the open source AI world. As it turns out, there’s a ton of overlap in the product domains of AI and metaverse, because while the core enabling technologies and their interaction modes look quite different from one another, the entire premise of the advancements and opportunities are grounded in emergent behaviors of computers simulating people and reality. They’re also both computationally expensive, paradigm-shifting, and deal with complex nuances of literally codifying human ethics into software.

I joke sometimes that my entire career to date has been about Learning How to Human – that I was drawn to social VR and metaverse platforms because my neurodivergent self wanted to experience a taste of a world that I could both understand, navigate, and flourish within. Working on XR gave me a unique perspective to study social and environmental simulations, and now, shifting to AI, I have the chance to more deeply study the translation of human learning and intelligence to machine simulations.

The hype in artificial intelligence seems more harmful than the hype that surrounded the metaverse several years ago, but the core challenges remain – centralization of technological capabilities and dependencies on a small number of providers, surveillance, unchecked advertising risks, scams, mis/dis-information, inequitable access to benefits, economic exploitation and privacy – to name a few. 3D virtual worlds bring joy to computational interactions that chat bots just can’t – but I digress.

Developing Human Intelligence

The human baby that I’m carrying in me right now, at just over 14 weeks gestation, is already showing signs of awareness of their (admittedly alien) environment. They are practicing goal-directed movements, recognizing how they can take an action to get closer to accomplishing a task. Before they are given any data beyond the fluid in which it resides, their senses are developing – something that many computers that are able to be programmed to act “intelligent” never get to.

It surprises me that I don’t see more research studying how to build technology that learns more akin to how humans develop. We restrict an agent’s “senses” to a particular set of inputs, and most of the time, we don’t give these agents unmediated environmental sensor information at all.

There’s a common misconception that our brains are “fully developed” at age 25. Ask anyone with experience in psychedelics, HRT, or therapy, and you’ll perhaps come to the realization that our neuroplasticity far outperforms what we understand, can monitor, and can classify. Reading about how pregnancy results in chimerism, for example, is just one area where we can see that we don’t fully understand the relationship between the self, one’s environment, and the other people we are around. Some language used to talk about how a pregnancy changes the brain compares it to a second period of puberty. I certainly already feel like the experience has profoundly changed me.

For one, I’ve now experienced what it’s like to be physically sick for weeks at a time. My perspective on ability and disability changed pretty substantially back in 2019 when I heard Adrian Roselli talk about how we’re all just temporarily able-bodied, and pregnancy exists (at least in the United States in 2023) in a place where massive physical and mental changes occur, but until you’re “far enough along”, you’re not really seen as experiencing a disability. Parental leave in the United States is atrocious, which means I’ve spent the last three months with what feels like a perpetual stomach flu, difficulty sleeping, and a level of exhaustion that I wasn’t able to imagine. No amount of reading about pregnancy prepared me for the reality of having to show up and work back to back, 16-hour days on a work week while throwing up and dealing with the worst depression of my life.

Developing Artificial Intelligence in the Service of Human Intelligence

If we invested half of the time, money, and energy that we’re collectively spending on “AI” technology and put that into educational experiences, we’d be in a far better place than we are today. The people championing AI as the solution to all of humanity’s problems have constructed themselves a reality where private corporations are the saving grace of a society at risk, under the misguided belief that the “free market” is the best and only way forward.

Technology has always had benefits and harms to society, but I’m increasingly reminded of the description of the United States government in Snowcrash as we face regulatory capture by the biggest tech companies.

All of that aside, I still come back to the power of open source and community collectivism. In some ways, it’s getting harder and harder to find the people whose views I share, and the regular conflict that emerges from YAOSWFOA (yet-another-open-source-wrapper-for-open-ai) being championed as the slightly improved option over AI services because they “might have support for more open models soon” is harder to bear than it used to be. Maybe because I’m tired all the time now.

To end on a more optimistic note, though, I’ll share some brightness that keeps me going:

  • The brilliant work of Bogdana Rakova, whose ‘Terms we Serve with‘ and ‘Speculative Friction‘ projects continue to showcase a brighter, more inclusive, and technofeminist future for the world
  • Monika Bielskyte’s ‘Protopia Futures‘ framework, which breaks out of utopia/dystopia binaries and into sustainable protopias
  • Everything that the Distributed AI Research institute publishes, especially their willingness to speak truth to power around Big Tech’s right-wing ‘TESCREAL‘ mentality

Things are hard right now, but these projects make me feel more optimistic.