AI is getting easier

The human questions are not

Waiting for the plane to move to our arrival gate, we sat motionless on the tarmac at San Francisco Airport. As I looked out of the window, I contemplated the potential discussions at the Wisdom & AI summit happening the next day. I anticipated discussions about the risks and opportunities for rapid AI acceleration and how we may influence these outcomes from a more human perspective. As I did so, I found myself increasingly aware of what was happening right in front of me. Infrastructure, fuel trucks, logistics, systems, traffic controllers, buildings, and metal walkways. It struck me that our life occurs at a very physical level, and that every experience rests on enormous unseen complexity.

The hidden infrastructure

Suddenly, the discussions about possible AI acceleration felt very lofty and abstract when compared to the largely invisible human and material infrastructure already underneath the AI in our pockets today. How interesting to be arriving at the Port of San Francisco for this event, I thought. This is a place that has seen other eras of significant growth and impact. I pondered the 1849 Gold Rush in California, where rapid growth and prosperity, based on the extraction of physical resources, occurred for some at the expense of others. Later, during the boom of Silicon Valley, the focus was on driving growth and prosperity in the use and provisioning of information. For decades, this too has had its own material infrastructure impacts, largely far away from consumer attention, as well as the more direct mental well-being concerns associated with the social media applications that now sit expectantly on our devices. I have worked on many aspects of how society intersects with the technology life-cycle and have seen these impacts play out. Now, with AI, the currency is not just material or psychological; we are talking about the use and provisioning of intelligence itself. So, as I waited to arrive in San Francisco that day, the story of this AI era felt all too familiar, with its own material and human impacts, yet this time also likely to have some wildly different consequences.

On information and relationships

Dinner conversation that night with conference attendees was equal parts fascinating and startling. Some participants were concerned with how to make sure that we can maintain the integrity of wisdom teachings as AI assistants become a default consulting and counselling interface for people. This concern feels valid to me. I have seen for myself that hallucinations in AI are a very real challenge and one that can be compounded if the user has no way of validating the response given. While the information age has enabled ancient wisdom to reach far more people than ever before, the risk of dilution and pollution of its teachings is already with us.

More startling to me, though, was a conversation with a young college researcher who casually mentioned people they knew in (romantic) relationships with AI. I was most startled by how normal this young researcher found that to be, even challenging me with why it would be a concern. The 2013 Spike Jonze movie “HER” comes to mind here. I want to believe that there is something special about human-to-human interaction. I believe in the co-regulation that occurs when sitting with another human being (i.e., one with a heartbeat). I want that to count for something, and it concerns me that we may be experiencing an approximation of emotional intimacy without its true benefits, and that this may have longer-term consequences on us as a species.

Still, that moment and other conversations through the event revealed something important: AI is not arriving as a tool or technology. It is arriving as companionship, workflow, a learning partner, and an invisible assistant. For many, especially younger generations, much of this already feels normal. Over a breakfast meeting with a friend who was also attending the conference, we discussed the lack of “friction” in AI interactions. This goes far beyond concerns of well-reported sycophancy, which are too real and very dangerous for normal minds, let alone vulnerable populations like children. The lack of friction with AI means that we may not be intellectually challenged by our collaboration partner. I am reminded of how animation giant Pixar would invite its best designers to watch new releases and openly critique each other’s work. That is an example of friction built into the design process of a movie. This type of process results in a better outcome. It’s possible to prompt your own “friction” into an AI interaction, but if we don’t do that, we will likely see people’s creativity flattened over time and outputs generalized by a lack of experience-based pressure testing.

The invitation to humanity

Participating in the conference itself was a little different from what I had anticipated. I arrived expecting debates about AI risk, governance, productivity, and how humans might remain safe and grounded through rapid technological change. There was an entirely deeper question on the table, though: not simply how we manage the acceleration of AI, which feels rather uncontrollable, but whether humanity itself must accelerate alongside it and what might that look like? Soren Gordhamer, the summit host, put it this way:

“Perhaps the emerging AI is an invitation of sorts for humanity to better understand who we are and what really matters. And there is no time to lose.”

I am not here to debate the more existential and political risks of AI nor whether we can put the genie back in the bottle. We are already in the middle of it all. Based on my conversations with people over these days, I see two things that are embedded within this invitation. Two things that I think we need to understand, and with the right awareness and engagement, we can use to nudge ourselves into the future.

Firstly, AI is getting easier and more comfortable to be with in all aspects of our lives. It’s happening at breakneck speed. That’s great. Truly, it is helping me and countless others every day to learn and work at a pace we didn’t think possible. Science is going to evolve in unprecedented ways, which is likely to have real human benefits. In terms of relationships, I know we are successfully democratizing a basic level of self-help and relational experience for people, and that too is wonderful. It might be too easy, though. We know in theory and in practice that “the magic lies outside our comfort zone”. We build resilience through exposure to difficulty, not through making things easier. Relationships with real people have conflict, and those with shared approaches to resolving conflict together are stronger. Are we willing to give up on the struggle and effort in how we learn and converse? Are we willing to sacrifice the quality of intelligence for speed and convenience? Will we evolve AI to be more constructive for us? Conversely, although we may want humans to be in the mix, are we actually going to be the most intelligent judges of what supports human well-being anymore? As I reflect on these questions myself, I am struck that many of us won’t have a choice as to how much AI is in our life. Those who can afford to choose a more traditional path for our work and creative endeavors may be a minority, and these won’t be where lucrative roles are found. Even getting employed today requires you to understand how to modify your resume so it gets past the machine. We are bending. Widespread adoption of AI systems for our work, contemplation, and relationships will become the norm. As this happens, we may need to meet different versions of ourselves with different values and understand which psychological capacities and tolerances will erode, whether we value emotional resilience, and what happens to creativity without tension.

Secondly, the large (and increasing) physical-world consequences of these seamless and intelligent companions are incredibly well-hidden in the moment that they are delivered. As eerily easy as AI feels at the surface, there are, and will likely continue to be, systemic impacts to labor, energy, and water consumption, extraction industries, politics, identity, and the concentration of power and wealth. To most, other than the people they impact directly, they will be invisible. How can we remain in a meaningful relationship with other unique humans, our communities, societies, and this beautiful ecosystem that we call home? How might we, from time to time, see the invisible AI infrastructure cost on the tarmac from the window seat of the AI plane? AI enthusiasts claim that these problems will benefit from investment in AI, which may be true, especially with a coalition of the willing, but right now, I am not seeing the incentives for anyone to do so.

Questions of the day

There are certainly more questions than answers here. In part, this summit carried the same purpose: to open our eyes a little. Growing up in the United Kingdom, our history classes at school taught us about The Luddites, a 19th century movement violently opposing automation in textile production. Putting their methods aside, I find it fascinating that even then, being a technophobe was not viewed kindly. I am no Luddite, and yet, like many in attendance at the summit last week, I meet the real and pervasive magnitude of these changes with both nervousness and curiosity. Many will dedicate their work towards these big questions. As someone oriented towards mindfulness and present-moment experience, I am compelled to stay focused on what’s here now and help people make small, wise decisions many times. This thinking, often attributed to meditation teacher Sharon Salzburg invites us to moments of awareness and wisdom as we choose how to interact and train our intelligent companions. This sentiment from her also feels appropriate in these times.

When we see the relatedness of ourselves to the universe, that we do not live as isolated entities, untouched by what is going on around us, not affecting what is going on around us, when we see through that, that we are interrelated, then we can see that to protect others is to protect ourselves, and to protect ourselves is to protect others.

In holding these questions on the blending of human and artificial intelligence, and how we can protect our shared humanity, mindfulness takes on its most expansive framing. Can we be mindful enough not to always choose comfort and convenience? Beneath the extraordinary seamlessness of these systems, are we willing to see the very real human and ecological worlds that make them possible? Are we willing to examine our lives more closely when the tools in front of us are so gleamingly transparent that we don’t see the pane of glass between us and the real world?

Let’s see.