Grappling with AI is an Adaptive Challenge
To tackle it, we must grow
Thank you for being here. Please consider supporting my work by sharing my writing with a friend or taking out a paid subscription.
I’ve been thinking recently about the Anthropic “keep thinking” ad campaign that was released in the fall of last year. It’s a good ad. I liked it then, I like it now. It’s got 1980’s Apple vibes and the kind of feel-good aesthetic that gets an engineer like me amped up.
There is real truth to the message of the ad. There has never been a better time to have a problem. I’ve seen it firsthand in my own work. I’ve been using AI to build out software and applications that I could only have dreamed of building before these tools. A Hermes AI agent is helping me handle logistics for one of my side hustles. Claude is helping me fix stuff around the house. A short back and forth about my broken dryer helped me to diagnose the problem, source the replacement part, and get a link to a relevant YouTube video with instructions on how to replace it, in about five minutes. I don’t even need to type the model number into Google anymore, I just snap a photo of the sticker and off we go. The replacement part was at my door the next day. Dryer fixed.
Is it a technical problem or an adaptive challenge?
But while AI can help to solve some problems, it cannot solve all problems. In fact, not only can AI not solve some problems, but it can actively make some of them worse. The real question is about the kind of problem that needs to be solved.
How might we distinguish between these kinds of problems? My friend and Praxis colleague Andy Crouch introduced me to one frame that I am finding helpful. The inspiration is from Ronald A. Heifetz and Donald L. Laurie’s distinction between technical problems and adaptive challenges. The table below provides a flavor of the difference between them.
As you can see from the table, the basic distinction that Heifetz and Laurie make between technical problems and adaptive challenges is less about the problems themselves than the kind of intervention needed to address them. Technical problems lend themselves to technical solutions. Even if they are very complex, the information needed to solve them is known, and once compiled and organized, can be readily applied to address the problem.
OpenAI’s recent announcement about the resolution to an Erdös problem, reported last week, is an example of a technical problem. While it remained unsolved for some time, the breakthrough came from pointing an AI model at a particular part of the solution space had not been fully explored. What prevented the problem from being solved was less about a new creative innovation, and more directly connected to the fact that no one had thought it worth the time and effort to explore the space with sufficient tenacity to discover the solution that the model was able to find. This writeup from Gary Marcus and Cal Newport has some more helpful discussion.
Adaptive challenges, on the other hand, are much gnarlier. Not only are they harder to identify, but they are, in a sense, fundamentally human challenges. They require the persons facing them first to be able to identify and acknowledge the challenge for what it is and then to grow in order to address it effectively.
Unfortunately, it is quite easy to miscast an adaptive challenge as a technical problem. In many ways it is the story of our current moment in history. So many of the challenges we face in society in the US and across the globe, whether they are about living at peace with our neighbors, caring for the forgotten and left behind, or trying to address the loneliness epidemic are adaptive challenges to which we dream of technical solutions. They will leave us wanting.
Consider the last decade. It is hard to argue that there has been a more significant cultural force in the last ten years than social media. Taken at its best, the goal of social media is to use digital technology to bring us together. Consider Facebook’s mission statement, “to give people the power to build community and bring the world closer together.” What a great vision. But it is hard to imagine a more adaptive challenge than building community. Building authentic community is perhaps one of the most adaptive challenges there is, requiring growth from all members of the community. Many of the forces driving AI right now are poised to repeat this mistake with even greater fervor.
Adaptive challenges and AI
As we think about the societal impact of AI, we’ll need to be clear eyed about whether the problems we are trying to solve with it are technical problems or adaptive challenges. To be clear, there are plenty of both. AI tools can solve many technical problems connected to understanding the connections within large amounts of data. AI can help us to be able to more flexibly interact with information.
And yet, the bigger promise embedded within the most optimistic vision of AI is that it will help us to tackle the adaptive challenges facing us. Consider just some of the consumer-facing applications of AI that are self-evidently adaptive challenges. Education. Mental wellness. Therapy. Relationships. Care for the elderly. Each of these areas is a domain of adaptive challenges, where the only way to make progress is for humans to grow.
Consider education, the space where I am most qualified to speak with some authority. There are ways to apply AI to help solve the technical problems that exist in the educational environment. For example, using AI to help analyze gaps in a student’s understanding of certain topics and to help curate a set of problems to help fill those gaps is a technical problem that AI is quite well equipped to solve. However, the much bigger challenges in education, like how to motivate and form young people as they build coherent visions of what makes their lives meaningful and how they can contribute to the good of the world, these deeper problems at the heart of education cannot be solved by AI. These are the kind of adaptive challenges that require that the student grow and continue to become someone rather than master something.
The question we are facing is whether we are willing as a society to face the challenge ahead of us head on with a willingness to grow. Adaptive challenges are not insurmountable. They are the kind of things that can be addressed if we are up to the task. But being up to the task will require a willingness to stretch ourselves and grow beyond the people we currently are.
The biggest challenges we face as a society are fundamentally human problems. They are about who we are becoming, not what we can do. AI can play a role in helping to solve the technical challenges that are wrapped up in these larger scale adaptive challenges. But at the end of the day, these challenges will require something from us that AI cannot do for us.
Got a thought? Leave a comment below.
Reading Recommendations
The big news of the week is from the Vatican around the release of Pope Leo’s first encyclical around AI. There will no shortage of takes and reactions. Here are a few that I’ve read so far:
Yuval Levin’s excellent essay in The New Atlantis, Idols of the Valley.
The appeal of idols has always been that they offer shortcuts. The God of the Bible demands that you live in a way that forms your mind and heart and soul toward your fullest human potential. This requires hard work but it yields a kind of person both capable and worthy of a flourishing life. The idol offers the material benefits of such a life without that formative work. And if all you care about are the benefits, not the form of your mind, heart, and soul, then the offer is awfully hard to resist.
This plainly rhymes with some of the deepest moral challenges posed to us by artificial intelligence. AI, at least used a certain way, offers us shortcuts around formative work, matching outputs with inputs without the need for the interceding effort of mind, heart, and soul. If all you care about are the outputs, not the form of your mind, heart, and soul, then the offer is awfully hard to resist.
The Book Nook
I’ve been spending some time this week in Arthur Holmes’ classic, The Idea of a Christian College.
The Professor Is In
Last week was a busy one. I was at the Praxis Summit Monday through Wednesday, and then drove home on Wednesday for the AI conference that I have been helping to co-organize at Harvey Mudd. Summit was amazing as always and the AI conference at Mudd was wonderful. It was especially refreshing to meet a variety of new folks from across the Claremont Colleges who are thinking about what AI means for their disciplines.
Leisure Line
Met this little guy on a hike last weekend in La Crescenta. Kids had a ball building him a little home in the stream.
Still Life
The peacocks are a staple attraction at the LA Arboretum. We were treated to a full display right when we walked in!
Excellent post. This seems to bring us to the whole idea of Human in the Loop, and how AI could help us do better, what we do best. A researcher, becomes at what he does the best. When it comes to education and educator, providing individualized education and not leaving anyone behind, can be a good example.