- Eric Ries
- Posts
- Can AI resist corruption?
For the Better comes to you with ideas about how and why to build companies focused on human flourishing and stories of the people who are doing it. Other enthusiasms may occasionally appear.
Last month, Anthropic's CEO Dario Amodei published a long essay called "The Adolescence of Technology". It's a follow-up to "Machines of Loving Grace", his 2024 post about all the ways in which AI has the potential to "transform the world for the better."
Writing now, Amodei still "believe[s] deeply in our ability to prevail, in humanity’s spirit and its nobility." But, he continues, "we must face the situation squarely and without illusions."
The situation, of course, is how to rein in and direct the incredibly powerful AI that Anthropic and others have built in order to realize its great benefits rather than falling prey to its potential dangers. And our current tools are far from fit for purpose. As Amodei writes:
"The sheer amount of capability embodied in powerful AI is such that ordinary corporate governance—which is designed to protect shareholders and prevent ordinary abuses such as fraud—is unlikely to be up to the task of governing AI companies. There may also be value in companies publicly committing to (perhaps even as part of corporate governance) not take certain actions, such as privately building or stockpiling military hardware, using large amounts of computing resources by single individuals in unaccountable ways, or using their AI products as propaganda to manipulate public opinion in their favor.”
Anthropic's response to these realities is the Claude Constitution. It is part training tool, part mission statement, embodying "Anthropic’s intentions for Claude’s values and behavior." Zvi Mowshowitz calls the Constitution, "the official version of what we were previously calling its 'soul document.'"
Essentially, it's applied philosophy in the realm of AI. It is also a new kind of governing document -- one that seeks to guide while remaining flexible in the face of ongoing change. It "provides structure and support while leaving room for organic growth. It’s meant to be a living framework, responsive to new understanding and capable of evolving over time."
If we want powerful technology to be aligned with human values, we must first ask if the organizations that create it are so aligned. The Claude Constitution shows how an organization can make that commitment public. (Anthropic is also making it available for anyone to use as a model.) It reflects Anthropic’s company structure, where an entity called a Long Term Benefit Trust protects its mission (and the values expressed in the Constitution) in perpetuity, regardless of leadership change or other factors.
The LTBT is a form of what I call constitutional governance. It’s the kind of structure we will all need to rely on as we move through this new era of technology, discovering how to be the citizens of a new world.
Join my live conversation with Reid Hoffman
Corruption in modern organizations isn’t primarily caused by bad people or moral collapse. It’s the predictable result of organizations that were never designed to withstand success, scale, or financial pressure.
Join Reid Hoffman and me for a candid conversation about which companies are failing. We’ll discuss the surprising tactics of companies who are withstanding the pressure and upholding their original missions.
This Wednesday, February 4
11am ET
I’m also making a special announcement. Can’t wait to share more about what I’ve been working on.
Hope you can join us,
Eric