Help for the Digitalist Papers From a Phenomenology of GenAI

The Digitalist Papers?

This week marked the emergence of the first 13 of the Digitalist Papers, a marvelous, timely project to bridge key disciplines when we face critical choices about the futures of democracy, artificial intelligence and therefore everything else. Co-led by two of the tech policy thinkers who most influenced me, Eric Brynjolfsson and Alex Pentland (collaborating with Condoleezza Rice and Law Professor Nathaniel Persily) the first 13 Digitalist Papers are authored by the broad range of business and policy leaders and thinkers who have been drawn into the critical debates about the future of AI in democracies and the future of civic life in the age of AI.?

Unlike the Federalist Papers (the basis here for GPTs critiquing MAGA, the current Supreme Court and the Federalist Society), the Digitalist Papers present an array of possible futures rather than creating and advocating a particular system of governance, so they provide a rich source for remixing, dialectics and innovation. Thus after devouring them eagerly, I created a wonderful GPT that did the same in its own way, Digitalist Ideator.

Lessig’s Sequestered Democracy

Fittingly, the first of the Digitalist Papers after the introduction is by Lawrence Lessig, who in recent years has been one of the keenest analysts of the challenges to American democracy. Remarkably, after reading all thirteen of the first Digitalist Papers, I found his essay the most incomplete on its own.

Lessig’s “Protected Democracy” proposes a radical restructuring of democratic processes through the creation of "protected assemblies"—groups of sequestered citizens who deliberate free from the influence of AI-driven disinformation and social media manipulation. Lessig depicts AI as a tool no more inherently good than a bulldozer. Left unchecked, Lessig rightly notes that such a neutral tech could worsen the polarization and erosion of trust already plaguing democracies, further manipulating public discourse and compromising elections. ?If Lessig is right to situate the current generative AI (GenAI) as just another bulldozer in the engagement economy—stipulating that such economy is problematic for public discourse, civic engagement and democracy—then his proposal to shield deliberative bodies from the distortions produced by these technologies makes sense as a potential general change in democratic processes.

?Lessig's bulldozer and his linking of social media to GenAI is unpersuasive to me for two reasons. The more trivial is that he ignores the incentives; of the four major AI foundation model companies, the two AI startups, OpenAI and Anthropic, are funded by subscription fees, not clicks on ads, and Sam Altman has expressed his opposition to ad-based funding. My early Jane Austen discussed the implications with Meta's Jane Austen. Of the two other foundation model companies, the deeper reason to follow will allow for differentiation even between Meta and Google.

Phenomenology of GenAI

My deeper reason for rejecting Lessig’s bulldozer is clear experiential differences between engaging as a “cyborg” with an LLM and doom-scrolling: Given my dual devotion to immersion in GenAI as much as feasible and to generating lots of long-form content for the greater KHive, I have been interspersing LLM work with way too much doom-scrolling in recent months, including on X.? For a while each day, I throw myself into the hurricane of symptoms and dopamine hits of social media engagement which the algorithms maximize: polarization, sensationalism, distraction, anxiety, depression, rage, elation, addiction. ?When I have had enough or too much, I beam up into a focused dialogue with GPTs that lead ultimately to the sort of co-created content the perfection of which is a goal of this newsletter. Phrases I learned as a teenager from Roland Barthes inform the latter experience: "the pleasure of the text" and "the death of the author."

The pleasure of the text seems appropriate because in writing with the AIs there is always dissonance between my perspective and that of the AI, at best surprises which are disruptive but relevant, which can on occasion lead me and the reader to jouissance—the disruptive pleasure that goes beyond simple satisfaction, leading to something unexpected and transformative. The dissonance creates a kind of dialectical relationship where the AI’s contributions provoke negotiation of meaning, encouraging reflection and recalibration of perspective, because they are relevant reactions not designed to get me to click on ads.?

The death of the author, the idea that the identity and intentions of the author should be stripped away, giving full interpretative control to the reader, has always seemed like a good cause to me, which I pursue both by trying to incorporate what I get from the AIs into my writing as seamlessly as possible, and by always offering my readers the AIs that I use to help them in further creation and exploration.? To Lessig, I argue that this use of GenAI fulfills the shift from "read-only" culture—where people consume static, finalized content—to a "read/write" culture, where users are empowered to modify, remix, and contribute to the creation of content themselves (an idea still worthy of an AI teacher like Lessig Remix Ideator).

A more contemporary description that Barthes’s which may be useful in appreciating and further shaping collaborative inquiry with GPTs is Mihaly Csikszentmihalyi’s concept of Flow, including:

1. Absorption in the Task: In a state of Flow, individuals are entirely absorbed in the task at hand, losing track of time and external distractions. Unlike doom-scrolling, where engagement is shallow and impulsive, working with GPTs creates a flow of ideas that sustains intellectual curiosity and productivity.
2. Clear Goals and Immediate Feedback: One of the core elements of Flow is having clear goals paired with immediate feedback. Dialogue with GPTs allows for constant feedback in real-time, as the AI responds to inquiries and suggests new directions for exploration. This creates a sense of progression and accomplishment.
3. Loss of Self-Consciousness and Distractions: Working with GPTs removes the distractions of external validation (likes, comments, shares) that dominate social media interactions. Instead, your attention is fully immersed in the creative and intellectual process, allowing for more authentic and productive work.
4. Intrinsic Motivation and Enjoyment: Flow is driven by intrinsic motivation—Unlike the extrinsic, attention-seeking engagement of social media, interaction with GenAI is rewarding because of the creative process, not because it seeks to maximize external engagement metrics.
5. Autonomy: Cyborg work enhances agency, focus and a rewarding experience of co-creation, as long as users do not just fall “asleep at the wheel.”

Now consider the difference between using GenAI on the two remaining foundation models, Google and Meta. With a Google search aided by Gemini, we are incentivized to click on ads, but it is in the context of a focused inquiry. We have greater agency, purpose and creativity than when doom-scrolling and responding.

Protect the GenAI, Not the Decisionmakers

My simple utilitarian response to Lessig's bulldozer is therefore that social media as it has evolved is less likely than a bulldozer to do good, and GenAI as I use it daily is more likely than a bulldozer to do good. More importantly, if we are focused on restoring productive civic participation or democracy, I find the cloistered sequestration of protected assemblies to be--while an understandable retrenchment in the face of firehoses of social media hatred and no doubt a valuable model in some circumstances--one of the most incomplete of all of Lessig's many remedies for the ills of contemporary America, mostly because by insulating the decisionmakers it also insulates everybody on the outside from the decisionmakers.? ??

If in GenAI we have a means of enhancing more focused, deeper collaborative inquiry as I suggest above, then why not use it as much as we can at the community level and beyond, using current crises and the greater threats of AI to focus more on making GenAI what our society needs than we did for the internet and social media? Indeed, several of the Digitalist Papers go precisely in that direction.

I particularly recommend "Rediscovering the Pleasures of Pluralism: The Potential of Digitally Mediated Civic Participation" by Lily L. Tsai and Alex Pentland, in part because it focuses on our big American issue of pluralism. They argue that digital platforms can be designed to reduce polarization by fostering engagement across different perspectives, allowing citizens to "watch, listen, and mull things over" rather than just express opinions. “Thinking slow” in Kahneman’s sense is valued as it should be. Protection, in this case of safe spaces against the emotional intensity of social media, is as important here as in Lessig’s protected assemblies.? I also loved their title, perhaps a reference to Stephen Jay Gould's 1997 article, especially in view of J.D. Vance's decree that pluralism always makes people unhappy.

Returning to Barthes for a second, disruptive pleasure (jouissance) may serve as a metaphor for the broader role AI might play in democracy—provoking civic engagement, challenging static viewpoints, and catalyzing innovation in the public discourse. Rather than seeing AI as a mere tool for efficiency or as a threat to coherence (as Lessig might suggest), it can be framed as an entity that facilitates a deeper, more complex engagement with content and ideas, precisely because of the friction it introduces. In this sense, AI democratizes not just access to knowledge but also the experience of engaging with that knowledge, turning every interaction into a potential source of creative tension.

Lessig begins "Protected Democracy" by simply positing: "AI will affect democracy. We should begin by understanding the democracy it will affect." A different starting point here yields opposite conclusions. Did Lessig devise protected assemblies with the social media polity in mind, and then shoehorn AI in to meet the moment of a sort of "keynote" for the Digitalist Papers? Are protected assemblies a hammer in search of nails? Social media is indeed such a nail; is GenAI?

?