My dissertation tackles just this topic by focusing on democratic decline in the Information Age. In it, I offer some ways to try and reverse democratic decline by using contemporary technology to change motivational processes in ways that facilitate democratic attitudes. I have placed asterisks (***) next to papers that will be a part of the dissertation.

I am happy to share drafts of works in progress. Feel free to email me at lmanuali [at] umich [dot] edu for a draft.

Published and Forthcoming

“Consorting with Bad Actors”, forthcoming in Business Ethics Quarterly

Why is it wrong to do business with unethical actors or accept donations from bad donors? In this paper, I reject five common explanations—reputation burnishing, harms from public knowledge, gains in self-repute, retributive punishment, and institutional mission—because they fail to provide extensionally adequate accounts of this wrongness. Instead, I argue that taking money from bad donors necessarily involves consorting—undertaking joint action with an unethical entity—which provides a unitary and necessary explanation for why such actions evoke ethical discomfort. I show that we care about consorting ethically because it compromises integrity and moral standing, and I conclude by applying these insights to commercial firms by posing the "Challenge of Firm Legitimacy".

“Blind Refusal: Language Models Refuse to Help Users Evade Unjust, Absurd, and Illegitimate Rules” (w/ Cameron Pattison (first author) and Seth Lazar), forthcoming at the Conference on Language Models (COLM), preprint on arXiv

When should AI models help users break rules? In this paper, we argue that current models exhibit what they call blind refusal — declining to help users evade unjust, illegitimate, or absurd rules even when the models themselves recognize the injustice. Testing 18 AI systems across 1,290 scenarios, we find models refuse around three quarters of the time, and call for alignment approaches that can actually distinguish rules worth enforcing from rules that aren't.

“Using LLMs to Enhance Democracy” (w/ Seth Lazar), Minds and Machines, 2026, accepted at FAccT 2025 (non-archival)

Examines (and categorizes) the various ways computer scientists are attempting to use LLMs in proposals for democratic processes and then assesses said uses in light of a few key democratic values.

“Addictive Motivational Scaffolds and the Structure of Social Media”, Synthese, 2025***

In the tradition of 4E cognition, I offer a psychiatric externalist account of behavioral addiction by developing the notion of a motivational scaffold (as opposed to cognitive or affective scaffolding). I then apply said account to elucidate why social media is addictive.

“The Coherent Dual Theory of Addictive Desire”, Philosophical Psychology, 2024

Offers an account of the dysfunction present in desire in addiction by attempting to attain coherence between empirical (esp. neuroscientific) accounts of addiction and phenomenological accounts of addiction.

Preprints

“AI and Political Theory” (w/ Linda Eggert, Jeffrey Howard, Ting-an Lin, and Rob Reich – equal contributions), preprint on APSA Preprints

A contribution to a larger edited volume on AI and Political Science, this chapter focuses on the intersection of AI and political theory. The chapter identifies significant contributions to date as well as areas for future research, concerning (1) theories of justice, (2) democracy, and (3) rights. It addresses issues of algorithmic injustice, the digital public sphere, digital-age human rights, as well as the contested possibility of AI rights. Along the way, the chapter assesses what existing theories imply for new policy questions in a world with AI and what basic theoretical commitments or assumptions may need to be rethought.

Other Papers In Progress

“Democratic Motivation and Mass Media”***, under review

Building on Susanna Siegel’s groundbreaking work concerning salience principles (roughly, norms that prescribe what is worthy of attention), I offer an account of how to create a journalistic mass media governed by salience principles that better facilitate democratic attitudes.

“When LLMs Threaten Democratic Autonomy”, under review

Computer scientists are testing various ways to use LLMs in democratic processes in order to support or enhance them. Pulling on Lovett and Zuehl (2022), I show how Large Language Models that are (1) used to facilitate deliberative processes and (2) used to represent us by extrapolating over unseen (policy) options threaten democratic autonomy.

“Philanthropic Power Requires Democratic Transparency”

With the rise of philanthropic organizations stemming from wealth made in the Information Revolution in the background, this paper makes the case that large philanthropic institutions and individuals (as well as some small ones) are subject to a democratic duty of transparency.

“The Negative Epistemic Argument for Democracy”

I present a new instrumentalist, epistemic argument for democracy over other feasible forms of government. I argue that democracies, unlike these other forms of government, possesses the epistemic virtue of minimal responsiveness, which allows it to avoid catastrophic outcomes that other forms of government fail to avoid. The main theoretical upshots of this argument are that it is resistant to several of the main critiques leveled at other epistemic arguments for democracy, and thus it provides us a basic reason to prefer democracy to other forms of government.

“Liberalism’s Negative: The Dualism of Political Evil”

I offer a new argument for a thin form of liberalism: that it is a bulwark against dystopia. Intervening in a dialectic begun by Judith Shklar and continued by David Enoch — who ask not how a government attains the best possible outcome (the summum bonum) but how it avoids the worst (the summum malum) — I object that Enoch's proposal, widespread and serious suffering, undergenerates: it cannot explain why dystopias like that of Brave New World, which contain little such suffering, count among the very worst outcomes a political system can produce. The missing ground, I argue, is a widespread, serious lack of autonomous agency. I thus defend a pluralist account on which the summum malum is grounded in both (1) widespread, serious suffering and (2) a widespread, serious lack of autonomous agency, and then argue — on both conceptual and empirical grounds — that liberalism avoids the latter. The main theoretical upshot is a new (if pro tanto) justification for liberalism.