Week 14: AI, Freedom, and the Kindly Slavemaster

DSAN 5450: Data Ethics and Policy
Spring 2026, Georgetown University

Jeff Jacobs

jj1088@georgetown.edu

Wednesday, April 22, 2026

A Pile of Simple Straightforward Topics Left to Cover

Today, In Class!

Discourse Ethics (Habermas)

Phenomenology (Sartre) \(\leadsto\) Phenomenology of gender (de Beauvoir) \(\leadsto\) Epistemological One-Way Mirrors (Du Bois)

Competing conceptions of Freedom (lowercase-r republican vs. lowercase-l liberal) \(\leadsto\) Kindly slavemasters are still slavemasters (Douglass) \(\leadsto\) Grapes of Wrath

Moved to Recommended Viewing:

Moved to Appendix Slides:

• Bertrand-Mullainathan-Pager discrimination experiments \(\leadsto\) Lily Hu critiques

• Fodor-Sperber Model: Culture as chains of internalized \(\leftrightarrow\) externalized representations

Discourse Ethics

A “General” Fairness Definition?

• May need to “descend” from 👆Platonic ideal fairness to 👇Aristotelian context-sensitive case-specific fairness 🤔
• (Hard enough to find “general” groupings in Canada, now do allocations of resources among them… and then the whole world)
• We saw this issue before, in different form! Rawls on “correct” ranking of rights
• [Rawls: No “correct” ordering; Different societies \(\leadsto\) different social value systems, power struggles \(\leadsto\) different orderings]

Raphael, The School of Athens (1511) rare 16th-century use of Georgia!

• [Me, I guess? 🙈: No “correct” fairness defn for racial discrimination; Different societies \(\leadsto\) different racial/caste/identity formations, power struggles \(\leadsto\) different fairness defns]

…But Wait! Here’s a German Guy!

• Speech Act Theory + Pragmatics \(\leadsto\) “Theory of Communicative Action” (Habermas 1990)
• Whether “globally” or (more realistically) in specific cases, what discursive settings might best enable “fusing of horizons”?

We must distinguish between the social fact that a norm is intersubjectively recognized and its worthiness to be recognized. [ought]

When we discuss moral-practical questions of the form “What ought I to do?” we presuppose that the answers need not be arbitrary; we trust our ability to distinguish in principle between right norms or commands and wrong ones. [Reflective Equilibrium]

[What structure undergirds this ability to distinguish?] What kind(s) of argument, what form(s) of reasoning is it proper for us to accept in support of moral decisions?

Habermas (1997) (Sole living person with a Cambridge Companion… until 1 month ago 😭)

Belief Formation: Coercive vs. Consensual

Beliefs derive from a complex mixture of rational insight and coercive force.

I distinguish between communicative and strategic action. Whereas in strategic action one actor seeks to influence the behavior of another by means of the threat of sanctions or the prospect of gratification in order to cause the interaction to [go in a direction that] the first actor desires, communicative acts [instead involve] communication oriented to reaching understanding [W02: fusing horizons!]

Only those norms can claim to be valid that meet (or could meet) with the approval of all affected in their capacity as participants in a discourse.

Like Kant, Rawls operationalizes 👀 “impartiality” in such a way that every individual can undertake to justify basic norms on his [sic] own [via “veil of ignorance” thought experiment]. The problems to be resolved in moral argumentation [however] cannot be handled monologically but require cooperation.

The categorical imperative needs to be reformulated as follows: “Rather than ascribing as valid to all others any maxim that I can will to be universal [HW1], I must submit my maxim to all others for purposes of discursively testing its claim to universality.

Phenomenology \(\leadsto\) Causal Pathways of Identity Formation

• Race as a Noun vs. Race as a Verb (“Racecraft”)
• Race as a static property vs. race as a social practice
• Subject-Object Distinction

But First… Phenomenology

“Objective” account: Roquentin sits down on bus seat; “Subjective” account:

(Image Source)

I lean my hand on the seat but pull it back hurriedly: it exists. This thing I’m sitting on, leaning my hand on, is called a seat. They made it purposely for people to sit on, they took leather, springs and cloth, they went to work with the idea of making a seat and when they finished, that was what they had made. They carried it here, into this car and the car is now rolling and jolting with its rattling windows, carrying this red thing in its bosom. I murmur: “It’s a seat” […] But the word stays on my lips: it refuses to go and put itself on the thing. It stays what it is, with its red plush, thousands of little red paws in the air, all still, little dead paws… (Sartre 1938)

W.E.B. Du Bois and the Epistemological One-Way Mirror

Black people in America are […] born with a veil […] in this American world—a world which yields him no true self-consciousness, but only lets him see himself through the revelation of the other world. It is a peculiar sensation, this double-consciousness, this sense of always looking at oneself through the eyes of others, of measuring one’s soul by the tape of a world that looks on in amused contempt and pity. One ever feels his two-ness—an American, a Negro; two souls, two thoughts, two unreconciled strivings; two warring ideals in one dark body. (Du Bois 1903)

• The veil: the world is seen and experienced differently on either side of the color line
• One-way mirror: Whites project their constructions of Blacks onto the veil and see their projections reflected on it \(\Rightarrow\) the power to define themselves and others
• The projections of whites onto the veil become realities (reification!) that Black subjects have to contend with in their self-formation.
• Twoness: in process of self-formation, the racialized subject must account for the views of two different social worlds—the Black world, constructed behind the veil, and the white world, which dehumanizes via lack of recognition of their humanity.

\(\textsf{Race}_{\textsf{Variable}}\) vs. \(\textsf{Race}_{\textsf{Construct}}\)

• Careful scientific, causal studies measure the effect that changing \(X\) (\(\text{do}(X)\)) has on \(Y\), controlling for \(C\) (via, at least under the hood, “\(\text{do}\)-Calculus”)
• But, even the most careful, controlled (and thus informative!) experiments must, at some level, partition variables into “race” and “not race”
• Keep in back of your mind as we look at example of how (measured by thorough, statistically-principled experiment), race can have direct, measurable, causal impacts on important aspects of our everyday lives

Racial Discrimination

• Marianne Bertrand and Sendhil Mullainathan. 2004. “Are Emily and Greg More Employable Than Lakisha and Jamal? A Field Experiment on Labor Market Discrimination.” American Economic Review. (Bertrand and Mullainathan 2004)

We study race in the labor market by sending fictitious resumes to help-wanted ads in Boston and Chicago newspapers. To manipulate perceived race, resumes are randomly assigned African-American- or White-sounding names. White names receive 50 percent more callbacks for interviews. Callbacks are also more responsive to resume quality for White names than for African-American ones. The racial gap is uniform across occupation, industry, and employer size. We also find little evidence that employers are inferring social class from the names. Differential treatment by race still appears to still be prominent in the U.S. labor market.

• So… Is [Solved-ness of problem] = [Closeness of racial gap to 0%]?
• Even on solely empirical, ahistorical basis (meaning, even without reparations for past harms in antecedents), there are reasons why 0% gap may not be the goal…
• To see why, we have to dig into race as a verb rather than a noun

“Cool Theory, I Guess…”

• Less pessimistic result of pessimistic conjecture: Some hope from Fodor-Sperber model (disclaimer: also terrifying Minority Report-style dystopian possibilities)
• “Good luck measuring ideas inside of people’s heads… I’ll be over here measuring real things and doing real data science!” -My innumerable Wile E. Coyote-style opps

Ordering of Topics is Important Here!

• Unfortunate-ness of white male teaching about race
• Unfortunate-ness of white male teaching about gender
• But, a counterpoint: Diversity / representation of marginalized voices \(\overset{?}{\longleftrightarrow}\) Labor of “speaking for” one’s identity group

The Standpoint Problem Revisited

• Problem statement: Jeff can’t possibly “teach” data-ethical issues, w.r.t. how they affect women, in the same manner he can teach e.g. how to take a derivative
• Solution 1: Have a woman teach a guest lecture \(\rightarrow\) (Possibility) Problem solved; (Possibility) Forcing additional labor onto women (see: 3 slides from now)
• Solution 2: Utilize the immense labor women have already put into trying to explain these issues to men with power, and amplify these already-existing products of this already-expended labor (next slide →)

Specifically-Chosen Examples

With Great Privilege Comes Great Responsibility

What is the most damage I can do, given my biography, abilities, and commitments, to the racial order and rule of capital? (Joel Olson)

The “Diversity in Tech”-Industrial Complex

• Problem: Not enough diversity in tech
• Solution 1: Intervene on the causal pathways leading to this outcome (incl. studying/tracing causal pathways)
  • Costs borne by tech companies; benefits accrue to marginalized ppl ❌🙅‍♂️⏹️
• Solution 2: Make marginalized ppl in tech jobs do tech jobs plus also extra job of explaining their marginalization to non-marginalized ppl (Third Shift?), who go home feeling good that they went to the diversity in tech panel (Brecht)
  • Costs borne by marginalized ppl; benefits accrue to tech companies ✅🎰🤑

(See Also)

Data Feminism (Epistemological One-Way Mirror 2.0)

Representation of the world, like the world itself, is the work of men; they describe it from their own point of view, which they confuse with the absolute truth.

(Beauvoir 1949)

“It Goes Without Saying”

Whiteness and maleness are implicit. They are unquestioned. They are the default. And this reality is inescapable for anyone whose identity does not go without saying […] For anyone who is used to jarring up against a world that has not been designed around them and their needs.

Belief in the objectivity, the rationality, the, as Catherine Mackinnon has it, “point-of-viewlessness” of the white, male perspective. Because this perspective is not articulated as white and male (because it doesn’t need to be), because it is the norm, it is presumed not to be subjective. (Perez 2019)

People = Male, Animal = Male

• “When I say ‘he’ I really mean ‘he or she’, obviously”
• Except… irrespective of what you really mean, or whether it’s ‘obvious’, it goes out into the world and has effects (reification!),
• From childhood (Vervecken et al. 2013)
• To job-hunting (Bem and Bem 1973)
• And beyond (Sczesny et al. 2016)
• A stuffed animal must be “super-feminine” before “even close to half of participants will refer to it as she rather than he”. (Lambdin et al. 2003)

The Cowan Paradox

For many ages to come the old Adam will be so strong in us that everybody will need to do some work if he [sic] is to be contented […] But beyond this, we shall endeavour to spread the bread thin on the butter—to make what work there is still to be done to be as widely shared as possible. Three-hour shifts or a fifteen-hour week may put off the problem for a great while. For three hours a day is quite enough to satisfy the old Adam in most of us!

(John Maynard Keynes, “Economic Possibilities for our Grandchildren”, 1930)

Cowan (1983)

A Kindly Slavemaster is Still a Slavemaster

Relevance for This Week (Where We Left Off)

• Can we develop policy interventions that equalize power, so that world looks like normative ethics from W03-W08 (“what is right”)?
• (Hidden antecedent: non-Nietzschean ethical framework e.g. Utilitarianism/Kant)
• Point of prev slide: From now til W14, keep in mind how definition of power (and hence “effectiveness” of policy intervention) depends on antecedent
  • liberal definition \(\Rightarrow\) focus on equilibria (no injustice if bad thing doesn’t happen in equilibrium)
  • republican definition \(\Rightarrow\) also take off-equilibrium possibilities into account (no injustice if bad thing doesn’t happen in equilibrium and doesn’t happen if one player deviates “on a whim”) (Jacobs and Naidu 2026 😉)
• Prisoners’ Dilemma 😫 \(\prec\) Assurance Game 🤨 \(\prec\) Invisible Hand Game 🥳

Thucydides and the Kindly Slavemaster

[What is] right, as the world goes, is only in question between equals in power; otherwise, the strong do as they please and the weak suffer what they must. (Thucydides 2013, chap. 411 BC) (Think of necessary vs. sufficient conditions!)

	liberalism	republicanism
Definition of Injustice	Strong do bad things (Berlin 1959)	Strong can do bad things (Skinner 1998; Pettit 1997; Lovett 2022)
Thucydides Question	Strong do as they please \(\overset{?}{\Rightarrow}\) Strong do bad things	Strong do as they please \(\overset{?}{\Rightarrow}\) Strong can do bad things
Answer	No, not necessarily!	Yes, necessarily!
Frederick Douglass	My feelings [towards slave masters] were not the result of any marked cruelty in the treatment I received…	…they sprung from the consideration of my being a slave in the first place. It was slavery—not its mere incidents—that I despised. (Douglass 1855)
A Doll’s House	Our home is nothing but a playroom. I have been your doll-wife, just as at home I was papa’s doll-child; and here the children have been my dolls. (Ibsen 1879)

(Plz notice the lowercase “l”, lowercase “r”!)

References

Beauvoir, Simone de. 1949. The Second Sex. Knopf Doubleday Publishing Group. https://books.google.com?id=OgMbKqJMzxcC.

Bem, Sandra L, and Daryl J. Bem. 1973. “Does Sex-biased Job Advertising ‘Aid and Abet’ Sex Discrimination?1.” Journal of Applied Social Psychology 3 (1): 6–18. https://doi.org/10.1111/j.1559-1816.1973.tb01290.x.

Berlin, Isaiah. 1959. Two Concepts of Liberty: An Inaugural Lecture Delivered Before the University of Oxford on 31 October 1958. Clarendon Press. https://books.google.com?id=cLaGAAAAMAAJ.

Bertrand, Marianne, and Sendhil Mullainathan. 2004. “Are Emily and Greg More Employable Than Lakisha and Jamal? A Field Experiment on Labor Market Discrimination.” American Economic Review 94 (4): 991–1013. https://doi.org/10.1257/0002828042002561.

Cowan, Ruth Schwartz. 1983. More Work For Mother. Basic Books. https://books.google.com?id=x8HuAAAAMAAJ.

Douglass, Frederick. 1855. My Bondage and My Freedom. Yale University Press. https://books.google.com?id=G8t0AgAAQBAJ.

Du Bois, W. E. B. 1903. The Souls of Black Folk: Essays and Sketches. A. C. McClurg. https://books.google.com?id=lTXYAAAAMAAJ.

Habermas, Jurgen. 1997. The Cambridge Companion to Habermas. Repr. Edited by Stephen K. White. Cambridge Companions. Cambridge Univ. Press.

Habermas, Jürgen. 1990. Moral Consciousness and Communicative Action. MIT Press. https://books.google.com?id=fmYjgiUMy7EC.

Ibsen, Henrik. 1879. The Doll’s House: A Play. D. Appleton & Company. https://books.google.com?id=afgNAAAAYAAJ.

Jacobs, Jeff, and Suresh Naidu. 2026. “Operationalizing Freedom as Non-Domination in the Labor Market.” In Perspectives on Labor Republicanism, edited by Philip Pettit. Cambridge University Press.

Lambdin, Jennifer R., Kristen M. Greer, Kari Selby Jibotian, Kelly Rice Wood, and Mykol C. Hamilton. 2003. “The Animal = Male Hypothesis: Children’s and Adults’ Beliefs About the Sex of Non–Sex-Specific Stuffed Animals.” Sex Roles 48 (11): 471–82. https://doi.org/10.1023/A:1023567010708.

Lovett, Frank. 2022. The Well-Ordered Republic. Oxford University Press. https://books.google.com?id=CdZ2EAAAQBAJ.

Perez, Caroline Criado. 2019. Invisible Women: Data Bias in a World Designed for Men. Abrams. https://books.google.com?id=GdmEDwAAQBAJ.

Pettit, Philip. 1997. Republicanism: A Theory of Freedom and Government. Oxford University Press.

Sartre, Jean-Paul. 1938. Nausea. New Directions Publishing. https://books.google.com?id=CGbBbtDOZbwC.

Sczesny, Sabine, Magda Formanowicz, and Franziska Moser. 2016. “Can Gender-Fair Language Reduce Gender Stereotyping and Discrimination?” Frontiers in Psychology 7 (February). https://doi.org/10.3389/fpsyg.2016.00025.

Skinner, Quentin. 1998. Liberty Before Liberalism. Cambridge University Press. https://www.dropbox.com/scl/fi/1oqaoig6ev602602384x1/LibertyBeforeLiberalism_Zoomed.pdf?rlkey=r8ntmmir8a1oxsnttgk5l6fq5&dl=1.

Thucydides. 2013. The War of the Peloponnesians and the Athenians. Cambridge University Press. https://books.google.com?id=dSYnCTf3rSAC.

Vervecken, Dries, Bettina Hannover, and Ilka Wolter. 2013. “Changing (S)expectations: How Gender Fair Job Descriptions Impact Children’s Perceptions and Interest Regarding Traditionally Male Occupations.” Journal of Vocational Behavior 82 (3): 208–20. https://doi.org/10.1016/j.jvb.2013.01.008.

Appendix 1: Experimental Evidence of Discrimination

“Controlling for” Everything Besides Race

• Economist assertion: everything is “same” except for [name \(\leadsto\) race]
• Weird part of assertion: only true if the “everything” is stripped of context… But, stripped of context, how would we get [name \(\leadsto\) race] in the first place?

Age Discrimination?

Figure 1: Based on Lily Hu, What is ‘Race’ in Algorithmic Discrimination on the Basis of Race? - UCLA IPAM (YouTube)

Fair \(\iff\) [\(\Pr(\text{Admit Presley}_{12}) = \Pr(\text{Admit Presley}_{22})\)]?

• Root of issue: [BA Stats, UCLA, 3.7] has no “free-floating” meaning—it’s attached to a person \(\Rightarrow\) affected by/interpreted w.r.t. their “protected” characteristics

Appendix 2: Fodor-Sperber Model

“Cool Theory, I Guess…”

(Brace yourself: Jeff’s Trying-My-Best Fodor-Sperber model of socially-constructed “race” on next few slides… I’m sorry in advance 🙈🙈🙈 Did you know you can italicize emojis)

Opening A Big Can Of Worms

• Social interactions among \(t^e_0\), \(t^e_1\), \(t^e_2\)…

Opening A Big Can Of Worms

• Social interactions among \(t^e_0\), \(t^e_1\), \(t^e_2\)…
• Mediated by external things \(o^e_3\) to \(o^e_8\) (giving rise to patterns of interaction)…

Opening A Big Can Of Worms

• Social interactions among \(t^e_0\), \(t^e_1\), \(t^e_2\)…
• Mediated by external things \(o^e_3\) to \(o^e_8\) (giving rise to patterns of interaction)…
• Each person \(x\) forming their own internal representations \(\widetilde{t^x_0}\), \(\widetilde{t^x_1}\), \(\widetilde{t^x_2}\) of one another based on patterns of interaction, then
• Generalizing to an internal representation of a “type of person” \(\widetilde{t^x_9}\)…

Opening A Big Can Of Worms

• Social interactions among \(t^e_0\), \(t^e_1\), \(t^e_2\)…
• Mediated by external things \(o^e_3\) to \(o^e_8\) (giving rise to patterns of interaction)…
• Each person \(x\) forming their own internal representations \(\widetilde{t^x_0}\), \(\widetilde{t^x_1}\), \(\widetilde{t^x_2}\) of one another based on patterns of interaction, then
• Generalizing to an internal representation of a “type of person” \(\widetilde{t^x_9}\)…
• Which they then externalize as \(t^x_9\).
• \(t^0_9\), \(t^1_9\), \(t^2_9\) “congeal” into a shared external representation \(t_9^e\) via social mechanism (discussion, media, culture, propaganda, parenting, religion, education, …) \(\Rightarrow t^e_9\) “reified” (causal effects on \(t_0\), \(t_1\), \(t_2\))