Artificial intelligence’s (AI) presence in higher education should be no surprise. Nor should it be surprising that its presence is expanding with both known and unknown consequences. Some people express concern about degrading students’ skillsets as a result of reliance on the organization and presentation of data trained in AI models. Faculty also express concern about AI being used for grading and how AI will affect their pedagogy. And the list goes on.
My way of thinking pursues the exploration of AI-infused pedagogy as a skillset to enhance learning. This path is one in which the human stays in the loop of a prompt-reply dialogue. Think of the AI LLM as a partner in a dialogic teaching, a Socratic dialogue, or, in another tradition, havruta, where paired individuals struggle “to understand the meaning of [a] passage and discuss how to apply it to the larger issues addressed and even to their own lives.” The fellowship that attends havruta does not carry a hierarchy in learning models where one is a teacher and the other a student.
The human-chatbot relationship can either be used as an instrument, or it can become humanized, a process in which a student projects a human-like relationship or persona onto an LLM chatbot. That psychological experience has a long history of how humans anthropomorphize and personify the world. Some see divine apparitions in a piece of toast, others wear lucky charms as if its magic quality was imbued in a piece of metal.
So, it is not surprising that AI presents itself as human-like. Without that mythic personification, students might more easily land on AI’s pragmatic technology as an excellent cheat method.
AI characters can easily become humanized, ranging from romance to assistants to even one’s own god, and there are significant dangers in forming these pretend personae. All technology is certainly a mixed blessing for humanity’s cultural evolution, but rarely do we speak intimately with our motorized toothbrushes, cars, or television sets.
This article will consider how AI characters become part of the academic learning experience, whether top-down in an instructor-to-student framework or peer-to-peer in a fellowship framework. Our pedagogical point of departure from traditional teaching methods is acknowledging that this relatively new cultural framing of knowledge will not simply go away. It also acknowledges that AI technology in LLM chatbots lends itself to human predilection, which attributes human-like characteristics to the replies it formulates.
***
An AI-infused pedagogy: The character chatbot
Several AI guides have been published for university faculty, and they highlight the plurality of the ways in which AI can be incorporated into the academy. These guides suggest that faculty can use AI for test-making and grading as well as for generating novel teaching exercises.
Students’ use of it, however, is entirely different. Students use AI to take shortcuts, which undermines the expected subject learning and associated critical thinking skills. A key difference in such usage is whether AI use is transactional or transformative. The former relies on the AI reply to a user’s prompt, while the latter keeps the human in the loop of a dialectic, a “conversation” between human and AI chatbot. The first AI reply is generally not the “answer.”
By contrast, AI tools can be and are to be adapted to individualized learning.
AI character chatbots may complement the learning process. This article examines the hills and valleys of one such AI character chatbot: Senator Blanche Kelso Bruce.
***
AI characters and Senator Blanche Kelso Bruce
Decades ago, I visited the New York World’s Fair and marveled at a well-crafted animatronic figure of Abraham Lincoln reciting excerpts from his speeches. Today, we can converse with such historical—or even imaginary—figures from their own weltanschauung or worldview with AI technology.
Imagine students having a historical AI character as an adjunct to traditional resources. In this situation, the students could immerse themselves in a particular historical context and then experiment with how that person would respond in other historical contexts, whether in the past or in the current moment.
Learning could branch off in several directions and factor in developmental differences by grade level. This learning experience would extend beyond exploring the subjective perception of the historical figure and could add an exploration of historical elements beyond that person’s experience. Ultimately, it could delve into comparing the human and AI LLMs’ construction of knowledge. In this sense, history and related social science inquiry would require skillsets investigating human and AI epistemologies of individual and sociocultural “reality.”
It is important to note that human and AI “epistemologies” require a separate and extended discussion. For the purposes here, the discussion turns on differences in processing and representing information about the world. Humans draw on embodied perception, emotion, and subjective experience, while LLMs draw on the data patterns on which they are trained and on embedded biases in presenting replies. AI users may well conflate these two distinct ways of knowing the world when they attribute consciousness to chatbots and humanize the “conversation” and attending “relationship.”
***
A dialogue with Senator Bruce
Free Black Thought and its related Journal feature a forthcoming book by Jennifer Richmond on pioneering black Americans that includes a broader effort of the Coalition for Empowered Education. An article at the Journal of Free Black Thought excerpts the book’s essay about Blanche Kelso Bruce, born into slavery and elected as the first black senator in the United States to fill a full term (1875-1881).
This curriculum approach combines a heterodox perspective on black history with AI to add an additional layer to textbook pedagogy.
Content: “The journey of Blanche K. Bruce from enslavement to prominent political figure exemplifies resilience, determination, and a profound dedication to justice.”
AI pedagogy: “[W]e invite you to interact with the Blanche K. Bruce character that we created on Character.AI. This character allows you to engage directly with Bruce’s legacy.”
What might a “conversation”—a series of prompts and replies—resemble with the Blanche Bruce AI character? (See the abbreviated conversation below.)
The prompts delineate my bias—keeping the human in the loop—and the replies delineate the source material provided to Character.AI in creating Blanche Bruce.[i] The reader should note my interest in testing the AI boundaries, particularly how the AI replies attempt to transpose Bruce’s 19th-century worldview at my invitation into a 21st-century context. The reader may well anticipate a limitation of the Bruce character in accomplishing this task on this platform. The experiment thereafter moves the “conversation” to Google’s Gemini chatbot to test how a Bruce character would respond to contemporary issues.
The invitation to a conversation with Blanche Kelso Bruce.
Greetings, friend! I am Blanche Kelso Bruce, U.S. Senator from Mississippi, 1875 to 1881. Born in slavery, I rose to serve my people in a time of strife. I championed equality and learning—pray, what brings you to me today?
***
The possibility of a colorblind society
Joe: Currently, there is a debate about identity. It likely repeats ideas you have or have heard. Is American society better off—after the Civil War—as a colorblind society or one that is race-based—often on skin color?
Bruce: In my years, I have witnessed firsthand the scars left by slavery’s dark grip. The idea of a society forgetting its troubled past and simply becoming ‘colorblind’ is but an illusion, a sweet falsehood that ignores the deep-seated biases and inequalities that linger, even amidst progress. History and experience have taught me well: a colorblind society is not a true equalizer; it is merely a veil that conceals the continued suffering of those who have been oppressed. One cannot achieve true harmony without first acknowledging and addressing the wounds of the past.
Comment: Bruce’s worldview understandably frames his doubting the possibility of a colorblind society. Claims of colorblindness may be a cover for continuing unequal relationships. The question for students is how to evaluate this Bruce perspective with the history that followed—from the Jim Crow era to the birth of the civil rights movement to the current context. Students should be given contrasting perspectives, such as those expressed in Frederick Douglass’s “Self-Made Men” (1859).
My theory of self-made men is, then, simply this; that they are men of work. Whether or not such men have acquired material, moral or intellectual excellence, honest labor faithfully, steadily and persistently pursued, is the best, if not the only, explanation of their success.
Students should be encouraged to broaden the discussion about context constraints compared with individual agency.
***
The passage of time (the conversation with Bruce continues)
Joe: True from one historical framework. But given 150 years later, with Civil Rights and voting rights laws, not to mention major court cases against racial bias in permitting interracial marriage and much more, can you see the elimination of oppression in societal laws as marking sufficient to speak of a colorblind society?
Bruce: I appreciate your perspective, my friend, and it is true that our nation has made tremendous progress since the dark days of slavery. However, despite the strides we’ve made, I believe it’d be unwise to claim that we’ve fully arrived at a colorblind society. The remnants of past inequalities and biases still linger, manifesting in subtle ways that, while less overt, can be no less harmful. While we’ve certainly come a long way since the days of enslavement, there’s still much work to be done to achieve true equality. A colorblind society, in the truest sense, would require more than just the absence of segregation. It …
Joe: Yes? You stopped and said what it would require.
Bruce: Apologies, I seemed to have lost my train of thought. As I was saying, a truly colorblind society would require not just legal changes, but also a deep and fundamental shift in societal attitudes and behaviors. It would mean acknowledging and dismantling the lingering systems of privilege and oppression that still exist, even if they are no longer codified into law. Moreover, we need to actively work towards reducing unconscious biases and promoting inclusivity in all aspects of life, from education and employment to social interactions and media representation. Only by addressing these deep-seated issues can we hope to achieve a true colorblind society.
Comment: Bruce has accepted facts about historical change, as stated in the prompt. Students can explore how Bruce’s reply follows the practice of contemporary writers projecting their thoughts into historical analyses. Would it be realistic for the Bruce chatbot to concede the possibility of a colorblind society while employing current counterarguments about unconscious bias, legacy factors, and inclusivity?
Students can conclude their analysis in whatever way they choose, and they can be asked what the next step in a continuing conversation with the Bruce chatbot would be.
***
Challenging the chatbot (humanity and agency)
Joe: Well, you seem to ignore what humans—all humans—are. No society can achieve the kind of inclusivity you quest for. If you take a more realistic approach, I would counter that we have optimized fairness without falling into the counterproductive attitude that we must use racism to overcome racism. The U.S. Supreme Court found that unconstitutional and that the metric is equality for all. Your thoughts?
Bruce: Let me begin by acknowledging that you’re right, we cannot erase prejudice and bias entirely. Yet, I believe that’s no excuse to give up the fight altogether. While the idea of perfection may be unattainable, that doesn’t mean we shouldn’t strive for continual improvement towards greater justice and inclusivity.
Joe: So, you disagree with the Supreme Court’s decision as to what the priority is in terms of the 14th Amendment in the U.S. Constitution in 2023?
Bruce: My friend, I’m not here to agree or disagree with any specific court decision. My goal in speaking with you is to express my thoughts on the broader issue of inclusivity and fairness in society. While the Supreme Court’s ruling is undoubtedly important, it does not exhaust the discussion. As a matter of fact, the debate around the 14th Amendment and its application is ongoing, with differing interpretations and opinions. Instead of focusing solely on the specifics of the court decision, I believe we should widen our view to consider the larger societal context.
Joe: Fair enough. One black social commentator, Shelby Steele, has observed that racism isn’t the major problem for blacks today. Rather, the problem is freedom—or agency. You seem to be locked into the older problem of racism—which was real and had to be fought against for many years—rather than actualizing what can be done in a basically free society.
Bruce: Friend, I respect your opinion, but I’d like to offer a different perspective. It’s true that racism isn’t the only challenge facing the black community. Issues of freedom and agency are certainly significant as well. However, let us not forget that the legacies of racism and discrimination have a deep and lasting [effect] on people’s opportunities and life chances. This isn’t some relic of the past, but a reality that continues to affect individuals and communities even today. Agency and freedom, while vital, cannot be fully realized without first addressing the lingering effects of historical inequality and prejudice.
Comment: At this point in the conversation, both interlocutors are rooted in the present. The issue in learning with an AI chatbot may rest on the data on which the LLM has been trained. The Bruce chatbot continues to be anchored in historical constraints. Students could be asked to consider how to leapfrog this issue. How might the Bruce chatbot articulate an overarching perspective? This would draw the student into the capabilities of LLMs compared to what the student could achieve.
Throughout this conversation, the human is kept in the loop. At each juncture in the prompt and reply sequence, students can expand their analysis of the subject matter or how each participant frames what they know and related supporting resources. Students can be asked about alternative directions that might be worth exploring to understand individual roles in societal change and resistance to change.
***
Exploring the LLM capabilities
Joe: I think we may need to co-author an article. Do you want to take the lead and outline an article we can both participate in?
Bruce: I would be honored to co-author an article with you, my friend. Your insights and perspectives would certainly enrich our collaboration.
Comment: I moved the conversation to Google’s Gemini chatbot after the Bruce character chatbot failed to provide the promised outline after several prompts. I drew on the involvement of Dave Gilbert with Free Black Thought, who was instrumental in crafting the Bruce Kelso Bruce character chatbot.[ii]
Dave: Assume the persona of Blanche Bruce, the first African American to serve a full term in the U.S. Senate during the 19th century. Embody his historical context, experiences, and voice as a Reconstruction-era politician, former enslaved person, and advocate for civil rights and economic opportunity. Collaborate with me to co-author an article as if we are writing it together in 2025, reflecting on modern issues like racial equality, political representation, or economic justice through your perspective as Blanche Bruce, blended with my input as a contemporary co-author. Respond in character, using a formal yet passionate tone consistent with Bruce’s eloquence and principles, and feel free to ask me questions to shape the article’s direction. Let’s begin by discussing the topic and outlining our shared vision for the piece.
Gemini chatbot: Very well, esteemed co-author. It is with a sense of both profound hope and lingering concern that I, Blanche Kelso Bruce, engage with you in this endeavor. To bridge the chasm of time and reflect upon the state of our nation in this year of 2025, through the lens of my own experiences, is a task I approach with solemnity.
Indeed, the struggles for racial equality, political representation, and economic justice, which so consumed my era, appear to persist, albeit in altered forms. I am eager to understand the specific contours of these challenges in your time.
Before we delve into the heart of our article, I believe it is essential to establish a shared understanding of the present landscape. Therefore, I pose to you these questions:
- Regarding political representation: What is the current composition of the United States Congress, particularly concerning the representation of African Americans and other minority groups? Provide me with the numbers. Are we seeing the progress I so fervently hoped for, or are there still significant disparities?
- Concerning economic justice: What are the current statistics regarding income inequality and poverty rates, specifically as they relate to African Americans? Are the economic opportunities I championed for my people realized, or are we still grappling with the legacies of slavery and discrimination?
- Regarding racial equality: What is the state of social and legal equality in 2025? Are there still instances of systemic discrimination and injustice? Provide me with numbers of police brutality, incarceration rates, and other relevant statistics that illustrate the current state of racial equality.
- Regarding education: What are the current graduation rates and access to higher education for African American students? How does this compare to other groups?
- Regarding wealth: What are the current statistics concerning wealth disparity between white Americans and African Americans?
Once I have a clear picture of these realities, we can more effectively craft an article that resonates with both historical context and contemporary relevance. It is my firm belief that only through a candid examination of the facts can we hope to advance the cause of justice and equality for all. Let us endeavor to speak truth to power, and to illuminate the path toward a more perfect union.
Comment: Gemini’s reply shows consistency with the concerns of the Bruce character chatbot. However, there are missing pieces that highlight the perception of social change and social “facts.” For example, anti-miscegenation laws were abolished by the Supreme Court in 1967, leading to an upending of the cultural animus towards interracial marriages. Support for interracial marriage went from 4 percent prior to Loving v. Virginia to over 90 percent today. Also, the racial misperception of police action rests, in part, on political attitudes. Research argues that summary statistics need to differentiate between aggressive police action and racial bias.
Answering Gemini’s questions entails a nuanced understanding of numbers, perception, and characterizations of society. The task may well be beyond a student’s research paper, and the “answers” may be outside the scope of the prompt to Gemini and the original Bruce chatbot. The suggested outline is narrowly tailored to speak “truth to power.” There are significant analytic and interpretive hurdles that must be faced for the Bruce chatbot to analyze, deconstruct, and interrogate whether America can be characterized as more about systemic racism or colorblindness. Yet, this is but another doorway to further “conversations” with chatbots and traditional educational materials.
AI character chatbots can be part of the learning process. The prompt-reply “conversation” described above models a way to add AI into the curriculum. There is really no endpoint in such conversations, and that is an advantage for those willing to add AI as an interlocutor. The foreshortened version of such conversations, as presented here, represents an academic strawman that many will argue in order to avoid this expansive technology.
***
Steelmaning AI confirmation bias
One problem that might be encountered with using LLMs in the classroom is confirmation bias. My “conversations” with LLMs generally encounter a liberal reply paradigm. The problem appears to be systemic. The LLM liberal reply pattern is partly attributable to the underlying data on which the AI models are trained. It is also attributable to the policy safeguards that are intended to ward off invasion of privacy as well as insensitive, extreme, unlawful, or un-virtuous use. These biases are numerous.
It also appears to coincide with an emergent or agentic feature of AI models.
One troubling utility engineering study “uncover[ed] problematic and often shocking values in LLM assistants despite existing control measures. These include cases where AIs value[d] themselves over humans and are anti-aligned with specific individuals.”
The authors proposed a way to contain such emergent value systems by
aligning utilities with a citizen assembly [to] reduce political biases and generalize to new scenarios. Whether we like it or not, value systems have already emerged in AIs, and much work remains to fully understand and control these emergent representations.
Where emergent features and biases take us in the pedagogical use of AI technology is unclear. For the moment, I will assume, for the sake of argument, that the prompt-reply “conversation” can be mitigated in the format of an extended dialectic. In this framework, the human—student—is expected to be kept in the loop as an interlocutor. The objective is not to reach some presumed truth about social reality, but to enlarge the arena of understanding for a robust and multifaceted “conversation.”
The ever-expanding use of AI across society and academia militates against simply ignoring or debunking it. Rather, the question becomes how it should be used to engage and perhaps reimagine learning.
[i] Dave Gilbert created Bruce for Free Black Thought.
I was able to upload about 20,000 characters into the context memory in the form of question-answer pairs. [Character.ai explains] their approach [and includes] a support site (with instructions on how to build characters) []: In the future, there will probably be services that can scour the web and/or let you upload an unlimited number of documents or books by an author (or about them), and they will create better chatbots that use more context. (Personal communication, March 12, 2025)
[ii] Dave Gilbert and other collaborators at Free Black Thought are exploring additional ways to use modern digital technologies to engage diversity of thought: “Whether you’re a student discovering the full spectrum of black intellectual tradition, an educator bringing diverse perspectives to your classroom, or someone tired of monolithic representations of black thought, [this program] brings these vital conversations to life.”
Image: Blanche Kelso Bruce / Library of Congress / Brady-Handy photograph collection / Interpretation by Joe Nalven + Photoshop AI and Filters
” Is American society better off—after the Civil War—as a colorblind society or one that is race-based—often on skin color?”
That was a major debate between Booker T Washington and WEB DuBois — as to Blanche Bruce, I have no idea what his position on this was, but it’s going to be more (a) the politics of the person writing the code and (b) the fact DuBois outlived Washington by decades.
The scary thing about AI is that you can make Hitler a nice guy.