Raymond Klassen – Ideals and Identities

I have noticed that many conversations about AI begin with a worry about what we will no longer be able to do. I was listening to a middle-aged DJ chastising a Gen Z server in a restaurant for not knowing how to do the mental math required to make change and link it to AI. I have been part of a writer’s group on social media, and the overriding claim there is that AI reduces writing capacity. Even in the post secondary institutions I teach, professors express this erosion of both certain kinds of capabilities and the formation of persons through the education. Students may stop learning to write. Workers may lose expertise. Teachers may outsource judgment. Writers may become dependent on a fluency they did not earn. These worries matter. But I am beginning to think they do not go deep enough.

Both worries matter because both belong to human agency. To be an agent is not merely to get something done. Agency means bringing a self into action. Agency includes capability, ownership, answerability, and presence. I act most fully when I can do something, when I can stand behind what I have done, and when others can meet me in the action.

That suggests a third worry, one that may be the most overlooked: AI threatens presence. It may thin out the human availability that makes speech, care, judgment, apology, friendship, teaching, and love trustworthy. The loss is easy to miss because the product may look excellent. A condolence note may be beautifully written. A sermon paragraph may be moving. A student reflection may sound sincere. A workplace email may strike exactly the right tone. But the deeper question remains: was the person actually present in the act?

AI did not create this crisis of presence. It arrived after several earlier technologies and social events had already trained us to accept partial presence as normal. The smartphone made presence interruptible. Social media made presence performative. COVID made mediated presence ordinary and often necessary. AI now risks taking the next step: it can automate presence.

The road to partial presence

The smartphone gave us a second elsewhere, another place to be that shadows us at our physical location. We could sit at a table, in a classroom, in a church service, or with a friend while remaining available to another conversation, another feed, another notification, another audience. Sherry Turkle argued that modern communication technologies can leave us sacrificing conversation for connection. Przybylski and Weinstein found that the mere presence of a mobile phone can reduce closeness, connection, and conversation quality, especially during personally meaningful conversations. Pew also found that many adults believed cellphone use in gatherings hurt the conversation and atmosphere.

Social media deepened the shift. Social media trained us to perform presence. We learned to appear before audiences, to curate impressions, to translate experience into shareable form, and to measure the self through response. This does not mean social media is only harmful. It has also preserved relationships, enabled public witness, and given people ways to find communities they could not find locally. But it has made the boundary between being with others and being seen by others more fragile.

COVID intensified mediation. This needs a careful word. Digital mediation was often merciful during the pandemic. It allowed families, congregations, classrooms, friends, and workplaces to maintain contact when physical gathering was dangerous or impossible. Still, the pandemic also made social isolation and loneliness more visible. The U.S. Surgeon General’s 2023 advisory named loneliness and isolation as a serious public health concern and called for renewed attention to social connection.

So AI arrived in a world already weakened in its habits of presence. We were already interruptible, performative, and mediated. As AI goes further, it offers to generate the response itself. That is new. It may compose the message that will be taken as mine.

1. The capability threat: Can I still do this?

The most obvious worry is capability. Will students stop learning to write? Will workers lose hard-won expertise? Will teachers, writers, pastors, coders, administrators, and artists become dependent on tools whose operations they cannot explain? This worry is real. A Microsoft Research study on generative AI and critical thinking surveyed knowledge workers and found that confidence in GenAI was associated with less critical thinking, while confidence in one’s own abilities was associated with more critical thinking. AI can help people draft, search, summarize, revise, translate, and imagine. The danger is that capability becomes outsourced rather than formed.

A tool can extend skill when the person remains active. A calculator can support mathematics without replacing mathematical understanding. A dictionary can support writing without replacing a writer’s ear. AI can do something similar when it helps a person think, test, revise, and learn. But it becomes dangerous when it offers the feeling of mastery before mastery has been formed. The work is finished, but the worker has not grown.

2. The authenticity threat: Is this truly mine?

The second worry asks whether the finished task still bears my responsible presence. This is the authenticity worry. AI-assisted work may be polished, fluent, and technically successful while still raising the question: whose voice, judgment, imagination, and risk are actually present here? Discussions of authorship and academic integrity are already pressing this point. Aras Bozkurt’s work on GenAI, authorship, ownership, and academic ethics highlights how AI complicates the boundary between co-creation, assistance, and misrepresentation. David Pereira similarly argues that authorship in AI-aided academic work should be treated as a qualitative threshold, not a simple yes-or-no category.

This distinction matters beyond school. A person can use AI to write a recommendation letter, a dating message, or an apology. Originality is too narrow a question. The deeper question is whether the person has inhabited the work enough to answer for it. Authentic agency requires ownership. It requires ownership, not in the possessive sense of branding the product, but in the moral sense of being willing to say: this is what I have seen, thought, risked, and can stand behind.

AI gives the false self unusually good tools. It can help us appear thoughtful before we have attended, vulnerable before we have risked, compassionate before we have entered the conditions of another person’s life, and knowledgeable before we have participated in learning. The reaches beyond fakery into familiar distance. The right words can come very near while the person remains far away from the conversion those words require.

3. The presence threat: Am I really here?

The third worry is the most overlooked because the loss may not show up in the finished product. Presence asks whether I am available through my words and actions to another person. Can I be questioned? Can I be corrected? Can I be trusted? Can I be affected by the response? Human agency is not complete until it appears in a shared world where others can receive it and answer it.

Research on AI-mediated communication helps name the issue. Hohenstein and colleagues found that algorithmic responses can change language and social relationships by increasing communication speed and positive emotional language, while also affecting how people evaluate one another. Daniel Battisti’s work on second-person authenticity argues that in intimate relations, the value of a message can depend not only on its quality, but on whether the other person recognizes you as the one actually present in the act. Sahebi’s work on AI-mediated communication and epistemic trust similarly warns that AI mediation can diminish trust in online communication.

This is an argument about substitution, not mediation. A letter can have presence. A phone call can have presence. A Zoom conversation can have presence. Even AI-assisted writing may preserve presence if the person remains genuinely involved, accountable, and responsive. Mediation is not the enemy. Substitution is. AI threatens presence when my words arrive but I have remained safely elsewhere.

Consider an apology. AI may produce a better apology than I could produce on my own. It may include empathy, responsibility, regret, and a commitment to change. But an apology is not merely a well-shaped text. It is a person becoming available to truth, consequence, repair, and the wounded other. If I use AI to help me find words for the repair I am actually seeking, the tool may serve presence. If I use AI to avoid the vulnerability of apology while sending language that looks repentant, the tool has helped me simulate presence.

What kind of agency do we want to preserve?

Capability is the worry we see most quickly. Authenticity is harder to name. Presence may be the one we have barely begun to understand. Together they show that the question of AI is anthropological. What kind of human beings are we becoming with these tools?

AI can serve agency when it helps people become more capable, more truthful, and more present. It can help a student revise a sentence while still requiring the student to think. It can help a teacher design a lesson while still requiring the teacher to know the learners. It can help a writer clarify an argument while still requiring the writer to answer for the argument. It can help a person find words for grief, apology, or gratitude while still requiring that person to enter the relationship more fully.

But AI threatens agency when it replaces formation with production. It threatens capability when the tool does the work without strengthening the worker. It threatens authenticity when the output sounds like me without becoming mine. It threatens presence when my words reach you while I remain safely elsewhere.

The central question, then, is what AI allows us to do. It is whether we remain answerable in the doing. Can I do this? Is this truly mine? Am I really here? Those three questions may help us discern whether AI is extending human agency or hollowing it out.