Location
South Ballroom, Hemmingson Center
Start Date
3-4-2025 4:15 PM
End Date
3-4-2025 5:00 PM
Description
Chaired by Julia French, Ph.D. (University of Notre Dame)
Almost everyone cares about respect. More specifically: almost everyone wants others to communicate with them in a respectful way. When they don’t feel respected, people shut down or retaliate. And when we feel that we’ve failed to show others respect, we can feel guilt and isolation. As LLMs increasingly become part of human-to-human communication, however, the question arises: does AI help or hinder communicative respect?
In this talk, I offer a framework for answering that question. The framework is inspired by the most influential philosopher of respect in the European tradition: Immanuel Kant.[1] Despite its historical inspiration, though, this framework is meant to provide guidance for those who design and implement AI within communicative technologies. Of course, frameworks aren’t always helpful – especially when they make a simple issue complicated. But respect, I argue, is not simple. It has, in fact, four dimensions. It is easy for all of us to overlook one or more dimension of respect. Within communication, the four dimensions are sincerity, receptivity, non-interference, and self-respect. To properly evaluate the impact of AI on person-to-person communication, we need to examine all four dimensions.
People sometimes talk of respect as though it were an absolute, all-or-nothing value. I suggest that is a mistake, and not just because it is multidimensional. At most, each dimension of respect corresponds to a default, defeasible moral demand: extreme circumstances and mutual agreements can make it permissible for people to bracket one or more dimensions of respect. As a simple example: when your life is in danger, you can lie and even harm the person threatening you. But these default demands are still strong. We don’t excuse lies and harm easily.
The first dimension of communicative respect is sincerity. A respectful communicator avoids lies, bullshit, and exaggeration. The words they use convey what they honestly think and feel. As anyone who has written wedding vows or an apology letter knows, sincerity is often difficult. Finding the right words can take hours. In principle, AI can assist with this, for example, by quickly providing a list of related phrases that we can consider using. AI can also undermine sincerity, however, if the technology does not prompt people to consider whether an AI-generated message really matches what they think. When looking for a phrase, it’s easy to just pick the simplest, nicest-sounding candidate from a given list, even when it doesn’t match one’s genuine but complex thoughts and feelings.
The second dimension of communicative respect is receptivity. A respectful communicator is willing to receive ideas, insights, and corrections from the person they’re communicating with. Genuine receptivity is difficult, especially when people start off with radically different perspectives, or when they have different approaches to communication. In principle, AI can facilitate receptivity, by translating or rewording messages to make them easier to understand. For example, a secular urban liberal could ask AI to ‘translate’ a message from a religious small town conservative into terms they’re more familiar with. It can also undermine this form of respect, however, by making it easier to respond to other’s messages without genuine receptivity to the others’ ideas. For example, the liberal might use AI to general a pleasant-enough response to the conservative’s message, without really engaging with its content.
The third dimension of communicative respect is non-interference. A respectful communicator avoids stepping on other people’s toes, that is, they avoid interfering with other people’s life projects. That is why we talk of being respectful of others’ time, since time is crucial for meeting the varied demands that are part of life. By making communication more efficient, AI can facilitate this form of respect. For example: if an administrator uses AI to the wordcount of an organization-wide message in half, it can leave employees more time for other tasks within work and outside of it. On the other hand, the very efficiency of AI-generated messages makes it easy for administrators to churn out vastly more words for employees to read, leaving them less time to meet their other professional demands, and to live their personal lives (“Sorry – I need to stay a little late to finish with my emails”).
The final dimension of communicative respect is self-respect. A respectful communicator recognizes and expresses their own value. The process of writing is often a powerful tool for self-respect, since composing a message can help someone crystalize their understanding of their own value. Assistive communicative technology can again help someone find the right words for this. Yet those same technologies can let someone skip on the task of reflection that facilitates that crystallization. As a result, someone who relies too much on AI in communication might never have a chance to understand their own priorities and their own knowledge. Without being able to do that cognitive work, someone might miss out on being able to say to themselves why they matter, and why they’re not simply replaceable.
In sum: there is no guarantee that AI-facilitated communication will either help or hinder communicative respect. Specific uses and design features can move things either way. The bad news is that this technology can easily make us less respectful in communication. The good news is that, developed, implemented, and used carefully, AI might open up novel and profound possibilities for communicative respect.
[1] I defend this framework in more detail in “Political persuasion is prima facie disrespectful” (Journal of Moral Philosophy, forthcoming) and Ethical Persuasion: How to Change Minds without Lies, Bullshit, or Moral Compromise (under contract with University of Chicago Press).
Recommended Citation
Marshall, Colin, "Can AI Foster Communicative Respect?" (2025). Value and Responsibility in AI Technologies. 4.
https://repository.gonzaga.edu/ai_ethics/2025/general/4
Can AI Foster Communicative Respect?
South Ballroom, Hemmingson Center
Chaired by Julia French, Ph.D. (University of Notre Dame)
Almost everyone cares about respect. More specifically: almost everyone wants others to communicate with them in a respectful way. When they don’t feel respected, people shut down or retaliate. And when we feel that we’ve failed to show others respect, we can feel guilt and isolation. As LLMs increasingly become part of human-to-human communication, however, the question arises: does AI help or hinder communicative respect?
In this talk, I offer a framework for answering that question. The framework is inspired by the most influential philosopher of respect in the European tradition: Immanuel Kant.[1] Despite its historical inspiration, though, this framework is meant to provide guidance for those who design and implement AI within communicative technologies. Of course, frameworks aren’t always helpful – especially when they make a simple issue complicated. But respect, I argue, is not simple. It has, in fact, four dimensions. It is easy for all of us to overlook one or more dimension of respect. Within communication, the four dimensions are sincerity, receptivity, non-interference, and self-respect. To properly evaluate the impact of AI on person-to-person communication, we need to examine all four dimensions.
People sometimes talk of respect as though it were an absolute, all-or-nothing value. I suggest that is a mistake, and not just because it is multidimensional. At most, each dimension of respect corresponds to a default, defeasible moral demand: extreme circumstances and mutual agreements can make it permissible for people to bracket one or more dimensions of respect. As a simple example: when your life is in danger, you can lie and even harm the person threatening you. But these default demands are still strong. We don’t excuse lies and harm easily.
The first dimension of communicative respect is sincerity. A respectful communicator avoids lies, bullshit, and exaggeration. The words they use convey what they honestly think and feel. As anyone who has written wedding vows or an apology letter knows, sincerity is often difficult. Finding the right words can take hours. In principle, AI can assist with this, for example, by quickly providing a list of related phrases that we can consider using. AI can also undermine sincerity, however, if the technology does not prompt people to consider whether an AI-generated message really matches what they think. When looking for a phrase, it’s easy to just pick the simplest, nicest-sounding candidate from a given list, even when it doesn’t match one’s genuine but complex thoughts and feelings.
The second dimension of communicative respect is receptivity. A respectful communicator is willing to receive ideas, insights, and corrections from the person they’re communicating with. Genuine receptivity is difficult, especially when people start off with radically different perspectives, or when they have different approaches to communication. In principle, AI can facilitate receptivity, by translating or rewording messages to make them easier to understand. For example, a secular urban liberal could ask AI to ‘translate’ a message from a religious small town conservative into terms they’re more familiar with. It can also undermine this form of respect, however, by making it easier to respond to other’s messages without genuine receptivity to the others’ ideas. For example, the liberal might use AI to general a pleasant-enough response to the conservative’s message, without really engaging with its content.
The third dimension of communicative respect is non-interference. A respectful communicator avoids stepping on other people’s toes, that is, they avoid interfering with other people’s life projects. That is why we talk of being respectful of others’ time, since time is crucial for meeting the varied demands that are part of life. By making communication more efficient, AI can facilitate this form of respect. For example: if an administrator uses AI to the wordcount of an organization-wide message in half, it can leave employees more time for other tasks within work and outside of it. On the other hand, the very efficiency of AI-generated messages makes it easy for administrators to churn out vastly more words for employees to read, leaving them less time to meet their other professional demands, and to live their personal lives (“Sorry – I need to stay a little late to finish with my emails”).
The final dimension of communicative respect is self-respect. A respectful communicator recognizes and expresses their own value. The process of writing is often a powerful tool for self-respect, since composing a message can help someone crystalize their understanding of their own value. Assistive communicative technology can again help someone find the right words for this. Yet those same technologies can let someone skip on the task of reflection that facilitates that crystallization. As a result, someone who relies too much on AI in communication might never have a chance to understand their own priorities and their own knowledge. Without being able to do that cognitive work, someone might miss out on being able to say to themselves why they matter, and why they’re not simply replaceable.
In sum: there is no guarantee that AI-facilitated communication will either help or hinder communicative respect. Specific uses and design features can move things either way. The bad news is that this technology can easily make us less respectful in communication. The good news is that, developed, implemented, and used carefully, AI might open up novel and profound possibilities for communicative respect.
[1] I defend this framework in more detail in “Political persuasion is prima facie disrespectful” (Journal of Moral Philosophy, forthcoming) and Ethical Persuasion: How to Change Minds without Lies, Bullshit, or Moral Compromise (under contract with University of Chicago Press).