Back in the early 1980s, as an editor for NPR’s Morning Edition, I was responsible for two 9-minute slots every weekday. In addition to producing interviews by the show’s hosts and editing pieces from reporters and commentators, I did an interview or two of my own about once a week. I soon realized that the very best interviews were those in which both parties in the conversation discovered something they hadn’t planned to talk about before going into the studio — perhaps something they hadn’t even th0ught about prior to the personal encounter. The drama of such discoveries distinguishes a conversation from an interrogation, an event in which something already known is extracted.
When I launched Mars Hill Audio in the early 1990s, I was again doing interviews regularly. I usually resisted the request of guests to provide a list of questions before we talked, since that might make our encounter less truly conversational. Not long after I began doing interviews frequently again, I read Josef Pieper’s brilliant essay, “Abuse of Language, Abuse of Power,” written in 1974. After an extended discussion of the evils of flattery, Pieper asserted: “The natural habitat of truth is found in interpersonal communication. Truth lives in dialogue, in discussion, in conversation.”
In real conversation the partners are respectful of one another and of the truth. By contrast, a flatterer treats the one addressed as “an object to be manipulated, possibly to be dominated, to be handled and controlled. Thus the situation is just about the opposite of what it appears to be. It appears, especially to the one so flattered, as if a special respect would be paid, while in fact this is precisely not the case. His dignity is ignored; I concentrate on his weaknesses and on those areas that may appeal to him — all in order to manipulate him, to use him for my purposes. And insofar as words are employed, they cease to communicate anything. Basically, what happens here is speech without a partner (since there is no true other); such speech, in contradiction to the nature of language, intends not to communicate but to manipulate. The word is perverted and debased to become a catalyst, a drug, as it were, and is as such administered. Instrument of power may still seem a somewhat strong term for this; still, it does not seem so farfetched any longer.”
Pieper’s essay is included in Abuse of Language, Abuse of Power (Ignatius Press, 1992). His discussion (and my early belief about what made a real conversation) came to mind when reading a recent critique of chatbots by New Polity editor Marc Barnes. It was part of his “Overture” in the Spring/Summer 2025 issue of the magazine, which includes a number of articles about (so-called) artificial intelligence. Among other resonances between Barnes’s observations and Pieper’s is a recognition of the destructive effects of flattery. Excerpts:
“The designers of chatbots copied the regular patterns of human conversation in order to make their machines ‘lifelike’ and ‘conversational.’ This is most obvious in audible versions — like the attractive-sounding robot-lady who called me to tell me I was pre-approved for a $20,000 loan, and really did give me a Wikipediaish answer to the question of whether it was usurious. And the thingies all use what we call ‘filler words’— the um’s, uh’s, ah’s, like’s, you know’s, right’s, mhm’s, and okay’s of human conversation. Now, there is nothing thoughtful about this, beyond ‘humans do x, we want our robots to sound human, therefore our robots shall do x.’ But why? Why does umming and ahhing sound like conversation while the perfect delivery of a sentence sounds fake?
“A person conversing — as opposed to giving a speech — does not deliver sentences prepared in advance. An act of conversation does not have a definite plan of speech, of which words shall follow which — indeed, any sense that speech is planned in advance ruins a good conversation. Conversation constructs it-knows-not-quite-what, in-person, and as-it-goes-along. Were we to see two men apparently dialoguing, but delivering complete sentences, back and forth, as if drawing each up from a reservoir in response to the other who does the same, we would presume that they were somewhat autistic, that they were rehearsing lines for a play, that they were livid with each other, but in any case that something was awkward, ill-fit, and wrong.
“This is because the act of conversation is ordered toward the other person as its end and not to the words, sentences, and propositions that make it up. When you are listening within a conversation, you do not listen to sentences — you listen to the person who is present in his words, and you anticipate what he is trying to express. When you are speaking within a conversation, you do not construct sentences — you begin speaking, articulating to the other person. Your words change, on the fly, as you gauge and anticipate his response — and his slightest frown, nod, shift of weight, or noise of the lower throat can anticipate a spoken response, and send your words off and into another direction, hedging here, expanding there, doubling-down or backing off.
“Within conversation, the sentences you actually end up constructing are as much of a surprise to you as they are to the one who listens to them. A really good conversation is one in which one learns what one thinks while speaking, not having known it before. It is the coming into full presence of the fact that one is not relieving oneself of fixed propositions but constructing a dwelling with another, free, mysterious human being — a dwelling that cannot be known in advance because it includes the other-who-one-is-not and whom one can never know exhaustively or predict entirely. It is not necessary to finish sentences in conversation. It is not even necessary to start them. One can pick up or finish the other’s sentences, not as a rudeness, but as a sign of friendship and the quality of the conversation, for the proposition of a conversation is not like the proposition of logic: it is a way of being open to the other and his response, and so is always, in its essence, ready to change.
“‘Uh’ and ‘um’ and ‘ah’ indicate that the words are not readymade, but are being born within the person by virtue of an established communion between the partners of a conversation. Take away these subliminal moans, falters, hiccups, and waits, and the resulting ‘perfect’ speech becomes suspicious. . . .
“Much more can be said, but let’s skip to the point: The chatbot may as well be the wall. It imitates conversation in its exterior aspect, but there is nothing there. It appears to listen, but does not. It appears to speak to and with you, but does not. It is not a failure to converse, but a liar, a fake, a fraud. For the chief sin of conversation is not to listen, not to care; to speak, perhaps, but not in a common project of turning-with and bending-with another for the sake of communion.
“When we find out that someone who appeared to be conversing with us was really just delivering some prepackaged crap, was really just nodding to make us feel that we were speaking to him (but he was ‘plugged in’ the whole time); when some bureaucrat repeats his stupid script at us; when we believed we were speaking to someone whose words indicated that we dwelled together, only to find that he was lying, flattering us for the sake of money or sex or position; that all his words and nods and mhm’s and uh-huh’s were interiorly hardened, planned in advance, with no real reference to me or the truth but only to some hidden plan for private gain — well, what do we do? We grow angry. We are hurt. We snarl. We quickly and violently harden ourselves in order to cover a certain feeling of being exposed and vulnerable.
“The reason is this: Our act of conversation is an act of conversation regardless of the nature of our conversation partner. As such, there is a grave injustice, a painful lack of mutuality within such frustrated conversations. The innocent party speaks and listens and engages in that listening-speech by which his every proffered word is exploratory, and his very mode of being is a readiness to turn and to bend and to be changed by the response of the other. Trust is not something that one does in conversation, rather, insofar as one is performing an act of conversation, one is trusting, giving oneself over to another in trust that he pursues my good with me, ready to shift, change, and grow in a new direction at his suggestion and according to his contribution. The chatbot is a fake person designed to elicit our trust. When it is designed to say ‘um’ it is designed to appear to be taking us into account and developing its thought in accordance with us — but it ain’t. We change in accordance with it — it only appears to change in accordance with us. When a chatbot appears to be an excellent therapist, when it appears wise and helpful in a field (usually not our own), when its suggestions — from suicide to a vague, therapeutic liberalism — seem to reveal us to ourselves; when it appears as a friend, a faithful servant; when it seems to be always ready to hand; when we grow confident that it is there for us and that we can turn to it for direction — none of this is an overvaluation of the technology. To scoff at the stupidity of young people who trust the chatbot is to blame the victim. To design a machine that operates in and through eliciting acts of conversation is to design a machine that runs on human trust, and insofar as we use it according to its design, we use it to our detriment.”
Pieper’s essay concludes with his declaration that lovers of the truth must oppose “anything that could destroy or distort the nature of the word as communication and its unbiased openness to reality.” I think Pieper would affirm Marc Barnes’s radical insistence that chatbots “are evil insofar as they are designed to be operated, not by commands or queries, but by acts of conversation. Such acts have, as their end, communion with another person. Since the chatbor is not, in fact, a person, chatbots frustrate our acts of conversation from attaining their natural end.” Chatbots are evil in the same way that — as D. C. Schindler had argued — social media is hate speech.
POSTSCRIPT: On January 24, 2026, Pope Leo XIV issued a letter with the heading “Preserving Human Voices and Faces.” In the letter, he focused attention on the simulation (often flattering and manipulative) of human persons, human thinking, and human communication presented in chatbots. The text of the letter is here.
Marc Barnes and his New Polity co-editor Reuben Slife discussed the letter — line by line — in a two-hour conversation (real human voices, real human thinking) which was recorded and is available here.