The scene is familiar. Two people are talking. One mentions a book; the other searches for its title on a phone. A message arrives, and the conversation briefly pauses. Soon, someone asks artificial intelligence a question, and its answer becomes a new topic. Within a few minutes, a personal conversation, communication with an absent person, and content produced by a digital system take turns shaping the encounter. The participants remain in the same physical space throughout, but their attention shifts.
Such encounters call for a more precise understanding. Counting how often someone looks at a phone is not enough to conclude that a conversation has gone badly. What matters is what the person was doing, when they did it, whether they explained it, and how they returned to the conversation. A phone can help us find an answer together. It can also distance us from someone who is telling us something important. The difference emerges in how the device is used and the meaning that use acquires within the relationship.
To explain this process, I propose the concept of hybrid interpersonal communication. It describes encounters in which, during shared physical presence, face-to-face conversation intertwines with digital messages and content. Generative artificial intelligence may help formulate a message, provide an answer, or become the subject of a shared discussion. The concept brings together existing insights into interpersonal and digital communication. Its purpose is to help us understand what happens to a relationship when we move between different communicative activities during an encounter.
Proximity and Availability
Interpersonal communication remains central to this discussion. Through conversation, people exchange information while also shaping their relationships. In responding to someone, we express interest, understanding, disagreement, or a willingness to continue listening. In his account of interpersonal communication, Joseph DeVito connects communicative choices with their consequences for messages and relationships. In our book Temelji međuljudske komunikacije [Foundations of Interpersonal Communication], Damir Jugo and I likewise emphasise the connection between the content and relational dimensions of communication.
This connection becomes particularly clear when a phone enters the conversation. The person picking it up may intend to check a fact useful to both participants. The other person, however, may feel that they have lost their conversational partner's attention. Intention and experience do not necessarily coincide. We therefore cannot assess communication solely from the perspective of the person using the device. We also need to understand the experience of the person waiting for a response.
Physical presence makes an immediate encounter possible, but it does not guarantee involvement. Someone may sit beside us, hear every word, and still miss what we mean. They may follow the facts without noticing the concern in our voice. In other circumstances, briefly looking at a screen will not disrupt the conversation because it forms part of a shared search for information. It is therefore useful to speak of communicative availability: a person's displayed readiness to receive, consider, and respond to a message from someone physically present.
Availability is conveyed through topic-relevant responses, questions that show understanding, and how we resume a conversation after an interruption. Explaining a temporary withdrawal also matters. Saying that we need to answer an urgent message gives the other person a framework for understanding the pause. However, an explanation does not guarantee acceptance. Expectations in a conversation about loss, fear, or a serious personal problem may differ from those in an informal discussion about dinner plans.
Listening becomes particularly demanding in these circumstances. Graham Bodie's approach to active-empathic listening connects message reception, processing, and response. Hearing the words is only part of the process. The other person needs some indication that we understand what those words mean to them. A brief nod may look like attention, but an answer that misses the question reveals that the connection has not been sustained. The quality of listening should therefore be assessed through the unfolding exchange rather than through a single gesture.
Digital Content at the Shared Table
Digital communication has long played a role in maintaining personal relationships. Joseph Walther demonstrated that computer-mediated communication can have interpersonal characteristics. Christian Licoppe describes connected presence: the maintenance of relationships through face-to-face encounters and exchanges at a distance. In everyday experience, this means that someone who is not in the room can nevertheless influence the conversation: their message is relayed to those present, their suggestion is considered, or their opinion changes a decision.
This raises the question of how content moves from a private activity into a shared conversation. Reading a message silently, recounting its meaning, and showing it to others are different practices. In each case, a person selects what to convey and how to connect it with the topic. That selection may facilitate understanding, but it may also leave out important context. Those present then rely on the intermediary rather than on the complete content. Making an activity understandable does not mean giving everyone access to private messages. We can explain why we are using a phone without disclosing someone else's confidential communication. Understanding the purpose and expected duration of a pause often matters more to a conversational partner than reading the content. This protects privacy while keeping the conversation coordinated. Difficulties arise when others cannot tell whether we are still engaged in the shared topic or have withdrawn into a separate exchange.
Changes also show up in nonverbal communication. A smile may respond to a joke from the person across the table or to a message on a phone. Without context, the other person cannot know which activity prompted the reaction. Looking at a screen does not reveal everything about listening, but it changes the cues the other person relies on. If no response follows, a brief shift in attention takes on a different meaning. The sequence of events is crucial to understanding the encounter.
Research by Ryan Dwyer and colleagues shows that smartphone use can reduce enjoyment of face-to-face encounters, with distraction playing an important role. This finding deserves serious consideration, but it does not justify the claim that every glance at a device harms a relationship equally. Anja Stevic and Jörg Matthes distinguish individual phone use from shared engagement with content. In their survey, shared use was positively associated with friendship satisfaction, which does not establish that it causes greater satisfaction.
This distinction also helps explain phubbing: neglecting a conversational partner while using a phone. Hybrid interpersonal communication encompasses such situations, as well as those in which digital content stimulates conversation. Showing a photograph, reading a message together, or checking a fact can expand opportunities for participation. The key question is whether the way the device is used leaves room for questions and responses. Showing a screen is not the same as involving another person.
Three Patterns of Interaction
For practical understanding, I propose three patterns: integrative, parallel, and disruptive. These are theoretical patterns of conversational episodes, rather than permanent labels for people. A single encounter may contain all three. Their value lies in recognising changes in the connections between content and in mutual availability, rather than in assigning every event to one category.
The integrative pattern emerges when digital activity enters a shared topic and allows those present to participate. While discussing a trip, we look at a map together and compare options. Someone opens an AI-generated answer, and we all consider it and ask a follow-up question. Digital content then becomes material for conversation. Integration is more successful when others can see the relevant content, comment on it, and connect it with their own knowledge. The parallel pattern means that face-to-face conversation and a separate digital activity occur within the same encounter, while availability is maintained or negotiated. A person announces that they will briefly answer a message and then return to the topic. The content remains largely separate, but the participants understand what is happening. The acceptability of such a pause depends on the relationship, the situation, and expectations. It may be routine in one group and require a different agreement in another.
The disruptive pattern arises when digital activity makes mutual responsiveness and conversational continuity more difficult. A question goes unanswered, someone repeats a sentence, or explicitly asks for attention. The digital content may still be related to the topic. For example, someone searches for an answer on behalf of the group but takes over the searching and reading to the point that others can no longer participate. A good initial intention does not, by itself, ensure good communication.
The boundaries between these patterns are fluid. A separate message may become a shared topic, moving a parallel episode towards integration. A joint search for an answer may become disruptive if one person begins to disregard others' comments. Device use duration is therefore insufficient as a criterion. A brief activity during personal disclosure may have a stronger relational effect than a longer pause everyone accepts.
Artificial Intelligence as Assistance and a Source of Content
Generative artificial intelligence expands the possibilities within these encounters. In their book Umjetna inteligencija u obrazovanju i robotici [Artificial Intelligence in Education and Robotics], Tomislav Volarić and Bojan Crnokić discuss the technological foundations of language processing and chatbots. To understand contemporary communicative situations, we also need to distinguish the roles a system assumes. Helping compose a message, responding to a user, and contributing an answer to a conversation among people are different processes.
In the first role, the system helps formulate a message intended for another person. It assists with translation, tone, or editing a response. Jeffrey Hancock, Mor Naaman, and Karen Levy explain such practices through the concept of AI-mediated communication. The human relationship remains the setting in which the message acquires meaning. The sender decides whether the text reflects their intention and takes responsibility for what they choose to accept. In the second role, the user interacts directly with the system. This falls under human–machine communication, as examined by Andrea Guzman and Seth Lewis. Such an exchange may take place while we are physically with other people, but it should be distinguished from interpersonal conversation among humans. A system can produce an answer that appears conversational. Its involvement nevertheless requires a different understanding than a human relationship and personal responsibility.
The third role arises when we bring the system's response into a shared discussion. Participants may formulate a question together, read the result, and compare it with their own experience. Alternatively, they may hear only a summary selected by one person. This distinction matters because access to the question and answer determines how well others can assess the content's relevance. The less they see, the more they rely on the person conveying it. Artificial intelligence is not a necessary condition of hybrid interpersonal communication. An encounter can already be hybrid through messaging or the inclusion of online content. The system is an additional possibility within a broader process. Nor is the mere presence of a phone on the table enough to classify an encounter as hybrid. Digital activity must unfold through content, a response to a notification, or an availability change.
Trust and the Authority of an Answer
When artificial intelligence helps write a personal message, the question of authenticity arises. How far does the text express the sender's intention, and how does the recipient understand the sender's personal contribution? Assistance with wording does not automatically imply a lack of sincerity. At the same time, polished phrasing does not guarantee understanding or a willingness to act. In relationships, people assess a message alongside the behaviour of the person sending it.
In experiments involving text-based communication, Jess Hohenstein and colleagues distinguish the actual use of suggested replies from the belief that a conversational partner used them. Their findings show why technological assistance must be distinguished from its attribution. This research does not establish the effects of every form of generative artificial intelligence in face-to-face encounters. It nevertheless raises a relevant question: how does our assessment of a message change when we believe that a system helped formulate it? Attributing authority to an answer is even more sensitive. The statement "Artificial intelligence says this is the case" can invite verification, but it can also attempt to close a discussion. In the latter case, content gains weight without its assumptions being considered together. Participants need opportunities to ask how the question was framed, what the answer omitted, and what it is based on. Confident wording is not enough to accept a statement as fact.
The person operating the device gains a certain advantage. They choose the prompt, see the answer, and decide what to convey. This does not mean that they are manipulating the conversation, but it creates unequal opportunities for assessment. If others receive only the final claim, they have little opportunity to consider how it was produced. Shared access to content allows further questions, although it does not, in itself, ensure accuracy. Decision quality still depends on human judgement and verification.
This is particularly important in professional communication. During a meeting, a system's answer may help formulate a proposal or identify an overlooked question. If someone presents it as definitive confirmation, others may stop questioning it. In public relations, this practice shapes meaning and determines who bears responsibility. A professional should explain the status of the content they convey and leave room for critical assessment.
Sustaining the Conversation
The practical value of this concept begins with recognising expectations. Before engaging in a digital activity, it helps to consider what the other person needs at that moment. Sometimes, a brief explanation of why we are using a device is enough. At other times, the conversation requires full attention, and the search should be postponed. In personally sensitive situations, this judgement carries greater weight than in a routine exchange of information. When digital content supports a shared topic, it should connect to what has already been said. Showing the relevant part, explaining why we are introducing it, and allowing a response is more effective than reading at length without a pause. When using artificial intelligence, it helps to distinguish a suggestion, a summary, and verified information. This gives the other person an opportunity to understand the kind of content being considered.
Returning after an interruption is itself a communicative act. Picking up where the other person left off, answering their question, or asking for clarification demonstrates a willingness to restore the exchange. If we have lost the thread, acknowledging this is more honest than pretending to understand. Such a return helps prevent a brief digital pause from becoming a lasting withdrawal during the encounter.
These recommendations follow from a theoretical framework; they are not universal rules established for every situation. The proposed typology remains open to further development and testing. Its value lies in connecting attention, content, and relationships. It helps us ask more precise questions about conversation and avoid simplistic judgements that devices always undermine an encounter or that every form of technological assistance improves it.
Hybrid interpersonal communication reveals the importance of human communication skills in a technologically rich environment. Listening, explaining, and responding determine whether digital content contributes to shared understanding. At the same table, we may be connected to many sources and people. The quality of the encounter becomes apparent in whether the person physically with us has an opportunity to be heard and to participate in the conversation.