The AI social paradox: Better at communicating, worse at connecting?

A text from a family member. An email from the boss. A friend seeking advice after an argument. These are ordinary moments that are part of everyday interactions. Yet, increasingly, there is another participant in these conversations, one that neither feels emotions nor has any personal stake in the relationship - artificial intelligence.For generations, people learnt social etiquette, understood different perspectives and navigated relationships through actual interactions. They often said the wrong things, misunderstood intentions and learnt from those experiences.Now, with AI offering an instant answer to every social dilemma, some of that effort can be outsourced.So, why has replying to an ordinary message become something we might need assistance with? Also while AI may make us more articulate, polite and confident, does that necessarily make us better at connecting?A 'social animal' with AI?Humans have long been described as social animals. Relationships are shaped not merely by what one says, but by how they listen, react, disagree and adapt to one another.AI, however, offers something human relationships rarely can: immediate attention, endless patience and responses tailored to specific concerns. There is no awkward silence, no raised eyebrow and seemingly no fear of embarrassment.This was highlighted by Shailesh K Nevatia, founder director of Grandeavour Communication, having spent nearly 25 years in the communication industry but described himself as shy when it comes to open communication concerning himself.

He said AI helps him navigate differences of opinion, offering perspectives he might otherwise overlook. "With humans we may have certain opinions, views, or thoughts, while with AI, the most valuable thing it offers is not the right answer but a different perspective than us," he said.

Nevertheless, he maintained that the emotional depth, vulnerability and shared experiences of human relationships cannot be replicated by a chatbot.When the robot does the talkingThe concern with AI-assisted communication is not just that people are using it to find the right words, but that repeated reliance on it could weaken their confidence in finding those words themselves.For instance, Riddhi turns to ChatGPT almost instinctively when responding at times. Whether she needs to sound polite yet firm in a professional email or find a way to decline plans with friends, AI helps her navigate the interaction.

"It's not like I can't reply without AI, but having someone to proofread the convos is better than just listening to how they sound in my head," she said.Yet, the reassurance comes with a trade-off for her. "AI has such a grasp on my writing that I am not very confident even while writing this, and that is, without being said, one of the biggest disadvantages." She also called to embrace AI but be careful.

Her experience pointed to a potential paradox: a tool intended to make communication easier can also leave users less confident about communicating independently.The stakes may be higher when AI is used not just to frame a response, but to interpret another person's intentions. Consider someone who turns to a chatbot after an argument and asks who was at fault. The response can sound considered and convincing, but it is based on the account presented to the system, which may omit details or reflect the user's own biases.In that context, Reshmithaa Nair, counselling psychologist at SPARSH Hospital, Hennur Road, warned that such interactions can reinforce confirmation bias, making people more convinced of their existing interpretation rather than open to another perspective.That matters because disagreements are also opportunities to develop social skills. Understanding another person's point of view, managing uncomfortable emotions and responding to unexpected reactions require practice, not simply a well-worded reply."Social skills are learned in actual situations that involve emotions, misunderstanding, conflict, and sometimes rejection," Nair explained.Repeatedly delegating these situations to AI could reduce opportunities to develop empathy, emotional intelligence and the ability to navigate uncertainty.

A person may learn to phrase a difficult message more effectively without necessarily becoming better at handling the conversation that follows.The missing human touchA message is not simply a collection of sentences. Its meaning can change with the relationship between the people involved, their shared experiences and even the pauses between words.Ritu Mehrotra, Founder and CEO of Shunya Labs, a voice AI company, pointed out that natural communication involves much more than language."Communication is not only about words, it’s also about tone, pauses, context, emotion and all the things we communicate without explicitly saying them."Advances in voice AI are making it possible for technology to interpret language, intent and context with greater sophistication, including across different accents and communication styles. Yet, recognising emotional signals is not the same as experiencing emotions."And as AI is becoming a part of more and more personal conversations, we have to realise that understanding emotion is not feeling it. Trust, privacy, and transparency must be fundamental," she said.

A chatbot may recognise expressions of sadness or affection and produce an empathetic-sounding response. But recognising emotional patterns is different from sharing the emotional responsibility of a human relationship. For ex, an AI chatbot was asked to give advice to reply on a (hypothetical) "text to a friend who was laid off"

While the response was considerate, it was also generic. It offered comfort without knowing the friend's emotional state, personality or the nature of the relationship.

Even a seemingly caring question like “Are you OK?” may fall flat when the person is looking for encouragement rather than a question whose answer seems obvious. A more personal response might acknowledge the setback while affirming the friend's abilities: “OMG, that's awful.

I know the job market is tough, but you're talented. You'll land on your feet.”The difference is not simply about choosing warmer words. It is about understanding what a particular person might need in that moment, something a chatbot may struggle to judge without sufficient context.There is also the question of authenticity. When a message is carefully constructed by AI, it may communicate the intended information effectively. But does it still sound like the person sending it?In this example, of AI removing the human touch, another chatbot was asked "Someone close to me is grieving. I don't know what to say, and I'm afraid of making things worse. What should I tell them?". It gave a response, which is not wrong but is too robotic. "I am here with you. I care about your pain. Speak if you wish; silence is acceptable." Perfect words but not someone who is comforting says in such "rigid tone"The little hesitations, imperfect expressions and familiar phrases that characterise someone's communication can disappear beneath polished language.

A message can become more grammatically correct while feeling less personal.

Not all badDismissing AI as a threat to social skills, however, would overlook how it can help people who struggle to express themselves.For Pooja, a college student who hesitates to speak up in front of the entire class or repeatedly ask friends to explain something, AI offers a space where she can ask questions without worrying about others' reactions.She uses it to understand complicated topics and brainstorm ideas, sometimes asking ChatGPT to explain something as though she were a 10-year-old.

Yet, she recognises the possibility of becoming overly reliant on it. "AI can give you answers in seconds, but actually understanding and learning something is still on us."For others, AI serves as a sounding board before difficult conversations. It can also help people become more mindful of how their words might affect others.

It can suggest more inclusive language, flag potentially insensitive phrasing and help users consider perspectives they might otherwise overlook. For someone unsure how to communicate across differences, this can offer a starting point for more thoughtful interactions.But knowing which words are considered appropriate is not necessarily the same as understanding why they matter. Genuine inclusivity requires empathy and a willingness to understand experiences different from one's own, qualities that must extend beyond the words suggested by a chatbot.Also, in such cases, might not give you the perfect answer, but may help you think in the right direction and give you options that you may not have thought about. In another hypothetical test, a chatbot was asked how to support a family member struggling with their mental health without pushing them away. While its response was not fully correct, it offered different approaches to consider, giving the person a starting point for thinking about how to respond.

Thus, the value, then, may not lie in AI providing the perfect answer, but in helping someone see possibilities they had not considered. The important distinction is whether people use these suggestions to develop their own judgement or simply follow them without question.As Shruti Rajan Kappil, clinical lead, excellence and quality at a mental health care platform Amaha, explained, that while the need to feel heard and supported is fundamental to human relationships, not every concern feels safe to share.

Shame, embarrassment and fear of judgement often prevent people from opening up, particularly where stigma remains prevalent.

AI offers an alternative, but Kappil pointed to an important distinction. "Good support often needs honest disagreement or gentle challenge with a different point of view. AI tends to be agreeable more often than not, and being agreed with is easily mistaken for being helped."She also noted that repeatedly seeking AI's approval before sending a message can become a form of reassurance seeking.

While it offers immediate relief, constantly checking whether one's words are "correct" may gradually undermine confidence in one's own judgement.Shristi Chaudhari, senior executive, regulatory intelligence and market research at Bureau Veritas, uses AI almost daily to condense regulatory updates. Increasingly, she also turns to it for interpersonal communication.From identifying emotional undertones to making messages warmer and more polite for colleagues and professional contacts, AI has become a useful sounding board.

These experiences suggest AI can help people prepare for conversations, overcome language barriers and articulate thoughts they might otherwise struggle to express.Perfect talks, weaker bonds?Communication is one of the ways human beings establish connection. But effective communication and meaningful connection are not necessarily the same thing.

One involves expressing something clearly. The other requires understanding, trust, emotional involvement and vulnerability.Tanvi Mankodi, assistant professor at SPJIMR, believes the deeper concern is not whether AI will replace human relationships, but what excessive dependence might do to our ability to participate in them."For me, the essence of human relationships is about our vulnerabilities, about our imperfections, and about how we try to relate with each other with our discomforts and uncertainties," she said.Misunderstandings, difficult conversations and moments when people struggle to find the right words are also experiences through which patience, compassion and empathy develop."Rather than worrying about AI replacing human relations, I am more worried about how if we use AI in a way that we are dependent on it, it might reduce our ability to be vulnerable and to be human," she added.

Her concern captures the paradox: in trying to eliminate the discomfort of human interaction, we may also remove experiences that make relationships meaningful.

Of course, not every AI-generated message weakens a relationship, and the long-term effects of outsourcing everyday communication are still being understood. AI may help some people overcome barriers to social participation while encouraging others to avoid interactions through which confidence develops. But, the concern is can it address the individuality and complexity of human behaviour?However, using AI to prepare for a difficult discussion is one thing.

Allowing it to decide how we interpret another person's intentions and what we say in every interaction is another.The point is not whether AI can offer thoughtful, nuanced or well-structured answers to human dilemmas. It is whether those answers can replicate the emotional understanding, vulnerability and lived experiences through which we learn to navigate relationships and grow socially. After all, relationships are not assignments to be completed with perfectly structured answers. They evolve through disagreement, compromise and sometimes saying something imperfectly but sincerely.Perhaps the real challenge is not learning how to communicate better with AI, but ensuring that, in the process, we do not forget how to connect without it.