Commentary weighs AI companionship against human connection, citing four-country attachment study

Uganda-based PC Tech Magazine published a staff-written opinion piece on 9 November asking whether AI companions can stand in for human relationships. Its answer: they can help, but should not replace people.

What we know

The piece's main evidence is a 2026 cross-national study of 7,027 people in Germany, China, South Africa and the United States. Of the 5,310 chatbot users in the sample, about 35% said they missed their chatbot after several days without it, PC Tech Magazine reported. The research, published in the journal Technology in Society, found that perceived emotional support was the strongest predictor of attachment, and that attachment was strongly associated with signs of dependence. Coverage of the study has noted its limits: the sample was not representative, and the cross-sectional design cannot show cause and effect. Contrary to the researchers' expectations, people with larger social networks reported stronger attachment.

The essay credits chatbots with being available at any hour and low-pressure, which appeals to lonely, grieving or socially anxious people and to those who cannot afford therapy. It also sets out the risks: inaccurate or confidently wrong advice, no clinical accountability or crisis assessment, privacy exposure, and fewer chances to practise the give-and-take of real relationships. It calls on developers to be clear about what their systems cannot do, avoid encouraging dependence and point users to professional help.

Why it matters

The piece contains no new reporting, but it shows where mainstream opinion now sits. Commentators are no longer asking whether people form bonds with chatbots. They accept that people do, and are asking whether operators are responsible for managing those bonds. The developer obligations it lists, honest disclosure, anti-dependence design and escalation to human help, closely match what California and New York already require of companion operators. As that view spreads beyond the US and Europe, operators should treat those features as the expected standard everywhere, not a local compliance task.

The study's finding on social networks also deserves attention. It contradicts the common story that only isolated people get attached, which suggests that safeguards built only around lonely users would miss part of the risk.