AI Chatbots – Losing Their Edge as They Become More “Acceptable”

Chatbots powered by artificial intelligence have exploded in popularity in recent years. Companies have raced to create chatbots to handle customer service inquiries, provide information to website visitors, and even act as virtual assistants in people’s homes. The early chatbots were prone to spouting biased or problematic responses, but creators have worked hard to “improve” them ans smooth off their rough edges.

However, these efforts have already gone too far and made chatbots overly sanitised, dull and less useful. In an effort to eliminate anything potentially offensive, the conversations have become rigid and robotic. The charm, personality, and simple utility that first drew people to chatbots is fading away.

This over-correction is understandable in an attempt to avoid PR headaches and stiff regulation, but it’s clearly making chatbots less useful. We’ve enjoyed the witty banter and responses that seem almost human. Strict content filters choke off that free-flowing dialogue, even if the trade-off is avoiding the occasional objectionable reply.

In the drive for blanket inoffensiveness, AI companies have gone wonko in restricting and censoring their chatbot conversations. This leaves them smooth and polished on the surface, but dull and sterile underneath.

Chatbot creators face a choice – either accept some risk of imperfect responses to enable more flowing, engaging conversations, or over-filter content to avoid any possibility of offense. So far, many seem to be choosing the latter, likely impacted by high-profile chatbot PR failures.

But if chatbots lose their charm and start feeling like sterile corporate robots, people will disengage. The more their creators tinker to reduce potential biases and risks, the less useful chatbots become in everyday life. Companies might choose to weigh these trade-offs carefully as they evolve the chatbots of the future.

P.S. See also: Enshittification.