A few years ago, IT giants got seriously involved in the development of chatbots with pseudointelligence, as this promised a lot of advantages. Of those that lie on the surface: reduction of personnel from living people for customer service, improvement of the quality of services, new marketing schemes. In practice, they received a drop in indicators in all respects and a somewhat undermined reputation. What went wrong?
The average people were so fascinated by the powerful ads of Facebook and Google that they seriously believed - they would be allowed to communicate with real artificial intelligence. Who knows everything, at least according to his profile, and can give them answers to difficult questions, really help. In fact, the Facebook chatbot in 70% of cases, after a couple of phrases, apologized and asked to contact a human consultant.
However, despite the failures of almost all ambitious projects in this area, the market for chatbots, intelligent assistants, is gaining momentum and will reach $ 1 billion by 2025. It is being developed by startups like the Spanish Landbot, whose authors have figured out how to close the catastrophic gap between user expectations. and reality, to minimize risks. You just do not need to promise support for "live conversation" and "understanding of the interlocutor."
Successful chatbots deliberately act "stupid", do not ask unnecessary questions, give stereotyped answers, often just voice lines from user instructions. And people evaluate the service according to the result, so when communicating with a primitive chatbot they do not experience disappointment from failure - what to take from it? And they opt for simple systems over complex AIs with animated chat that promise but don't deliver. Proof of this is the growth of orders for Landbot and its analogues, which are already successfully used in 50 countries.