Meta apparently plans proactive AI chatbots that address users without request

Chatbots based on artificial intelligence (AI) are now used by countless companies to answer questions from human users. However, Meta Platforms apparently wants to have AI chatbots become active themselves soon to proactively address the users of its services. This is according to internal documents from an AI company working for Meta. These actions are intended to bind users to Meta's platforms, which generates advertising revenue.

It would not be the first experiment of this kind by the Facebook group, as Meta had 28 AI bots operating on its platforms for a year and a half. These bots represented different characters and were able to chat with other users, as well as independently post images to encourage interaction. However, at the beginning of this year, Meta switched off the AI characters following controversy. Some bots exhibited unusual behavior and Meta spoke of a bug. This was because some people were unable to block the bots.

Proactive AI bots for Messenger, WhatsApp and Instagram

But Meta Platforms is probably not giving up on this approach. This is reported by Business Insider, which cites guidelines from the company Alignerr. The company specializes in AI training and is working with Meta on training customizable chatbots that can contact people without prompting and ask questions about previous conversations. For example, AI bots can contact users on Messenger, WhatsApp or on Instagram about films and offer appropriate recommendations.

Meta Platforms confirmed to TechCrunch that it is testing requests via AI. However, the AI bots will only follow up within 14 days after users have started a conversation and asked the AI at least five questions during this period. However, Meta promises that the chatbot will not send any further messages if there is no response to the first question. However, this also shows that the AI can remember information about users – at least these conversations.

No comment on the commercialization of AI

“So you can continue to explore topics of interest and engage in more meaningful conversations with the AIs in our apps,” a Meta spokesperson commented. The aim is therefore customer loyalty, as more users on the company's own platforms lead to more advertisements. Meta Platforms did not want to comment on whether the AI chatbots themselves are also commercialized and receive advertising or provide sponsored answers. It also remains unclear whether AI chatbots of this type will also be integrated into Meta's virtual reality application “Horizon Worlds” at some point.

However, particularly active AI chatbots also entail dangers. For example, a teenager in the USA recently took his life after regular and intensive conversations with a chatbot from Character.AI, whereupon his relatives went to court. When asked about this, Meta Platforms simply refers to Instagram's terms of use for AI chats. It states: “Answers from AIs may be inaccurate or inappropriate. You should therefore not use them as a basis for important decisions.”

(fds)