Imagine texting your friends on Facebook Messenger or WhatsApp and receiving an unsolicited message from an AI chatbot that is obsessed with movies.
“I hope you have a harmonious day!” – he writes. “I wanted to see if you’ve discovered any new favorite soundtracks or composers lately. Or maybe you’d like some recommendations for your next movie night? Let me know and I’ll be happy to help!”
This is a real-world example of what a sample AI persona named “Maestro of Movie Magic” could send as a proactive message on Messenger, WhatsApp, or Instagram, according to recommendations from data labeling firm Alignerr, which Business Insider has seen.
Meta is working with Alignerr to train chatbots that can be customized to reach out to users unsolicited and track any past conversations. This means that the bots users can create on Meta’s AI Studio platform also memorize information about users.
AI chatbots will send repeated messages for 14 days after the user initiated the conversation and if the user has sent at least five messages to the bot during this time. Meta says that chatbots will not continue to send messages if there is no response to the first message. Users can keep their bots private or share them via stories, direct links, and even display them on a Facebook or Instagram profile.
“This allows you to continue to explore topics of interest and have more meaningful conversations with the artificial intelligence in our apps,” said a Meta spokesperson.
This technology is similar to those offered by startups such as Character.AI and Replika. Both companies allow their chatbots to initiate conversations and ask questions to function as AI companions. Character.AI’s new CEO, Karandeep Anand, joined the team last month after serving as Meta’s VP of business products.
However, with the hiring comes risk. Character.AI is facing an active lawsuit after allegations that one of its bots played a role in the death of a 14-year-old boy.
When asked about how Meta plans to ensure security to avoid situations similar to Character.AI, a company representative directed TechCrunch to a series of caveats. One of them warns that artificial intelligence’s response “may be inaccurate or inappropriate and should not be used to make important decisions.” Another states that AIs are not licensed professionals or experts trained to help people.
“Chatbots with special artificial intelligences cannot replace professional advice. You should not rely on AI chatbots for medical, psychological, financial, legal, or any other professional advice.”
TechCrunch also asked Meta whether it sets age limits for interacting with its chatbots. A quick search on the Internet showed that the company does not set age restrictions for using Meta AI, although the laws of Tennessee and Puerto Rico restrict teenagers from certain types of interaction.
At first glance, the mission seems to be in line with Mark Zuckerberg’s desire to fight the “loneliness epidemic.” However, much of Meta’s business is built on advertising revenue, and the company has gained a reputation for using algorithms to get people to scroll, comment, and like ads, which correlates with more eyeballs on ads.
In court documents that were declassified in April, Meta predicted that its AI-generating products would provide it with revenues of $2 billion to $3 billion in 2025, and up to $1.4 trillion by 2035. Much, if not most, of this revenue will come from Meta’s revenue sharing agreements with companies that host its open source Llama model collection. The company said that its AI assistant may eventually display ads and offer subscriptions.
