Imagine that you send messages in the Messenger application on Facebook or WhatsApp, and you get an unmistakable message from Chatbot AI, which is obsessed with movies.
“I hope you have a harmonious day!” I'm writing. “I wanted to check in and check if you have recently discovered any new favorite soundtracks or composers. Or maybe you would like a few recommendations for the next movie evening? Let me know and I will gladly help!”
This is a real example of what the AI Persona sample called “Maestro of Movie Magic” can send as a proactive message about Messenger, WhatsApp or Instagram, in accordance with the guidelines from the ALIGNERR data labeling company Business Insider.
He learned leaking documents on sale that Meta cooperates with Alignerr in order to train configurable chatbots to reach users that are free of released and following any previous conversations. This means that the bots that users can create on the Studio AI META platform also remembers information about users.
The meta confirmed that it is testing AIS contacts to TechCrunch.
Ai Chatbots sent followed only within 14 days after initiating the conversation and if the user sent at least five messages to the bot in this frame. The meta claims that chatbots will not send messages if there is no response to the first observation. Users can keep their private bots or share them via history, direct links, and even display them on Facebook or Instagram.
“This allows you to continue to study interesting topics and get involved in more significant conversations with AI in our applications,” said Meta spokesman.
The technology is similar to the one offered by AI startups, such as character.ai and replica. Both companies allow their chatbots to initiate conversations and ask questions to function as comrades AI. The new general director. AI, Karandeep Anand, joined the team last month after he was the Vice President for Meta Business Products.
But there is a risk with commitment. Character.
When asked how Meta plans to deal with safety to avoid situations such as a character. A series of reservations. One of them warns that AI's answer “may be inaccurate or inappropriate and should not be used to make important decisions.” Another says that AI are not licensed specialists or experts trained to help people.
“Chat with non -standard AIS cannot replace professional advice. You should not rely on AI chats on medicine, psychological, financial, legal or any other professional advice.”
TechCrunch also asked the finish line if he imposes an age limit for chatbot involvement. Short internet diving does not invent any age restrictions on them imposed by the company Tennessee and Puerto Rico Limit teenagers from some commitment.
On the surface, the mission is in line with Marek Zuckerberg's pursuit to combat the “epidemic of loneliness”. However, most of the meta activities are based on advertising revenues, and the company has gained a reputation for the use of algorithms to rewind people, commenting and preferences, which correlates with larger eyes at advertising.
In court documents, which were unpaid in April, Meta predicted that his generative AI products would provide him from $ 2 billion to $ 3 billion in 2025, and up to $ 1.4 trillion by 2035. Many, if not most, it would result from the contracts for the division of meta revenues with companies that organize an open collection of Llam models. The company said that her assistant AI can ultimately show ads and offer the subscription option.
The metal refused to comment on TechCrunch's questions about how it plans to commercialize his AI chatbots, regardless of whether it plans to turn on advertising or sponsored answers, and whether the long -term strategy of the company with AI companions involves integration with Horizon, the virtual community of the meta.
Do you have a sensitive clue or confidential documents? We report the internal functioning of the AI industry – from companies shaping their future to people affected by their decisions. Contact Rebecca Bellan at rebecca.bellan@techcrunch.com and Maxwell Zeff at maxwell.zeff@techcrunch.com. To get safe communication, you can contact us via Signal at @Rebeccabellan.491 and @mzeff.88.
