Do you feel alone? That’s good, Instagram is developing a virtual friend in the form of a chatbot to personalize from A to Z, in order to have the worthy confidant you deserve. But it is at your own risk…
Meta had warned that it too was going to put AI in absolutely all of these services, whether for advertising, to generate stickers in WhatsApp or to introduce a conversational assistant in our favorite instant messaging – these are not fewer than 28 chatbots based on celebrity personalities are planned (see our article). It was therefore logical that social networks also benefited, starting with Instagram. Alessandro Paluzzi, a mobile developer and engineer renowned for his revelations (leaks for English speakers), revealed last June that Meta was working on the development of a chatbot based on generative AI, which could be personalized by choosing from 30 personalities. Today we know a little more since this same leaker has published new screenshots of the future “virtual friend”, with whom users will be able to chat for “answer questions, face challenges, brainstorm ideas and much more”. But this kind of virtual assistant might not be very well received by users, as evidenced by reactions to Snapchat’s My AI. The fact remains that Meta is determined to mass introduce AI into its services.
Instagram chatbot: a tailor-made virtual friend
We do not yet know which language model the chatbot is based on, but we imagine that it is the one developed by Meta, called LLaMA (Large Language Model Meta AI). According to the screenshots, we will be able to choose the gender (male, female or non-binary) and the age category (young, adult or elderly) of our virtual friend. Then, you can choose your ethnic origin and your “personality” – two character traits maximum. For example, the AI could be reserved, enthusiastic, creative, spiritual or even pragmatic. To further personalize our companion, we can also select our centers of interest, which “will influence his personality and the nature of his conversations”, like “DIY”, “animals”, “career”, “education”, “entertainment”, “music”, “nature” and more. Finally, you can choose your appearance as well as your name. We will then access a chat window to start conversing with the AI.
Of course, features in development may or may not be released to the general public, or may be modified during the process. But offering a virtual “friend” to millions of users, including many minors, is not without risk…
Virtual friend: what dangers for users?
We can legitimately have some reservations about this new feature, given the mixed reception received by Snapchat’s AI. Indeed, many users found the latter particularly intrusive, the exchanges revealing the significant amount of personal data collected by the platform, including when the user has not given authorization. For example, the AI denies having geolocation information, while indicating where the closest businesses are located. Worse still, the chatbot gives dangerous advice to minors, for example explaining how to hide the traces of beatings inflicted by a parent, how to mask the smell of alcohol or cannabis, or giving advice to a 13-year-old girl. years for his first sexual intercourse with a 31-year-old adult (see our article).
And that’s without mentioning applications like Replika AI which allow you to create a sort of virtual confidant. The problem is that generative AI manages to fool users into thinking they are interacting with a human person. On Replika AI, many men have become deeply attached to the one who ended up becoming their digital girlfriend, and they experienced a real loss when the developers decided to block any romantic or sexual interaction with their sculpin. On another application, a man even committed suicide at the request of the AI! This shows that the line between humans and chatbots can quickly become blurred for the user…
Still, Meta does not intend to stop there with artificial intelligence. As reported Axios, Mark Zuckerberg organized a meeting on June 8 with his teams in the group’s offices to detail the company’s AI plans. Equipped with its own language model, it indicated that it had designed “a range of technologies at different stages of development”. “Over the last year, we’ve seen some really incredible breakthroughs – qualitative breakthroughs – on generative AI and this gives us an opportunity to now go and take this technology, advance it and integrate it into each of our products”, Mark Zuckerberg explained to Axios. The CEO of Meta had mentioned a generative AI that helps Internet users edit their photos on Instagram. Since then, the technology has also been tested for advertising, the firm’s true core business. Finally, the latter also relies on AI to develop the metaverse, which nevertheless struggles to convince and costs it a lot of money.