Why not stay and talk with a bot instead of ever leaving its apps? Mark Zuckerberg revealed an update to Meta’s AI models this past week, saying that they were now among the best in the business in some areas. He described the goals his organization has set for pursuing artificial general intelligence, or AGI, and offered some more precise projections: “By the end of the decade, I believe a lot of people will use smart glasses like the ones we’re developing with Ray-Ban Meta to communicate with AIs frequently throughout the day.”
Perhaps so! However, the business has other plans for the time being. Facebook, Instagram, WhatsApp, Messenger, and other of Meta’s most popular apps are among those where the chatbot is being implemented. The chatbot may appear to users as suggestions in social feeds, comments on Facebook posts, or voice opinions when tagged in group chats. It allows for direct chat, similar to ChatGPT. Similar to how Microsoft and Google have integrated AI assistants into their productivity tools, Meta has incorporated helpers into many social scenarios. It can create images and compose messages on your behalf. Meta will find out quite quickly whether or not people utilize them in these circumstances, which will be genuinely intriguing to see.
It has been said that this move is both cunning and desperate. Is Meta attempting to catch up, investing heavily in a passing trend, and pushing unfinished technology onto its users? Or is Meta, with its functional approach, relevant hardware business, and greater user base than anyone else, now the de facto leader in AI? Claims like these, like AI models themselves, are difficult to measure since everyone involved in AI is racing toward an unclear goal that they—or at least their investors—think would bring immense wealth.
Similar to ChatGPT, you can ask it anything and get a response that is synthesised. Unlike some other chatbots, Meta’s AI will frequently return something that looks like search results, presented as a summary with footnoted links sourced from the internet. This is similar to the kinds of results you might get from an AI-powered search engine like Perplexity or Google’s Search Generative Experience. When it functions properly, the goal is rather obvious: these services aim to lessen the necessity of ever leaving Facebook or Instagram, rather than offering users additional activities within the platform. You can use Meta’s search bar to get the solution to your question without leaving Instagram to look it up on Google or to browse the web for a while.
Meta isn’t just trying to maximize engagement here, though that’s undoubtedly a factor. At this size, Meta is probably losing a significant amount of money on the deployment of this kind of AI, which is costly to train and requires a lot of processing power to operate, which is why OpenAI charges customers for comparable capabilities. It’s also a blueprint for an imagined future in which the web, or publicly available websites that exist outside of walled gardens like Meta’s, is smaller, less navigable, and less essential to the majority of people’s online lives. These days, smartphone users utilize web browsers inside apps and switch between apps and online browsers. Apps that don’t truly communicate with one another otherwise lack connective tissue, and most apps at least partially refer to the web as a common resource. Here, Meta provides a glimpse into a future when the internet is essentially a source of summaries and references, rather than something you browse yourself. Instead, it’s a collection of data that a machine searches through on your behalf.
This wouldn’t be good for the web or the different parties that currently contribute to it; in fact, the rapacious approach that AI firms take to any and all sources of data that are currently available and available could make it harder to find such data and less likely for those who create it to produce it, or at least share it (since Meta’s AI is currently based on results from Google and Bing). Let’s not get ahead of ourselves: when I initially used Instagram’s new function, I typed in “New York,” which brought me a list of profiles and a few suggested searches, one of which was, oddly enough, “New York fries near me.”
It’s probably a good thing that my true desire was not for fries. In another instance, Meta’s AI is advising parents on Facebook, posing as the parent of a child who attends a public school in New York City who is both gifted and impaired.
Perhaps Zuckerberg is correct when he predicts that by the end of the decade, we will be conversing with AIs in our Ray-Bans on a daily basis. Currently, though, Meta expects us to have such conversations—regardless of whether we find the responses necessary, agreeable, or understandable. We’re forced to test both the AI and ourselves.
