Sign up for the Slatest to get the most insightful analysis, criticism, and advice out there, delivered to your inbox daily.

Not too long ago, I was messaging a friend on Instagram about hangout plans. The photo app’s DMs are a common place for us to chat, since we frequently share memes with each other there and it usually makes sense to consolidate those communications in one spot. At least, it did until very recently, when we fully beheld the wretched manifestations of Meta’s desperate A.I. push.

If you’re still a regular Instagrammer who hasn’t quit in protest of Meta CEO Mark Zuckerberg’s Trumpy turn, you’ve doubtless encountered various artificial intelligence integrations in your everyday uses of the app. The primary means by which Meta has encouraged adoption of its A.I. can be generously described as digital force-feeding. Symbolized by a little blue circle, Meta AI is automatically embedded in your Facebook and Instagram search bars. You have no choice as to whether the apps will conduct a proper search for you or just choose to “ask” LLaMA instead. Threads’ “trending topics” are also automatically A.I.-generated.

For a while, you could hide the A.I. button in WhatsApp, but that option is now no longer available, much to international users’ chagrin. (It’s also part of WhatsApp’s search.) There’s also Meta AI in those Ray-Ban sunglasses—and this is all outside of that solo app that’s reportedly coming sometime this year.

Instagram, however, provides the most instructive examples of Meta AI’s forced ubiquity. On top of all the A.I. “characters” and user replicas—which appear to be getting rather uncomfortable and sexual these days—there are rollouts going on for A.I.-suggested comments, image and video editing, and even a “task assignment” option in your DMs.

I was already familiar with the sparkling white pencil in the blue circle that pops up when you’re typing a message and allows you to “write” using Meta AI—and I take care to avoid tapping it at all costs. I was also familiar with the encircled plus sign that appears at the end of the text bar before you actually type anything, which you can tap for a small options menu: “imagine” a generated photo, or just do something with “Meta AI”—nothing is specified. But even though I’d long steered away from these unholy buttons, I wasn’t in the clear.

At one point, when I was about to respond to a DM, a random prompt appeared beneath the message prompting me to “Tap and imagine with AI. Try it.” Because I had not pressed that plus button, I’d assumed that the supposedly all-knowing Instagram would have realized I did not wish, in fact, to tap and imagine with A.I. Silly me! I thought I’d at least parlay this moment for humor, so I took a screenshot of that text. Then I began typing my gentlemanly response to the prompt (“fuck no I won’t ‘tap and hold’ ”) before remembering that, alas, when you start typing in Insta chat, you cannot include an image in that same message, because all the attachment options are immediately replaced with just that white pencil. So I decided to highlight that text to cut and paste later, and lo and behold:

Somehow it wasn’t clear that I had not wished to tap and hold; now, I was being prompted once again, to “Write With AI,” even with the white pencil in plain view! Unamused, I then took a screenshot of that moment to send to my friend, and to ensure the picture went through, I tapped on the image after it sent. I was then given an option to “Edit with AI” at the bottom of said screenshot, which was already sent.

I tried to inform my friend of the rage-inducing incident and started typing, “even just tapping on the imag”—welp, I didn’t get to finish that word, much less the phrase. The Instagram chat automatically caught that last word and invited me to select its “/imagine” prompt, so that I could “Describe an image for Meta AI to generate.” No! Absolutely not! No!!!

So instead, I took a screenshot of the original “Edit with AI” prompt as it showed up in the chat. I edited that image in my phone’s gallery—not by using A.I., just by using a draw tool to circle the prompt. Once I sent the edited screenshot in the chat and tapped on it for a closer view, the app again prompted me to edit the “Edit with AI” photo with A.I.

The real kicker came near the end of the convo, when I accidentally pressed a backslash and had to confront the whole suite of A.I. prompts: “/silent,” to send a message without a notification to my friend with whom I was making plans, and then a general Meta AI option, blue ring and all, available for me to “Ask questions.”

I gotta say: This all sucks! Even if I were less of an A.I. skeptic, and even if I were enthusiastic about Meta AI in particular, I would still be very annoyed by the repeated, nonstop prompting within a small DM window that’s already hard to use. Options that would actually be of use to me in regular messaging (non-A.I.-image attachments, saving some draft text to my phone’s clipboard) are increasingly getting pushed out for … for what, precisely? Ways to “imagine” stuff that’s not likely at all to be pertinent to what I wish to convey?

The worst part of all this is the realization that it’s inevitably going to become shittier. Meta is pouring billions upon billions of dollars into its A.I. operations, and while it’s still a wildly valuable company, it can’t afford another metaverse-style implosion—especially at a moment when it’s already battling plenty of public scrutiny and adverse economic circumstances. Zuckerberg knows this and, in classic fashion, appears to be fudging the A.I.-usage metrics a bit. Through the past year, the A.I.-bullish CEO has touted constant user growth: reaching 400 million monthly active users by August, getting to “almost 500 million monthly activities” (emphasis mine) not even a month later, and earning “nearly” 600 million monthly active users by December. All impressive-sounding, and all tethered to some steep caveats: When reporting that 400 million figure in August, the company also acknowledged that Meta AI only had 40 million daily active users, and Meta’s vice president of product for generative A.I. admitted that the metrics include people who “run across it while scrolling through Facebook.” And there is, of course, the blurred classification between “activities” versus “active users”—does accidentally tapping that weird blue ring mean you are suddenly an active user, because you inadvertently engaged in a Meta AI activity?

There are already signs that the A.I. efforts are cracking a bit: The company’s head of A.I. research announced her departure at the start of this month, an event that prompted several Meta employees to anonymously tell Fortune that Meta’s specialized A.I. labs are “dying a slow death.” (A characterization that the firm denies.) A few days after that, Meta launched an update to its large language model and faced accusations that it had artificially boosted the quality of the LLaMA models’ outputs, forcing the company to issue a rejection of those rumors that didn’t really convince anybody.

So obviously, as such changes and critiques overwhelm its systems, Meta’s going to keep inserting A.I. wherever it can, making the buttons and prompts inescapable and collecting your accidental taps as indicators of success. As Big Tech has known for a long time, sometimes you have to massage the numbers in order to get what you want. Even if it means worsening the user experience of their apps for simple tasks like making plans.

Get the best of news and politics

Sign up for Slate’s evening newsletter.

Please enable javascript to sign up for newsletters.

Email address:

Send me updates about Slate special offers.

By signing up, you agree to our
Privacy
Policy

and
Terms.

Sign Up