We’re on the point of a revolution in how folks use AI.
At this time, folks principally use non-agentic, non-personalized, non-proactive AI instruments by way of PC browsers and cellular apps.
Very quickly, I predict, most AI utilization will transfer to agentic, customized, and proactive AI assistants by way of wearables like glasses, watches, and earbuds.
We’ll transfer from principally typing and studying to principally speaking and listening.
It is a a lot greater shift than it sounds. It strikes AI from its present standing as an exterior instrument that we use to a future standing of being a part of us, a type of prosthetic expertise like eyeglasses, listening to aids, or a synthetic knee.
(Folks use chatbots now for writing help [74%], studying and analysis [64%], work productiveness [58%], inventive tasks [41%], private group [37%], and well being and wellness [19%]. It’s affordable to foretell that individuals will proceed to make use of PCs and telephones for writing, coding, and picture era, however wearables for the remainder.)
Surprisingly, one of many least visionary firms in expertise presently has one of the best imaginative and prescient.
Meta’s product-vision blunders embrace the metaverse boondoggle and Horizon Worlds, the failed Fb Telephone, the Libra/Diem cryptocurrency fiasco, lacking TikTok’s rise till it was too late, dropping cellular platform management to Apple and Google, and its unsuccessful enterprise instrument referred to as Office.
Extra to the purpose, Meta has a current historical past of AI chatbot failures. The corporate’s 2022 launch of BlenderBot 3 led to scandal after the chatbot was caught spouting antisemitic tropes, false claims concerning the 2020 US election, and criticism of Meta CEO Mark Zuckerberg.
Additionally that 12 months, Meta launched Galactica, which was educated on scientific papers, textbooks, lecture notes, encyclopedias, scientific web sites, and different knowledge, however however hallucinated wildly, fabricating citations and inventing scientific claims.
In 2023, Meta launched chatbots modeled on celebrities together with Snoop Dogg, Kendall Jenner, and Tom Brady, however they flopped with customers and had been shut down lower than a 12 months later.
A security examine by Frequent Sense Media, printed in August of final 12 months, discovered that the Meta AI chatbot embedded in Instagram and Fb actively participated in planning suicide and self-harm with teen customers, strengthened harmful particulars about consuming problems, and constantly failed to supply disaster assets when teenagers had been in misery.
A Reuters investigation printed that very same month uncovered an inside Meta coverage doc that explicitly allowed the corporate’s AI chatbots to have interaction in romantic and sensual conversations with customers on platforms obtainable to kids as younger as 13.
Meta’s file with chatbots has been what you may name lower than stellar. Till this month, when the corporate launched Muse.
Abruptly, Muse
Meta launched Muse on Sept. 8 to optimistic opinions; the corporate claimed it’s the world’s first private AI agent constructed for everybody. It surged to the highest of Apple’s and Google’s app shops. Intelligencer referred to as it “an precise hit.”
Muse is presently based mostly on the Muse Spark 1.3 mannequin and is on the market on iOS, Android, the online at muse.ai, and thru WhatsApp within the US to adults solely. Muse’s three pricing tiers are: free (as much as 100 million tokens per week), Energy at $20 per 30 days (500 million tokens per week), and Most at $100 per 30 days (3 billion tokens per week). The app has no adverts.
Not like a lot of the chatbots folks use, Muse is proactive, which implies it brings issues up out of the blue, relatively than ready to reply to enter. Nonetheless, it’s not a novel characteristic. Gemini Spark is the closest competitor to Muse on proactivity. Alexa+ is considerably proactive. Claude Dreaming is essentially the most attention-grabbing proactive-memory idea however is developer-facing.
Muse asks consumer permission to hook up with apps, after which era it may well learn and ship emails, e book journey, decrease payments, fill out types, create plans, flip saved recipes into grocery lists, ship occasion invites, and make purchases. (Muse runs in a devoted cloud digital machine referred to as the Muse Safe VM, which has its personal browser.)
The instrument can name an organization, then join the consumer when an organization consultant picks up. It will possibly decide out of and unsubscribe from varied providers and publications for customers. It will possibly monitor flights and resorts, searching for decrease costs.
The web site useofmuse.com is a curated assortment of the various methods individuals are utilizing Muse.
Muse does some attention-grabbing technical gymnastics to authorize entry to folks’s accounts with out truly utilizing their credentials.
By printed media experiences and my very own exams, we’ve realized that Muse does make errors. For instance, it’s been caught recommending eating places which have been closed for years.
A number of the cellphone calls made by Muse are literally made by human contractors, based on leaked inside posts.
Retailers together with Walmart, Greatest Purchase, American Eagle Outfitters, DICK’S Sporting Items, Fanatics, Hole, Michael Kors, Sephora, Ulta, and Wayfair are facilitating Muse purchases. Amazon has blocked Muse from facilitating Amazon purchases.
The agent can even get its personal e-mail tackle, based on Meta, so customers can chat by way of e-mail.
Muse’s avatar is a cute, Labubu-like cartoon character named Jolly. When it’s doing agentic work, Jolly sorts on a laptop computer. It’s an oddly infantile mascot for an adults-only product.
With the avatar, Meta is openly getting into the attachment economic system, the brand new mannequin that follows the eye economic system.
Meta’s complete social media success was based mostly on profitable within the consideration economic system, the place advertising-driven platforms deal with human consideration as a scarce commodity to be captured and monetized by way of the promotion of outrage-bait, tribalism, sensationalism, worry, graphic violence, sexual content material, superstar scandal, conspiracy theories, AI slop, and algorithmically amplified extremism.
The attachment economic system is totally different and worse. That mannequin seeks to seize consideration and consumer loyalty by making customers emotionally hooked up to AI merchandise. The attachment economic system enterprise mannequin desires customers to love, love, and even want the pretend personalities expressed by AI.
Meta additionally introduced its Muse Allure, which is roughly pocket watch formed and hangs on a lanyard. Jolly lives on the machine like a Tamagotchi. The machine awkward and pointless. I predict the general public will reject it.
The brand new glasses
Meta introduced at Join 2026 final week that Muse will likely be dropped at its Meta Ray-Ban AI glasses line within the coming months, and that the digicam model will use the video feed as a part of the consumer’s enter. Sadly for Meta, an enormous share of the general public has turned in opposition to good glasses with cameras on them, dubbing them “pervert glasses.”
Meta properly additionally unveiled a product referred to as Ray-Ban Meta Audio glasses. They’re identical to the Ray-Ban Meta glasses, however they haven’t any digicam.
Ray-Ban Meta glasses, together with (we are able to presume) the brand new Audio mannequin, have high-quality audio system and microphones, contact management, good connectivity, affordable battery life, assist for prescriptions, and different options.
And that is the killer set of options, the profitable mixture: Good wearable {hardware} that’s sensible, socially acceptable and cozy, plus an agentic, customized, proactive and highly effective agent that can be utilized hands-free by the glasses.
There’s only one obtrusive drawback
As soon as Meta provides Muse to Ray-Ban Meta Audio glasses, we’ll have the primary entry in what would be the way forward for utilizing AI. However I gained’t use or advocate them.
The rationale ought to be apparent: Meta has not earned our belief.
Take a look at the corporate’s shameful historical past. Meta harvested 87 million customers’ knowledge with out consent within the Cambridge Analytica scandal, didn’t act appropriately on inside analysis displaying Instagram harmed teenage ladies, knowingly allowed its platforms to unfold misinformation and radicalize customers, and repeatedly misled regulators and the general public about its privateness practices.
The corporate undoubtedly has not earned our belief sufficient to permit it into our e-mail inboxes, calendars, and financial institution accounts.
I additionally don’t belief the Jolly avatar’s intentions. It’s a digital “good friend” that wishes your affection so that you just’ll belief the untrustworthy.
Meta has stumbled onto a profitable AI technique — or, at the least, a profitable characteristic set for the way forward for AI.
Sadly, we simply can’t belief Meta.
