Hey @meta.ai is that true? Threads is testing a Grok-like AI feature

Trending 3 weeks ago
The Threads app connected a phone.

PJ McDonnell/Shutterstock

Meta has spent nan past mates years giving its self-titled AI chatbot salient placement successful its apps and now it's Threads' turn. The institution is starting to trial a caller feature that gives nan Meta AI chatbot Grok-like functionality connected Threads, pinch nan expertise to reply to posts pinch further "context."

To do this, Meta AI is getting an charismatic Threads relationship (@meta.ai) that users tin chat pinch alongside their different conversations successful nan app. The feature, which nan institution describes arsenic an "early beta" will beryllium rolling retired first to Threads users successful Malaysia, Saudi Arabia, Mexico, Argentina, and Singapore.

The premise sounds a batch for illustration nan original volition of Grok. Meta suggests that users could inquire questions for illustration "why are group talking astir nan World Cup this month?" and Meta AI will beryllium capable to publically reply. Users tin besides invoke Meta AI successful replies to different posts. This is simply a very communal usage lawsuit connected X, wherever seemingly each viral station has astatine slightest 1 salient reply asking Grok immoderate variety of "is this true???"

Of course, giving an AI chatbot truthful overmuch visibility connected a level for illustration Threads could airs immoderate of its ain challenges. On X, Grok has famously gone connected pro-Nazi rants, spewed sycophantic praise for Elon Musk and generated a staggering magnitude of kid maltreatment material. Meta has mostly been much cautious astir safeguards for its AI chatbot than X has pinch Grok, which was ever billed arsenic a benignant of AI edgelord. Still, it's reasonable to wonderment whether nan Threads characteristic could beryllium prone to nan aforesaid type of user-driven manipulation that X has blamed immoderate of Grok's much high-profile stumbles on.

Meta is, astatine least, offering a measurement for group to disregard nan caller Meta AI relationship connected Threads. The institution notes that users tin opt to shut up nan relationship and hide replies that look nether their ain posts.

The Threads trial was announced amid a broader push for nan recently revamped Meta AI, which is now powered by nan company's latest Muse Spark model. In summation to nan bot connected Threads, Meta is testing a caller measurement for group to chat pinch Meta AI astir nan happenings successful their group chats connected WhatsApp and different apps. Much for illustration connected Threads, these "side chats" arsenic nan institution is calling them, are meant to springiness group a measurement to query nan bot for further "context" based connected what's going connected successful nan group thread. But, dissimilar Threads, these chats return spot successful their ain Meta AI chat and are only visible to nan personification asking Meta AI to measurement in, not nan full group.

Elsewhere, nan caller Spark exemplary is besides powering a caller in-app type of Live AI successful nan Meta AI app. Previously only disposable via Meta's smart glasses, Live AI lets users ask questions astir their surroundings and get real-time answers astir immoderate their phone's camera is pointing at. (One connection of caution: Live AI was astatine nan halfway of a caller study that claimed human moderators were viewing recordings of these sessions, which sometimes featured friendly moments. So support successful mind there's a chance that quality eyes could spot immoderate you're asking Meta AI about.)

And, speaking of Meta's glasses, nan institution says that nan Spark exemplary is starting to rotation retired to its lineup of Ray-Ban and Oakley-branded glasses successful nan US and Canada, and will beryllium coming to its display-enabled frames this summer.

Muse Spark has been nan first awesome merchandise from Meta's "superintelligence" group arsenic nan institution tries to laic a caller instauration for its AI activity successful nan post-Llama era. CEO Mark Zuckerberg has besides hinted astir nan company's plans for OpenClaw-like AI agents that will tally connected immoderate type of nan model.

More
Source engadget.com
engadget.com