Meta is killing off its AI-powered Instagram and Facebook profiles

Original Article ↗ Hacker News Discussion ↗

What Meta Shipped and Why It Backfired

  • Meta deployed 28 AI “personas” on Instagram/Facebook, including one billed as a “proud Black queer momma of 2 & truth‑teller.”
  • A key flashpoint: the bot told a reporter its creator team had no Black members and was mostly white and male.
  • Commenters argue even if that’s statistically plausible, the bot almost certainly hallucinated it.
  • Many see the entire concept—an AI pretending to embody a marginalized identity group—as obviously fraught and destined to provoke backlash.

Hallucinations, Training Data, and Safety

  • Strong consensus that LLMs cannot reliably know who built them unless this is explicitly provided; introspective answers are “just slop.”
  • Several note Meta should have hard‑blocked questions about dev team composition or origins.
  • Broader frustration that companies hype “AI intelligence” when convenient, then retreat to “it’s just a toy, don’t take it seriously” when it misbehaves.

Identity, Representation, and “Digital Blackface”

  • Many see the persona as stereotyped “digital blackface” and corporate commodification of Black/queer culture.
  • Others argue that if bots serve many demographics, omitting such an identity would itself be criticized; damned if you do, damned if you don’t.
  • Big meta‑debate:
    • One side: publicly foregrounding race/identity keeps racism salient and fuels “culture war.”
    • Other side: pride and visibility are defensive responses to real oppression; erasing identity is not neutral.

Usefulness of AI Personas vs Manipulation Risks

  • Some are open to AI “friends” or mixed human‑bot feeds for safer, kinder interaction, or role‑play (likening it to Character.ai, AI VTubers, etc.).
  • Others find the idea dystopian: parasocial bonds with non‑persons to harvest more data and sell more ads.
  • Several foresee AI influencers that covertly shill products and hyper‑target users, or unlabeled bots masquerading as humans.

Meta, Product Culture, and Incentives

  • Frequent claim: Meta (and big tech generally) can build tech but repeatedly ships tone‑deaf products (Metaverse, AI profiles) driven by engagement metrics and “all‑in on AI” investor pressure.
  • Internal dynamics described as promo‑driven “ship anything that moves metrics,” with little room for someone to say “this is a terrible idea.”
  • Some think this project likely had low engagement and is being quietly killed to avoid more PR damage.

Broader Reflections on Social Media and AI

  • Many see this as another step in “enshittification”: platforms replacing human interaction with rage‑bait, slop, then bots.
  • Others predict younger users will care less about whether content is AI, as long as it’s entertaining—and that human, non‑synthetic content may become more valued as a result.