Meta is killing off its AI-powered Instagram and Facebook profiles
Meta’s decision to shut down its AI-powered Instagram and Facebook personas, including a “proud Black queer momma” character, has triggered criticism of both the product concept and its execution. Commenters highlight risks of LLM “hallucinations” about sensitive topics like race and creator identity, worry about corporate stereotyping and cultural commodification, and question why users should be nudged into parasocial relationships with bots at all. Many see the project as emblematic of a broader trend: large platforms deploying half-baked AI features to drive engagement and ad revenue, even at the cost of trust and meaningful human interaction.
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.