A notification for you, Apple: There is no husband
Apple’s new on-device AI features, such as message summarization in iOS, are drawing criticism for hallucinations and tone-deaf behavior, like inventing a “husband” where none exists and making personal communication feel dehumanized. Commenters question whether anyone really wants AI-altered notifications, arguing that unreliable summaries add cognitive load, amplify spam, and are being pushed too aggressively by Apple for marketing and shareholder reasons. The thread broadens into skepticism about large language models’ actual intelligence, concerns about Apple’s internal bureaucracy and product quality drift, and speculation that “AI‑free” products and services may become a selling point.
LLM Notification Summaries & Usefulness
- Many see Apple’s text-notification summaries as low-quality and often wrong (e.g., hallucinating a “husband”), eroding trust.
- Several users say they don’t want summaries of short, personal messages at all; they prefer verbatim texts from people they care about.
- Some find summaries genuinely useful:
- For partners who send many short messages, summaries help triage what needs attention now vs “just chat.”
- For long group chats or in CarPlay, they give quick context without reading the entire thread.
- Others turned the feature off due to inaccuracy, or back on purely for entertainment value.
- Concern that summarization can make spam or scams sound more credible.
Notification Overload vs Better UX
- Many argue AI is a band-aid over poor notification design.
- Suggestions: merge rapid-fire messages into a single notification, make group chats silent by default, or let AI only gate non-urgent messages until later.
- Some feel summarizing personal messages is dehumanizing; summarizing work email/Slack is seen as fine.
Perception of LLMs & “AI” Hype
- Several posters view LLM “smartness” as overrated and similar to bad targeted ads: occasionally impressive but mostly off-target.
- Others say summarization and coding help are among the most practically useful LLM features.
- Debate over whether there is a true “killer app” yet; some claim chatbots already are, others strongly disagree.
Apple’s AI Push & Product Direction
- Frustration that Apple Intelligence and other AI features are aggressively promoted or default-on, with “Not now” instead of a clear “No.”
- Some users feel Apple has shifted from user-centric design to rent extraction and metric-driven feature rollouts (e.g., opt-out defaults used as “adoption” stats).
- A few say recent macOS/iOS releases and AI integration make them consider leaving the ecosystem.
Apple Culture, Efficiency & Quality
- Multiple ex-employees describe heavy bureaucracy, redundant teams chasing the same “hot” topics, and “rest-and-vest” roles where underperformers persist.
- Others counter that such dysfunction exists at all large companies and Apple still ships a lot of integrated software.
- Some see “pets” and politics protecting weak performers and tying this to visible product missteps (e.g., buggy features, controversial design choices).