How to disable or avoid intrusive AI

Original Article ↗ Hacker News Discussion ↗

Intrusive AI features being embedded across operating systems, search engines, productivity tools, and car infotainment systems are prompting users to look for ways to disable or avoid them. Commenters trade practical tactics—like browser filters, alternative apps, de-Googled phones, and Linux desktops—while arguing that the core issue is loss of user control as companies bundle AI into essential functionality and remove non‑AI fallbacks. Some see value in AI when used intentionally (e.g., standalone chatbots or specific tools), but there is broad frustration with being forced into data‑hungry assistants and opaque automation they never asked for.

Overall sentiment on AI vs. intrusive integrations

  • Many distinguish between choosing to use AI (e.g., ChatGPT/Claude) and having “AI everywhere” forced into existing tools.
  • Users object to animated/copilot-style buttons, AI summaries, and “rewrite/summarize” features that clutter workflows and remove or gate existing functionality.
  • Others push back on “nobody wants this,” citing widespread paid use of AI chat tools, while agreeing that embedded AI in every app is often unwanted.

Business and market incentives

  • Some see AI features as investor bait: added to signal innovation rather than meet user demand, and ideally not heavily used because they’re expensive to run.
  • Others argue it’s partly defensive: enterprise vendors fear becoming “dumb pipes” to external LLMs and want their own AI to increase lock‑in and pricing power.
  • Comparisons are drawn to previous tech fads (“engagement,” blockchain), with concern that large public companies still must show returns to shareholders.

Disabling and avoiding intrusive AI

  • Common tactics include:
    • Browser filters: uBlock Origin rules (e.g., blocking Google AI summary CSS, EasyList annoyances “AI widgets”), element hiders, specific extensions (e.g., to kill Amazon’s Alexa shopping panel).
    • Search tweaks: udm=14 on Google queries, alternative engines (Kagi, DuckDuckGo’s no‑AI endpoint, Qwant, Ecosia).
    • OS/app choices: Linux desktops, privacy‑focused or AI‑stripped browsers (LibreWolf, Waterfox), LibreOffice, VSCodium.
  • Some prefer to abandon products that push AI too hard rather than fight the UI.

CarPlay, Siri, and in‑car systems

  • Several complain that CarPlay and some car UIs now gate basic functions (search, messaging, maps input) behind Siri or voice assistants.
  • Supporters argue voice-first design is meant to reduce distraction while driving and is better than legacy in‑car infotainment systems.
  • Others dislike any mandatory assistant, lit screens, or alerts, preferring “appliance‑like” cars with minimal electronics.

Responsibility and ethics

  • There’s debate over whether to blame “devs” or “the business” for intrusive or harmful AI deployments.
  • One side emphasizes that executives and product managers set incentives and policies; another argues individual developers are still complicit in implementing harmful systems.

Broader concerns

  • Threads note surveillance risks, AI being tied to non-AI features (e.g., Gmail, Apple Intelligence), and the burden on users to constantly opt out, block, or reconfigure.