Tesla's self-driving tech ditched by 98 percent of customers that tried it

Original Article ↗ Hacker News Discussion ↗

What “self-driving” should mean

  • Many define “real” self-driving as:
    • The manufacturer assumes legal liability for crashes while the system is active.
    • The human need not constantly supervise or be ready to instantly take over.
  • By this standard, commenters argue Tesla does not have true self-driving and is unlikely to in the near term; current systems are seen as advanced driver assistance.

User experiences with FSD

  • Strongly mixed reports:
    • Negative: phantom braking, abrupt speed changes, poor city behavior, trouble with pedestrians, parked cars, on-ramps, roundabouts, and narrow residential streets; often described as more stressful than manual driving.
    • Positive: recent v12 versions seen by some as a big leap, “drives like a human,” excellent on highways, good at lane keeping, passing, exits, and long trips; some use it almost all the time.
  • Several analogies liken it to riding with a nervous teenage driver: capable, but unpredictable.

Pricing, adoption, and the “98% ditched” claim

  • FSD costs around $8k upfront or ~$100–$200/month.
  • Only ~2% of those given a 30‑day trial bought it; some call that a solid conversion rate for a high-ticket add‑on, others see it as disastrous given how central FSD is to Tesla’s story.
  • Many say the feature is interesting but not worth the price, especially when cars are already underwater in value.

Safety, legality, and accountability

  • Debate over legal frameworks: retail vehicles generally require a supervising driver, while some robotaxis operate without one in constrained areas.
  • Concerns about Tesla allegedly disabling Autopilot right before crashes and about long-running, overly optimistic FSD marketing.
  • References to regulatory and legal actions, including federal probes and class-action efforts, with arbitration clauses seen as a barrier.

Competition and alternatives

  • Waymo is widely viewed as leading in true driverless service, but only as a taxi, not an owned product.
  • Other EVs (VW, Hyundai, Kia, BMW, Polestar, etc.) are discussed; some find them inferior overall to Tesla on value, software, and charging, others think they now match or beat Tesla.

Human behavior and usability

  • Wheel “nags” and constant supervision make FSD feel pointless to some; planned removal of steering-wheel nudges raises fears of increased inattention.
  • Tension between strict speed-limit adherence and real-world traffic norms is unresolved and seen as a hard problem for autonomy.

Outlook on full autonomy

  • Optimists believe recent progress suggests robotaxis are only a few years away.
  • Skeptics argue driving is effectively AGI-level, neural nets handle edge cases poorly, and the decade-plus of missed timelines undermines credibility.