Some AI experts say fully autonomous cars may be further away than expected and that it may be years before self-driving systems can reliably avoid accidents
Russell Brandom / The Verge :
Context & Ripple Effects
This 2018 report lands mid-arc in a debate that had been running since at least the [[a:867184|2016 argument over whether incremental semi-autonomous features beat waiting for full automation]]. Russell Brandom's reporting captures the moment experts began publicly walking back the near-term autonomy timelines the industry had been marketing.
The skepticism proved durable rather than dated: by 2021 the industry was formally resetting expectations after its 'commonplace by 2021' predictions missed, and analysts were arguing steady ADAS improvement beats a moonshot Level 4 bet. By late 2024, the same critique resurfaced against Tesla specifically, with experts saying its imitation-learning approach still needs AI breakthroughs that may not arrive soon.
First-order effects
- Aggressive developers like Uber and Tesla face a widening gap between promised timelines and expert-assessed technical reality — a gap that, as later coverage argued, risks alienating a public already wary of self-driving vehicles.
Second-order effects
- The industry's center of gravity shifts toward driver-assistance as the scalable product: the Financial Times' reporting on experts favoring incremental ADAS over Level 4 moonshots is the direct commercial response to this expert consensus.
- Deployment bottlenecks beyond the AI itself come to the fore — Bloomberg's reporting on how police pull over autonomous cars shows reliability is only one of several unsolved prerequisites for widespread driverless operation.
Third-order effects
- If the pattern holds, autonomy becomes a decades-long capability curve rather than a launch-date product, structurally favoring automakers who monetize assistance features along the way over those betting on a single full-self-driving breakthrough — exactly the critique later leveled at Tesla's approach.
The trend: Autonomous driving is settling into a long incremental-improvement trajectory, with each round of expert timeline corrections pushing the industry from full-autonomy promises toward driver-assistance products.