Four-wheeler driving test candidates at three RTOs in Ahmedabad and one in Gandhinagar are failing at a 94% rate since the offices switched to AI-powered test tracks, up from 55% under the older sensor-based system, according to DeshGujarat. Two-wheeler failures have also risen, to 25% from roughly 10%. At the Subhash Bridge RTO, where one of the new tracks was inaugurated, some early test batches recorded near-total failure among four-wheeler applicants, as reported by CarToq.
What changed
The older tracks relied on ground sensors and human examiners to judge whether a candidate stayed within lane markings, managed reverse parking, and handled an incline start without rolling back. The new AI system uses cameras and automated scoring to evaluate the same manoeuvres, reportedly with far less tolerance for the small deviations examiners might previously have overlooked. RTO officials told reporters the stricter reading is deliberate, meant to weed out drivers who scrape through on lenient human judgment and end up unsafe on real roads.
That reasoning has some basis: pass rates around 90-95% under the manual system, common at many Gujarat RTOs before this rollout, look generous next to global norms, where first-attempt failure is routine. But a sudden jump from roughly half of four-wheeler applicants passing to just 6% doing so is a large swing for a system introduced with, by officials’ own account, no public dry runs, no published scoring criteria, and no advance guidance to driving schools on what standard they were now expected to train candidates against.
What it means for Ahmedabad residents
For anyone applying for a licence in Ahmedabad or Gandhinagar right now, the practical effect is a longer, costlier road to a licence. Each failed attempt typically means a new appointment slot, a re-test fee, and in Ahmedabad’s case, appointment backlogs that were already stretched before the AI rollout. The Transport Department has reportedly begun planning longer daily testing windows and a second track at Subhash Bridge specifically to absorb the resulting overflow, an implicit admission that the current pass rate is creating a bottleneck rather than simply enforcing a higher standard.
There is also an equity dimension. Driving schools that have not yet adapted their training to the AI system’s exact tolerances put their students at a disadvantage against those trained on the newer standard, and officials have acknowledged they have not surveyed applicants or examined failure patterns to see whether the system is rejecting genuinely unsafe driving or simply penalising minor technical infractions, such as a slightly wide turn, that would not meaningfully affect road safety.
Gujarat is not alone in this transition. Automated, camera-based test tracks have been rolled out at RTOs elsewhere in India, including a similar system installed in Rajkot earlier this year, and have produced comparable failure-rate spikes in other states as well. The pattern suggests this is a predictable adjustment period for any AI rollout of this kind, not something specific to Ahmedabad’s implementation.
Outlook
Whether the higher bar sticks or gets recalibrated will depend on what the Transport Department does next. If failure rates stay this elevated for months without a published explanation of what the AI system is actually scoring, pressure will likely build from driving schools and applicants for either a review of the scoring thresholds or clearer guidance on what standard candidates need to meet. The planned second track and extended hours address the appointment backlog, but they do not address the underlying question of whether the near-90% jump in failures reflects a genuine safety gain or an unexplained recalibration that Ahmedabad’s licence applicants are absorbing without much say.
Sources: DeshGujarat, CarToq
This report was compiled and written with AI assistance from publicly reported sources, and reviewed for accuracy.