2026-06-18 — views

AV Passenger Experience — What It Feels Like to Ride Waymo and Tesla Robotaxi

Rider surveys and G-force data reveal how Waymo and Tesla robotaxi comfort directly drives adoption rates and repeat usage — the human dimension of physical AI.

Article 51 in the Physical AI Benchmark Series — The Consumer Dimension

The Physical AI Benchmark series has tracked operational metrics since Article 40: weekly ride counts, fleet size, disengagement rates per million miles, crash-adjusted safety ratios. These numbers measure what the machines do. This article measures what the humans experience. Passenger comfort is not a soft metric. It is the most direct driver of repeat usage rates, net promoter scores, and the pace at which driverless vehicles cross from early-adopter curiosity to mass-market utility. A robotaxi system that is technically impressive but physically uncomfortable will fail commercially just as surely as one that cannot navigate a roundabout. Understanding the ride experience — what users report, why AV braking feels different from human braking, what the psychology of the first driverless ride looks like, and how comfort converts into revenue — is the prerequisite for understanding whether the current operational leaders can hold their advantage at scale.

All figures marked (est.) are estimates based on published surveys, academic research, and industry reporting. They have not been independently verified and should be treated as directional rather than precise.

Section 1 — The Driverless Ride Experience: What Users Report

Published Waymo One rider surveys, independent academic studies of AV passenger acceptance, and online review communities provide the clearest available picture of what autonomous vehicle passengers actually experience. The table below compares reported dimensions across the two leading systems as of mid-2026.

Dimension	Waymo One (reported experience)	Tesla Robotaxi Austin (early supervised, est.)
First ride reaction	Majority report initial anxiety, then relaxation within 5–10 minutes (Waymo user research, est.)	Similar pattern — supervised safety driver presence reduces initial anxiety
Smoothness of braking	Early Waymo rides noted for “robotic” braking — slightly abrupt at lights; improving with each software generation	FSD v12/v13 noted for smoother acceleration and braking than earlier versions; occasional phantom braking persists
Routing decisions	Some users report AVs take unexpected routes or pause at edge-case intersections	FSD routes generally efficient; occasional overcautious lane changes noted
Safety perception	73% of Waymo riders report feeling as safe or safer than human-driven rides after completing a trip (Waymo survey, est.)	Supervised — safety perception anchored to the presence of a human safety driver
Comfort: no driver interaction	Highly valued — no small talk required, no tipping pressure, genuine privacy in the cabin	Not applicable until driverless mode reaches Austin
App experience	Clean Waymo One app; live ride tracking, remote door unlock, climate control (est.)	Tesla account integration for authorized account holders; early robotaxi UI in development (est.)
Accessibility	Full trip with zero driver interaction — ideal for users with social anxiety, hearing impairment, or non-English speakers	Requires interaction with safety driver during supervised phase

The most consistent finding across rider surveys is the arc from pre-ride anxiety to post-ride reassurance. First-time riders arrive nervous and leave converted at a rate that no marketing campaign could match. This experiential shift is the core mechanism behind Waymo’s word-of-mouth growth in Phoenix and San Francisco.

Section 2 — The Physics of AV Comfort: G-Force Profiles

Human drivers unconsciously smooth their driving based on passenger feedback — visual cues, social pressure, and years of muscle memory tuned by passengers shifting in their seats, grabbing door handles, or making eye contact. AV systems optimize primarily for safety and rule compliance rather than passenger comfort, unless comfort has been explicitly incorporated as a training objective. The result is a measurable difference in the G-force profile of common maneuvers.

Maneuver	Skilled human driver	Early AV systems (v1–v2 gen)	Modern AV (Waymo Gen 6 / FSD v13, est.)
Normal braking	Gradual, predictable deceleration; smoothly reduces G-force	Abrupt at yellow/red lights; “robotic” feel reported by most early riders	Significantly smoother; anticipatory braking using road context (est.)
Acceleration from stop	Smooth, proportional to surrounding traffic flow	Firm and efficient; adequate but not comfort-optimized	Closer to human feel; still slightly firm at full-torque EV starts (est.)
Lane change	Fluid; uses natural gaps and mirrors passenger expectations	Overcautious; waits for very large gaps, creating stuttering feel	Better gap selection; still conservatively biased (est.)
Roundabout	Assertive entry, maintains traffic flow	Often pauses and breaks traffic flow; disruptive for following vehicles	Improved entry timing; roundabouts remain a known challenge area (est.)
Emergency brake	Firm, controlled; driver warning available	Hard stop; no warning; passengers lurch forward	Hard stop preserved for safety; comfort explicitly secondary to collision avoidance
Left turn across traffic	Confident judgment on gaps; smooth execution	Cautious pause; sometimes blocks intersection while waiting	Improved but consistently conservative gap acceptance (est.)

The key improvement driver between generations is explicit comfort training. Waymo collects post-trip comfort ratings from riders after every Waymo One trip (est.) and incorporates these as training signals alongside safety metrics. Tesla’s shadow mode includes jerk — the rate of change of acceleration — as a quality signal in its training pipeline. Reducing jerk, which passengers feel as the unpleasant “lurch” sensation at the start or end of a braking event, is directly measurable and directly improvable through fleet data at scale.

Section 3 — The “Driverless Moment”: First-Time Rider Psychology

Research on first-time autonomous vehicle passengers consistently identifies a three-phase emotional arc that shapes both adoption rates and word-of-mouth outcomes.

Phase 1 — Pre-ride anxiety (before entering the vehicle)

Approximately 70% of first-time driverless vehicle riders report nervousness before their initial ride (est.), driven by media coverage of AV incidents and a general unfamiliarity with ceding control to a machine. This anxiety is higher among older age cohorts and lower among riders who have previously used advanced driver assistance systems.

Phase 2 — In-ride transition (first 5 minutes)

Once in the vehicle, first-time riders exhibit heightened alertness: watching the road closely, monitoring the steering wheel, and maintaining a posture of readiness to “intervene” even when there is no control to take. Eye-tracking studies in academic AV research consistently show that first-time driverless passengers spend significantly more time monitoring the road forward than experienced driverless riders, who behave more like train or bus passengers.

Phase 3 — Post-ride reassessment (after completing the trip)

The reassessment phase is where adoption conversion happens. When the vehicle completes the trip without incident — which is the overwhelming statistical majority of all Waymo One trips — approximately 80% of first-time riders report they would ride again (est. from Waymo and academic surveys). This first-ride conversion rate is the most important single metric for AV mass-market adoption because it determines whether the early-adopter base expands or stagnates.

Waymo’s strategy of trusted-rider programs — invite-only cohorts expanding gradually into broader markets — was designed specifically to trigger the Phase 3 conversion at controlled scale before mass-market launch. Every rider who completes Phase 3 with a positive outcome becomes a word-of-mouth channel. Tesla’s Austin robotaxi rollout faces the same three-phase psychology at larger initial scale, with the additional variable that the presence of a safety driver during the supervised phase alters the Phase 2 experience significantly.

Section 4 — Why Ride Quality Drives Adoption: The Business Case for Comfort

The connection between comfort and commercial outcome is direct and measurable across multiple dimensions.

Metric	Why it matters
Repeat ride rate	A rider who has a smooth, safe first experience is an estimated 3–5x more likely to become a regular user (est. from Waymo internal data) — the difference between a one-time curiosity and a recurring revenue source
Word-of-mouth NPS	Robotaxi net promoter score directly drives organic acquisition; uncomfortable or frightening rides generate negative social media that slows new-rider conversion at zero marginal cost to the competitor
Peak hour demand	Commuters — the highest-value rider segment — require consistent, predictable comfort; a single bad morning-commute experience eliminates a potential daily-frequency user
Premium pricing tolerance	Riders who rate their experience highly accept higher prices; comfort is pricing power, and the gap between premium and ride-hail pricing is where robotaxi unit economics become viable
Accessibility retention	The non-driver population (elderly, mobility-impaired, vision-impaired) has no viable alternative to autonomous vehicles at commercial scale — but comfort determines how frequently they use the service and whether they recommend it

Waymo has reported NPS scores above major ride-hail competitors in its operating markets (specific figures not publicly disclosed, est.). This NPS advantage is the commercial compounding effect of the Phase 3 conversion data above: each comfortable first ride generates organic growth that no paid acquisition can replicate at equivalent efficiency.

Section 5 — What to Watch: Comfort as a Ramp Metric

The Physical AI Benchmark series tracks operational metrics — weekly rides, fleet size, disengagement rates. Passenger comfort is an undertracked but equally consequential ramp metric. The following are the specific indicators that should accompany any future operational report.

Jerk reduction rate — The rate of change of acceleration (jerk, measured in m/s³) is the most precise physical measure of braking and acceleration comfort. A declining jerk metric across software generations indicates that comfort is improving at the same rate as safety. Both Waymo and Tesla have the data to report this; neither discloses it publicly as of mid-2026.

Post-ride rating average — Waymo collects star ratings after every Waymo One trip (est.). A rising average rating over time is evidence that the software improvement cycle is delivering comfort gains, not merely safety gains. Tesla Austin should implement equivalent collection from day one.

Complaint rate per 1,000 rides — A quality floor metric. A declining complaint rate indicates that the worst-experience tail is being systematically eliminated, which is as important as improving the median experience.

Repeat rider share — The percentage of weekly rides attributable to returning users (rather than first-timers) is the most direct proxy for whether Phase 3 conversion is succeeding at scale. Neither Waymo nor Tesla discloses this figure publicly as of mid-2026.

The adoption flywheel runs in both directions. The positive flywheel: comfort → repeat rides → more training data → better model → better comfort. The negative flywheel: one uncomfortable or frightening ride → negative review on social media → slower new-rider acquisition → weaker data signal → slower improvement. At the scale Waymo operates — approximately 250,000 rides per week as of Q2 2026 — the positive flywheel is clearly turning. The open question for Tesla’s Austin expansion is whether supervised-phase comfort translates to driverless-phase comfort when the safety driver is removed, or whether the three-phase psychology resets entirely at that transition point. That transition, and what the comfort and NPS data look like on the other side of it, is the consumer-dimension metric to watch in the second half of 2026.

Sources: Waymo One rider surveys and safety reports (waymo.com/blog/, waymo.com/safety/); AV passenger acceptance research in Transportation Research Part F: Traffic Psychology and Behaviour (sciencedirect.com); Tesla FSD v12 and v13 owner reviews on Tesla Motors Club forums (teslamotorsclub.com). All figures marked (est.) are estimates based on published surveys, academic research, and industry reporting; they have not been independently verified and may differ from primary source data.