-
17:50
-
17:20
-
16:50
-
16:23
-
16:20
-
16:20
-
15:50
-
15:20
-
14:50
-
14:47
-
14:20
-
13:50
-
13:20
-
12:50
-
12:30
-
12:20
-
12:00
-
11:50
-
11:30
-
11:20
-
11:00
-
10:50
-
10:30
-
10:20
-
10:00
-
09:30
-
09:20
-
09:00
-
08:50
-
08:48
-
08:30
-
08:20
-
08:00
-
07:50
-
07:30
-
07:00
Tesla's ex-AI head says Waymo can't match coast-to-coast self-driving feat
Andrej Karpathy, former head of AI at Tesla, stated on December 31 that Waymo lacks the capability to replicate Tesla's recent milestone of a fully autonomous drive across the United States. He highlighted fundamental technological differences between the two companies' approaches to self-driving vehicles.
The remarks came shortly after Tesla owner David Moss completed a 2,732-mile journey from Los Angeles to Myrtle Beach, South Carolina, using the company's Full Self-Driving version 14.2 without any human intervention. When asked on X if Waymo could achieve something similar, Karpathy replied affirmatively, noting that Waymo's modular system prevents such end-to-end performance.
This exchange underscores the competing philosophies in autonomous vehicle development. Alphabet's Waymo relies on high-definition maps, LiDAR sensors, 5G connectivity, and multiple neural networks within a modular framework. Tesla, by contrast, employs a single end-to-end neural network that converts camera data directly into driving commands.
These architectural distinctions surfaced during a December power outage in San Francisco, where Waymo vehicles struggled at intersections with non-functional traffic lights. The company admitted its robotaxis sometimes required confirmation checks for dark signals, leading to response delays and traffic backups. Waymo reported successfully navigating over 7,000 such intersections during the blackout and has since rolled out software updates.
Karpathy's assessment arrives amid ongoing tensions with Tesla CEO Elon Musk, who on December 23 dismissed the ex-AI director's views as outdated on X. Musk asserted that Tesla's AI software has advanced far beyond Karpathy's tenure, boasting an intelligence density per gigabyte that outpaces rivals by at least an order of magnitude.
Despite Musk's earlier July invitation for Karpathy to rejoin the team, recent comments point to lingering friction. Karpathy left Tesla in 2022 after five years leading the Autopilot team and now heads AI education startup Eureka Labs. He praised Moss's achievement on December 30, calling it special as a coast-to-coast trip was a key early goal for the Autopilot group, recalling late-night video review marathons to analyze past interventions.
