When Self-Driving Fails: Evaluating Social Media Posts Regarding Problems and Misconceptions about Tesla’s FSD Mode

Abstract

With the recent deployment of the latest generation of Tesla’s Full Self-Driving (FSD) mode, consumers are using semi-autonomous vehicles in both highway and residential driving for the first time. As a result, drivers are facing complex and unanticipated situations with an unproven technology, which is a central challenge for cooperative cognition. One way to support cooperative cognition in such situations is to inform and educate the user about potential limitations. Because these limitations are not always easily discovered, users have turned to the internet and social media to document their experiences, seek answers to questions they have, provide advice on features to others, and assist other drivers with less FSD experience. In this paper, we explore a novel approach to supporting cooperative cognition: Using social media posts can help characterize the limitations of the automation in order to get information about the limitations of the system and explanations and workarounds for how to deal with these limitations. Ultimately, our goal is to determine the kinds of problems being reported via social media that might be useful in helping users anticipate and develop a better mental model of an AI system that they rely on. To do so, we examine a corpus of social media posts about FSD problems to identify (1) the typical problems reported, (2) the kinds of explanations or answers provided by users, and (3) the feasibility of using such user-generated information to provide training and assistance for new drivers. The results reveal a number of limitations of the FSD system (e.g., lane-keeping and phantom braking) that may be anticipated by drivers, enabling them to predict and avoid the problems, thus allowing better mental models of the system and supporting cooperative cognition of the human-AI system in more situations.

Letter to the Editor: Update on Experiences of a Driver with Vision Impairment When Using a Tesla Car - Full Self-Driving (Beta) in City Driving

“…It is important to note that the FSD Beta system is currently in a beta testing phase and still in development, and many users have reported usage problems and potential safety concerns. 3,6 As automation technology continues to advance, it is possible that some of the issues and use challenges identified by this patient and other users may be remedied in future versions of the Tesla FSD system or in other semiautonomous vehicles. Our main goal in reporting this patient's entire experience across 2 years is to demonstrate the possibilities of using a car with level 2 automation to support driving safety and mobility opportunities for drivers with vision impairment.…”

“…It is important to note that the FSD Beta system is currently in a beta testing phase and still in development, and many users have reported usage problems and potential safety concerns 3,6 . As automation technology continues to advance, it is possible that some of the issues and use challenges identified by this patient and other users may be remedied in future versions of the Tesla FSD system or in other semiautonomous vehicles.…”

The Use of Social Forums to Train Users About Shortcomings of Tesla Full Self-Driving (FSD)

“…We used a common set of materials across both experiments. From the 196-communication corpus (Linja et al, 2022), 33 base posts with comments (associated with erratic driving circumstances posted by Tesla FSD users) were used in the experimental condition. The veracity of the communication corpus cannot be independently validated, but the purpose of this report is to determine whether or not content from the social forum can be utilized as a training platform for new AI users.…”

“…Furthermore, such training might backfire, because it will tend to focus on the failure modes of the system rather than the successes. In the case of self-driving vehicles, besides numerous third-party training organizations, such as Drivyn, the extensive traffic on user discussion boards centering on Tesla’s FSD mode (see Linja et al (2022) ) illustrates that the community is stepping in and educating themselves via social Q&A platforms (Oh, 2018) to provide the training that the vendor is unwilling or unable to provide, and this represents a potential case of Collaborative Explainable AI (CXAI)” - (Mamun et al, 2021)”. Although these forums are active and include both positive and negative examples of how self-driving vehicles fail in the real world, it has not been demonstrated whether drivers can benefit from this.…”

“…Safety is perhaps the most salient expectation that users have about autonomous vehicles. However, Linja et al (2022) argued that along with safety, users have several other expectations: they expect systems to operate according to laws and rules of the road; they expect systems to provide information consistent with the information being used by the system to make decisions (i.e., transparency), and they expect the system to otherwise behave like a human driver (i.e., cognitive anthropomorphism; (Mueller, 2020). Violations of any of these expectations may lead to reluctance to adopt the system, and any can lead to accidents For example, Dixit et al (2016) compiled a summary of accidents involving autonomous vehicles.…”