Automotive and Artificial Intelligence: From Driver Assistance to Driverless Service, and How Much Evidence Is Enough

Purpose. The car is both a product that AI is entering, in driver assistance and driverless service, and a product that AI helps to make. This article asks where artificial intelligence stands in the automotive industry in 2026, what the evidence on safety shows, and where the industry may be by about 2035. Design and methods. A review that combines the share of the automotive research literature that involves AI (OpenAlex, 2010–2025), Eurostat's survey of AI use in EU vehicle manufacturers (2021–2025), peer-reviewed safety analyses of driverless service, and a classic analysis of the evidence needed to demonstrate safety. Claims taken from the literature were checked against the sources cited. The scenarios are judgements built on this evidence. Findings. The share of automotive works that mention AI rose from 6.4% in 2018 to 18.6% in 2025, the smallest relative increase of the 20 industries (2.9 times as much) because the base was already high. In the EU, 23.9% of enterprises making motor vehicles used AI in 2025, up from 11.4% in 2021 and above the 19.9% for all enterprises, mostly for text mining (13.5%) and language generation (11.1%), with 4.9% using autonomous robots or vehicles. In a peer-reviewed analysis of 56.7 million rider-only miles through January 2025, the Waymo service had statistically significantly lower crash rates than human benchmarks for crashes with any injury, airbag deployment and suspected serious injury or worse, with a 96% reduction in injury-reported intersection crashes with other vehicles; the authors are employees of the operator, the data cover a few cities, and the benchmark comparison is retrospective. A classic analysis showed why such evidence takes so long to accumulate: demonstrating reliability on fatalities and injuries requires hundreds of millions to hundreds of billions of miles. We expect driver assistance to be universal and driverless service to expand city by city through 2028, the pace to depend on accumulated safety evidence and public acceptance in 2029–2032, and long-distance and personal driverless cars to remain open. Originality. The article keeps the manufacturing and product stories apart, reads the safety evidence for driverless service with the limits of its source in view, and sets a quantitative adoption series for vehicle makers beside it.