Driverless cars won't become widespread until automakers and police determine how to pull over autonomous cars, how to program the AI to recognize authorities
with a big asterisk Tweets: Thomas Baekdal / @baekdal : This should be the easiest thing to solve. Just have the police car beam a command to stop to the car. http://twitter.com/... Martin Bryant / @martinsfp : Do driverless cars mean no more car chases? I suppose movies will need to have a 'hacking the car's software to avoid the cops' scene. http://twitter.com/...
Context & Ripple Effects
The pull-over problem is old news in autonomy circles but newly urgent: Robert Scoble's 2016 rundown of why self-driving cars wouldn't replace human drivers listed police and laws alongside maps and technology as unsolved blockers, and this Bloomberg piece shows that blocker still standing years later.
What changed since then is the industry's answer to it. Rather than crack problems like authority recognition head-on, experts pushed automakers toward incremental driver-assistance instead of a moonshot Level 4 system, and the ~$100B investors poured into self-driving startups has produced little progress. Police interaction is now less a footnote and more a case study in why full autonomy slipped.
First-order effects
- Automakers deploying driverless fleets must design an explicit authority interface into their vehicles — programming the AI to detect and comply with police commands — before any city will tolerate wide deployment.
- Police departments gain a new operational burden: without a standardized way to signal a driverless car to stop, officers face vehicles that may not respond to conventional traffic-stop procedure.
Second-order effects
- Automakers and police forces are pushed toward negotiating a shared stop-signal protocol — likely a broadcast command standard — turning what looks like a software detail into an inter-industry coordination problem.
- The unresolved pull-over question strengthens the case for the incremental advanced-driver-assistance path over a Level 4 moonshot, since assisted-driving cars keep a human who can be pulled over the traditional way.
Third-order effects
- If the pattern holds, public-safety interoperability becomes a formal regulatory gate for autonomy: no certification without a demonstrated compliance-with-authorities mechanism, adding a governance layer on top of the safety-testing layer.
- The broader lesson from the capital-heavy autonomy bet is that deployment timelines are set by civic and legal infrastructure as much as by AI capability — a structural drag that favors companies selling partial autonomy today.
The trend: Autonomy deployment is increasingly gated by governance gaps — police procedure, law, and standards — rather than by raw AI capability, which is why the industry has drifted toward driver-assistance over full self-driving.