Mr. Mayor, you must not waste time on the detours and distractions that dogged your first term.

Last week’s terrorist attack in New York City is a stark reminder that the use of vehicles as weapons is on the rise. But we don’t have to remain vulnerable to this new terrorist tactic if we smartly — and quickly — harness technology to protect human life.

This much is clear: Cars and trucks are increasingly being used to kill innocent people. The Global Terrorism Database at the University of Maryland has documented more than 350 vehicle-as-weapon deaths since 2006, and many more injuries.

According to the Counter Extremism Project, a nonpartisan international policy group, such attacks have increased each year over the past decade. So far in 2017, at least 32 people are dead and more than 240 wounded from vehicles used as weapons.

These numbers are particularly chilling because victims are ordinary people engaged in the most innocent of activities — biking and walking in the case of the attack in lower Manhattan.

The advancement of autonomous driving technology and artificial intelligence means similar attacks theoretically could accelerate, with faceless perpetrators operating several vehicles at once in different locations.

However, there is a way to make our streets safer using existing and developing technologies, if we plan ahead.

Some, but not most, vehicle attacks could be deterred through better street design and existing mechanisms. I recently observed a car on the same bike path Sayfullo Saipov allegedly used to mow down about 20 people, killing eight. I know of at least two other vehicle-bike fatalities, both in 2006, that happened as a result of driver error on this pathway; in one, a car traveled for almost a mile before striking a young man at Houston St. just blocks from the current tragedy.

The concrete barriers installed after the attack will stop vehicles from intruding in the bike path. Creating still more secure bike paths in the short term could involve placing permanent bollards bike-widths apart.

However, we can’t ring every public space with protective infrastructure. Instead, we should focus our energy on the tool used criminally or accidentally to kill and maim people: the motor vehicle itself.

Most 2017 car models are computerized, and many are equipped with lane control, automatic braking, GPS and other features. In just a few years, all cars will be partially or fully autonomous.

We are a short step away from cars that can be programmed not to go the wrong way down a oneway street, travel on a bike lane, or speed through residential areas. Uber is already using some of this technology in Pittsburgh.

Researchers are looking at ways to connect smart cars to other smart cars and to link those cars to smart infrastructure.

This could and should be an opportunity to prevent cars from doing what they shouldn’t be doing. The infrastructure, along with GPS and mapping, can control the vehicle. In most cases, this will lead to a large drop in crashes.

But in light of the terrorist threat, we need to go a step further, and have a government override built into new cars. Bern Grush, an autonomous-vehicle researcher, points out that highly automated vehicles that will hit our streets in the next few years will have abilities of communication and oversight to prevent such criminal use.

I propose a protocol that all auto manufacturers would agree to, ensuring their vehicles’ technology can communicate with each other and with the local infrastructure. Speeds would be controlled, along with access to auto-free zones. A central traffic control center would monitor compliance.

If someone figures out a way to override their vehicle’s safeguards, an alarm would be triggered so that law enforcement could be notified to take action.

The inevitable argument about loss of privacy could be mitigated by guaranteeing that whatever system is chosen wouldn’t prevent honest people from going places they want or need to go. Important information could be anonymized. Nor would anyone use an individual’s travel patterns for unwanted commercial purposes.

Tesla and SpaceX chief executive Elon Musk has warned that artificial intelligence poses “vastly more risk” than a nuclear North Korea. If we start planning now, we can prove that prediction wrong.