It’s impossible to spend a day in Silicon Valley without hearing about the business potential of autonomous cars.

But I don’t hear as much debate about how cities should plan their autonomous vehicle strategy. This discussion is incredibly important: cities must plan ahead to make the technology work for their people, instead of reacting to technology and business models as they arise.

I discussed this topic at the Worlds Fair Nano with my good friend Elliot Katz (who is the co-chair of DLA Piper’s Connected and Self Driving Car Practice).

Here are the four key points from our discussion:

1. Driverless cars are the means, not the end

Autonomous vehicles shouldn’t be an end in themself. Smart Cities should aim for shared mobility and deploy autonomous vehicles towards this goal.

Shared mobility is essential for cities, like Singapore, that struggle with land constraints and intensifying populations. To solve the problem of congested roads and unhappy commuters, we need fewer vehicles on the road. This requires a strong push towards shared mobility (and away from vehicle ownership).

Fully autonomous vehicles will contribute significantly to the objective of shared mobility, especially if they are deployed in fleets (think Uber, but with autonomous vehicles instead of drivers). Operators can dynamically size the fleet — injecting new cars when demand surges, and deploying them to less-crowded outskirts when demand falls.

Theoretically, you would get faster rides and increasingly reliability, while cutting down on “surge pricing”. These cars will be constantly on the move, never taking up precious real estate by parking for lunch or waiting at the curbside for a next job. This will free up land for other purposes, especially in the crowded downtown areas.

Cities should carefully consider the allocation of private versus shared autonomous vehicles. If the objective is shared mobility, the optimal scenario is for all autonomous vehicles to be shared: either deployed in fleets, or for privately-owned vehicles to be shared among a smaller group of family members and friends.

While this does not necessitate excluding private autonomous vehicles, cities should consider how they allocate road-space to private vs shared vehicles, as this will determine the extent to which they will achieve shared mobility.

There are many policy levers to consider: from quotas for privately-owned vehicles (which Singapore employs), to incentives for private car owners to share their vehicles.

When cars are connected, it is easy to design these incentives based on real-time information. For example, a car owner can receive an offer to avoid a road toll if they pick-up someone else along the route.

2. Win everyone over

Cities must introduce autonomous vehicles in a way that builds broad-based public acceptance — rather than just appeal to the 20% of early adopters.

Many have pointed out that one of the biggest challenges to autonomous vehicle deployment is broad-based societal acceptance. There will be maximum efficiency and safety gains when autonomous vehicles are deployed at scale, and this can only be achieved if a large majority of city dwellers is comfortable riding autonomous vehicles.

Cities need to introduce autonomous vehicles in a way that builds broad societal acceptance. Unlike the US, where autonomous vehicles are tested on public roads, Singapore has introduced them in designated trial areas. For example, autonomous taxis by Nutonomy ply 12km of public road space in the One-North neighbourhood in Singapore. Autonomous electric buses ply public roads in the Jurong district. Soon, on-demand autonomous shuttles provide rides on Sentosa, an island dedicated to tourism and recreation.

The Government works closely with AV companies to ensure that the testing routes provide sufficient challenge, but are not too far out of the vehicles capabilities such that it creates dangerous scenarios. We also work closely on the requirements for autonomous vehicle testing, and on after-action reviews when accidents occur.

While this seems like an arduous process for both company and Government, it is a long-term investment towards shared mobility. It is also better for accidents to happen within limited contexts. In this early stage of testing, accidents provide valuable lessons that will lead to improvements in the technology. In late 2016, Nutonomy had its first accident with a lorry in the One-North trial area.

Fortunately — and to some extent by design — the impact was limited and no one was injured. It is incredibly important to public perception that these accidents happen in a limited context with no fatalities, unlike what happened with Tesla earlier last year. The Government’s commitment to working with AV companies gives assurance that there is an added layer of accountability for the safety of these vehicles.

These trials have arguably piqued Singaporeans’ interest in riding autonomous vehicles. While Nutonomy limits the pool of people who can bid on their autonomous taxi service, it is still more accessible than having to purchase a private autonomous car to experience riding in one.

3. Co-Design your regulations

Cities must work with AV companies on their regulatory approach to autonomous vehicles.

One additional benefit of a city working closely with AV companies on autonomous vehicle trials is that it creates a space for a meeting of minds.

Unlike many US states, Singapore has not yet introduced regulations towards autonomous vehicles. Instead, we work with companies on trials that shape our thinking on the appropriate regulatory approach.

At the same time, companies have given feedback that it is useful to understand the Government’s concerns so that they can address these concerns at the design stage.

4. It isn’t just about cars!

While autonomous cars receive the most attention, other forms of autonomous vehicles hold incredible promise for the objective of shared mobility. For example, if a city has autonomous freight and autonomous utility vehicles, these activities can be done in the dead of the night, when commuters aren’t trying to move around. This frees up road space in the day, and makes for a better commuting experience for everyone.

But a final point to remember: Smart Cities should never deploy technology for its own sake. They should define their objectives, and target their time, money and policy interventions in ways that achieve those objectives.

Transport is just one area, of course. This applies to healthcare, education, housing, and any issue that a Smart City deals with!

Karen Tay wrote this blog in a personal capacity, and is Singapore’s Smart Nation director for North America.