Driving is so last century, let’s bring back the horse

Our equine past might hold the secret to learning to love autonomous cars


Spend any time in the autonomous car conversation and you’ll encounter a deceptively simple question: “How will the driver hand off control to the car?”

Seems simple enough. Push a button or something. Right?

Kind of. But the challenge of switching control mid-flight is both complex and banal at the same time. It’s one of those vexing problems that gets harder the longer you think about it.

It’s difficult in part because we don’t yet have norms for how to interact with robots. So a simple exchange like “Here, take the wheel” has become a topic that is occupying many hours of thought and experimentation in labs and innovation centers around the world.

For all the scientists and engineers who are focused on the dynamics of that one moment — the handoff between human and machine — we think there is a much larger challenge which is being overlooked. What’s really needed is to change the nature of the relationship between human and machine: to create a driving experience in which car and driver are working together toward the same goal, rather than driver commanding the car or the car acting independently from the driver.

Designing these types of interactions requires a shift in the current paradigm. Today we think about transferring control. The human has control or the autonomous car has control. A lot of energy is being spent thinking about how to create trust during this handoff. But if this one interaction is going to feel trustworthy, then it has to be part of something larger: a new category of human to machine interaction. One that describes the relationship between the car and driver as a partnership.

The shift in thinking is this: how can those interactions build a connection between driver and car instead of just transferring of control? The opportunity is to design the relationship, not the interactions. Because once the relationship is understood, the interactions and trust will grow naturally.

Thinking of the car as a partner changes everything. Not just the types of interactions the driver can have with the car but the nature of those encounters and how the car might interact with pedestrians, for example. Handing over control of the car to the car is not equivalent to turning on cruise control. Approaching this handoff as if it were is to misunderstand the nature of the interaction at its core.

A transactional interaction (like pushing a button) actually erodes trust because it requires an awkward “handshake” to verify the transfer. It is made worse because that interaction never evolves. The car never learns. Every time you hand over control to the car, the same clunky protocol must be followed to ensure it was done correctly.

This robotic “handshake” is a countervailing force, diminishing trust with each encounter because it’s a constant reminder that the machine requires oversight. Humans don’t work like this. We learn. We build relationships and trust grows with every interaction. Transfer of responsibility gets faster and more fluid as a result.

It begs the question: what is a model for an interaction between human and machine that feels like a partnership? How do we design this transfer so that each interaction builds trust and instills confidence?

For guidance, we might begin by modeling another relationship where transfer of control is at once fluid and trusted. Ironically, one source of inspiration might be the one that was abandoned with the advent of the car 100 years ago. The relationship between horse and rider is at once more subtle and more advanced than the command and control paradigm we have with cars and offers a view into a type of interaction that addresses many of the concerns that autonomous cars surface.

There are countless examples of what this looks like in real life. My colleague, Danny, tells a story of his grandfather who used to ride horseback to monitor fence lines for a gas company in Colorado in the early 1900s. During a blizzard he became so disoriented that he couldn’t find his way back to the field house. After hours of fruitless searching, the situation became dire. As a last resort he let go of the reins and relinquished control back to the horse. Left to its own devices, the horse found its way back to the field house in half an hour. The horse’s intuition and the rider’s trust in the horse saved both of their lives.

The clarity of roles in that example stands in stark contrast to the current approach to driver handoff. There’s never a question of roles between rider and horse. The rider picks the destination. The horse picks the steps. Each task is equally distinct and important and leverages their respective strengths. There’s an interdependency that could be instructive for the relationship between autonomous car and driver. It’s a relationship built upon trust and intuition, one that evolves and grows stronger over time.

Meanwhile, cars are amnesic, reverting to zero at every interaction and never evolving past stranger status. It is the living definition of untrustworthiness.

Designers create better solutions by asking better questions. Framing the question of handoff as one of binary control misses a chance to solve a more interesting problem. Reframing the question as an opportunity to build a relationship between car and driver allows us explore and define the nature of control.

Reframing the question as an opportunity to build a relationship between car and driver allows us explore and define the nature of control.

What if the car could emulate the rider/horse relationship? In that relationship, the rider does not control the horse. The horse is not controlled by the rider. They have a common goal with a shared outcome. It’s a partnership.

The technical challenge of the handoff is formidable, but it’s just one of a thousand similar engineering triumphs that the automotive industry can herald. On the other hand, car as partner is an invitation to explore the nature of the relationship between humans and technology; an opportunity to understand the very nature of consciousness. It’s a pathway to answer one of the most difficult and compelling questions of our time and it’s time to saddle up to the challenge.


Special thanks to IDEO colleagues @DannyStillion, @racheltobias, @DarjaWendel, for help creating this article