Ford’s self-driving cars are really good, but are they good enough to win?

It was a humid, overcast day in Miami, and I needed to get from the city’s crowded downtown to the trendy, mural-splashed neighborhood of Wynwood. I pulled out my phone, tapped an app, and hailed a self-driving car. A few minutes later, a Ford Fusion crowned with cameras and high-powered sensors crawled up to the curb. The street sign above me read “No Parking: Autonomous Vehicles Only.” Good thing: I got in, buckled up, and away we went.

Of course, none of it was real. The app, the sign, my destinations… all of it was part of a day-long, intricately choreographed pageant put on by Ford to offer a glimpse of its autonomous future. It was like Ford carved out a mini-Westworld in the middle of downtown Miami to show off what was possible. But weirdly enough, the only thing that was real was the self-driving cars.

On Wednesday, Ford summoned myself and about two dozen other reporters to Miami to show off the progress it has made over the last year of testing in the city. But more than that, the event was designed to broadcast a message to the public that Ford’s technology is advanced and road-ready, its business model is solid and well-thought out, and its future is brighter than recent headlines would suggest.

A quick look at some of those headlines, and you can see why Ford is so keen on changing the narrative. Its profits are plunging, its credit rating is one step above junk status, and it’s widely seen as lagging behind rivals like Waymo and General Motors in the quality of its autonomous technology. Waymo is launching its first commercial service next month; GM in 2019. Ford says it won’t be ready until 2021. The latest rumor is that Ford will team up with Volkwagen in a joint self-driving car venture.

In the meantime, we have the powerpoint promises of Ford’s top executives: lots of talk about platforms, mobility clouds, transportation pain points, and the cars themselves — a dozen in Miami, and a few more in Detroit, Pittsburgh, and Washington, DC. Ford says it will have 100 vehicles on the road by the end of the year, and is currently working on a purpose-built autonomous vehicle without traditional controls like a steering wheel and pedals.

Here’s what I’ll say: I’ve ridden in about a dozen autonomous vehicles in a variety of settings over the past three years, and this was one of the most complex environments I’ve ever seen an AV navigate. I’ve been an AV passenger in San Francisco, Mountain View, Las Vegas, Detroit, and Pittsburgh, but nothing like Miami — cars and trucks and candy-colored mopeds jockeying for space with bicycles, scooters, and old ladies with laundry-filled pushcarts. Despite the heat, it’s the type of environment that would make most AV operators break out in a cold sweat.

But the ride was a marvel. Ford’s self-driving vehicles deftly handled a variety of challenging scenarios that have been known to trip up even the most skilled AVs, from unprotected left-hand turns, to construction zones, to narrow, two-way roads without lane markings. The car rode confidently alongside other drivers, braked for pedestrians, and inched its way assertively into intersections. The only time the vehicle seemed confused was when a passing flatbed truck kicked up an enormous cloud of dust, forcing the safety driver to momentarily take control.

Brian Salesky, CEO of Argo AI, the Pittsburgh-based startup in which Ford invested $1 billion last year, said the dust-cloud disengagement was evidence that the experience was authentic, and not scripted.

“We’re not gaming this for you today,” Salesky said. “This is what life is like in the real city. We could have rerouted you so you wouldn’t encounter the dust cloud. We saw the same thing yesterday and it’s like, ‘Yeah, these are things we need to address.’”

Salesky’s list of things that Ford’s AVs need to “address” is long, but not insurmountable. “When you make a list of the 2,000 features that need to be implemented,” he said, “I want you to give us credit for the 1,800; the 200 that are left, dust is in them.” He added, “We’re focused on building a safe, confident driver.”

The cars were Ford Fusion sedans equipped with an array of high-powered sensors and cameras strapped to the roof, and enough computer power in the trunk that the back seats got noticeably toasty. (As did my tush.) I took four trips with Ford, each time in a new AV. Each trip featured a safety driver behind the wheel, and a co-pilot in the passenger seat with a laptop, calling the play-by-play. A touchscreen was mounted behind the headrest of each seat, and I was instructed to tap the big “Go” button to initiate each ride.

But again, it was all for show. The touchscreens didn’t really do anything other than give me an idea of what it will eventually be like. Ford forbid me from photographing the inside of the vehicle, because the design around the user experience wasn’t final. And we didn’t have the freedom to ride the cars wherever we wanted, but rather were limited to four possible destinations within a 6-mile section of Downtown Miami. While the destinations were canned, Ford claimed the routes taken were still dynamic and unscripted.

The first stop was Ford’s AV terminal, where the cars are cleaned and the sensors are calibrated between trips. But with its gleaming white interior, and the dozen or so workers tapping away at laptops, it felt more like a co-working space than a oil-streaked maintenance garage. Here, Ford’s fleet management and transportation-as-a-service experts walked us through a typical day in the life of its autonomous vehicles. We discussed concepts such as uptime, utilization rates, and the age-old question of what to do when someone pukes in the AV.

“This is dirty work,” said Don Conroy, Ford’s director of AV terminals, “and we plan to be really good at it.”

Next was Wynwood, where Ford had representatives from Domino’s and Postmates, two of its corporate partners, as well as a few local businesses — a florist and a dry cleaner — who had participated in a weeks-long pilot testing Ford’s autonomous vehicles for deliveries. That morning, Ford announced it was teaming up with Walmart and Postmates to further test out using autonomous vehicles to make grocery deliveries.

There are still a lot of kinks to work out. The vehicles Ford is using aren’t really autonomous, but designed to look that way for the purposes of the pilot. Moreover, the vehicles that Ford plans to use when it launches in 2021 haven’t even been built yet. But the user experience will be similar: input a code into a touchscreen embedded in the side of the vehicle and a steel shutter slides open revealing a paper takeout bag (empty), a stack of pizza boxes (also empty), or a two-liter bottle of Coke (warm).

The idea was to demonstrate the experience of retrieving a delivery from a Ford autonomous vehicle. Ford is betting that people will be willing to walk to the curb to fetch their food or flowers, if it means interacting with a self-driving vehicle. But novelties wear off, and people will always prefer convenience and affordability. And in the markets that Ford is targeting — dense, urban — most people live in tall apartment buildings, or have mobility issues. There’s a lot still to work out.

Last, we were shepherded to a midtown neighborhood, where Ford’s head of digital content sermonized on the in-car experience for AV passengers. In short, Ford envisions earning revenue from partnerships with local businesses and concert venues that serve up ads and digital coupons to riders of autonomous vehicles. On your way home in a Ford AV? Stop by the local artisanal soda shop for 50 cents off a cherry lemonade. No need to get out of your car, the waiter will bring it right to your window. It’s a 21st century, self-driving version of a carhop.

And this was what Ford really was interested in showcasing: its business model. Ride-hailing, fleet management, deliveries, and digital content — all the pieces Ford says will make its $4 billion bet on autonomous vehicles pay off and will send it rocketing past its supposed superior rivals. Today, ride-hailing costs around $2.50 per mile. Sherif Marakby, president and CEO of Ford Autonomous Vehicle LLC, estimated that the company’s autonomous vehicles could drive that price down to $1 per mile. Marakby said Ford wasn’t interested in following in the footsteps of Waymo and GM by launching a commercial service in one city in the near-term.

“If we wanted to call a launch 100 vehicles next year and go into some business, we could do that,” Marakby said. “I don’t think that’s what we want to do.”

Ford’s plan is far grander and will involve up to 100,000 self-driving cars — with no steering wheel, accelerator, or brake pedal — operating in multiple U.S. cities, delivering people and packages around the clock. The event in Miami was designed to leave one with the impression that Ford is approaching this challenge cautiously and deliberately. And it did that.

But it also left me questioning where this technology was going. If your vision of the future is a self-driving car that can be summoned at any time to take you anywhere you want to go, I have some bad news for you: that is many years away, decades, possibly more. Instead, what we are likely to get are self-driving taxis and shuttles operating in small, geofenced areas of major cities. Some will be on fixed routes, like a bus, others will be more dynamic and possibly more expensive. Those vehicles will likely retain human safety drivers for fear of liability. Years of mapping and data collection will be needed before any service can launch in any new city. And the rides may be cheap, but they are unlikely to be more convenient than a human-driven Uber or Lyft.

We’re just now coming off the sugar-high of the early, inflated expectations of self-driving cars. But Ford has too much invested in those expectations to slow down. Earlier in the day, the company’s CEO Jim Hackett said that AVs have the potential to become a $10 trillion business — which is around $3 trillion more than the conclusion in a widely cited report by Intel. Which number is real, and which one is just for show?