Like many of you, I’ve been amazed to watch the pace and progress of Elon Musk’s Tesla and SpaceX ventures in the last decade. In many ways, both companies have consistently delivered on the promise of a fast-moving future where challenges like renewable energy, or long-distance space travel in reusable rockets, are actually solvable. The 100,000 or so Teslas on the road today, the brand loyalty of their customers, and the glowing reports of blogs and magazines everywhere are testament to his success at making the previously-impossible possible. The nearly half-million downpayments for the upcoming Model 3 blew everyone away, including Musk. It’s an inspiration to product managers and technologists everywhere.
Sci-fi tells us to expect the future Any Day Now. That said, Tesla seems to have hit a serious bump in the road recently, with the combined challenges of production issues on its Model X SUV and recent reports of possible driver deaths due to use/overuse of the company’s Autopilot feature. Tesla’s recent bid for Solar City has also encountered serious objections from the financial community with a major hit to the company’s stock price.
Musk’s vision here is probably impeccable. He’s got too good a track record at this point for anyone to seriously question it, in any case. The bigger issue is the sequence and pacing of some of the decisions, many of which appear to be driven by optics rather than reality. The two biggest cases in point are some of the design choices for the Model X (distance-sensing gullwing doors, new seat technologies) and the rapid deployment of Autopilot.
The fact is, after the early cult hit of the Roadster and the broad upmarket success of the Model S, Tesla already had the momentum to get it where it needed to go. Many of the steady stream of announcements since those core products shipped (‘They’re building a Gigafactory!’, ‘Teslas can filter air to biohazard levels!’, ‘Teslas can float!’) have contributed to this unprecedented momentum. The fact that the Model 3 looks pretty much like a downsized Model S, and generated the massive orders that it has, is testament to this.
The problem is that every great product guy, including Musk, needs to have a near-perfect sense of pacing, and in some recent cases he and the company have gotten ahead of themselves. This is completely understandable – they’ve been on such a roll that it’s to be expected. In particular, the Model X design choices such as the unique door and seat technologies which added real risk to the product ship dates, are of a piece with the can-do approach that made the company successful. Even Musk has acknowledged that some of these decisions amounted to ‘hubris’, a term not often found in dry Wall Street disclaimers, and the company will undoubtedly muscle their way through the manufacturing issues over time, despite the short term financial hit.
The bigger challenge will be for Tesla to step back from its leading-edge promotion of Autopilot technology. As Drivers Ed teachers everywhere point out, a car moving at highway speed is a deadly weapon, and nobody is advocating that (for instance) semiautomatic rifles be self-deployed and controlled by autonomous systems. It’s risky enough that they’re controlled by people at this point. But somehow, partly because of the rapid advancement of this class of technology and by Tesla’s great track record, everyone involved has hurtled forward on this one, common sense be damned.
It’s especially odd because Musk himself has been such a vocal critic of the long-term risks of AI for humanity. But maybe that speaks to a consistent over-estimation of how ready this stuff is, too. The reality is, there are far too many edge cases and challenging problems still to be simulated, tested and resolved for this particular technology to be cut loose today. The propensity of enthusiastic customers to really count on the new stuff, even do things like watch videos while driving based on their faith in the system, also indicates that more work and education is needed.
Taking a step back, all technologies evolve over time. The deep learning/self-navigation/neural net tech that car vendors are using for driver assist and autonomous vehicles have taken many decades to mature; the sensors and GPUs that are used to make them a reality are improving rapidly every year. The crossover point for real software and hardware, in the real world, to work well enough for consumers to use safely every day is definitely approaching. We just don’t know how soon.
As a longtime software entrepreneur (www.axlevideo.com) and trained physician, I feel like I have some perspective on the cautious side of things. There are some fields, like clinical therapies, where the pace of change has been glacial for too many years, which are really overdue for regulatory reform and moon-shot initiatives – thanks, Joe Biden. You can argue that the auto industry was one of these fields ten years ago; today, not so much. I’m not sure about you, but I somehow trust Volvo or Mercedes when to tell me they’ve curated and tested the heck out of these technologies, more than Tesla. That’s a real gap for Musk’s company.
Great product manager/CEOs like Steve Jobs, Bill Gates and Henry Ford, warts and all, need a clear sense of pacing and balance. There are times to leap forward with waves of technology, and times to consolidate and iterate a winning formula. The latter is much less glamorous, but I’d argue that it’s what’s needed here. A key part of this journey will be to recalibrate the use of the Autopilot technology, and determine where/when it can be successful. If the company can sensibly walk its customers, investors and critics through this process, it will earn its way to the next level of massive commercial success and influence.
For now, it may be wise for Tesla to slow down from its self-proclaimed Ridiculous mode. In short, the Autopilot Sci-Fi blind spot may have been in thinking that just because something should be true Any Day Now, it’s true already.