Falling Asleep at the Wheel

If you’ve read my previous two articles, you’ve most likely inferred that I hate driving. It’s true-I do. Aside from my fascination with Artificial Intelligence and its implications on significantly altering society over the next 50 years, I am personally invested in diving deep into the sub-topic of autonomous vehicles so I can know when taking a nap while driving is possible. After researching online, I was overwhelmed with the range of estimates for when to expect to fulfill the lifelong dream of never driving again, and decided to connect directly with experts to gain some first-hand insight. I had a couple of fascinating conversations with experts whose backgrounds and work with the topic come from two radically different perspectives.

First, I spoke to a UPS competitor, getting a first-hand account of how mature freight companies are readying themselves for this massive seachange. Second, I spoke with Deepak Verma who is a Partner at Innospark Ventures-a Boston based venture capital firm focusing on early-stage artificial intelligence companies. In this conversation I gathered an understanding of what needs to occur from an AI, technical evolution perspective to enable the ability for automated vehicles to run as intended (and safely). Unfortunately for my dream, much of what I gathered from my conversations was a less rosy outlook for a short-term reality of full-fledged autonomous vehicles, but the reasons why are quite fascinating. Specifically, the shift most necessary for AI to facilitate the transition to fully autonomous vehicles will also provide tremendous benefits-and potentially dubious prospects-for society at large.

Within the freight industry, the time horizon for two initiatives were highlighted during my conversation: fully autonomous trucks and platooning. Because of where AI is, the short-term reality for the first initiative is having these vehicles stationed in a parking lot outside of a city like Boston (say Newton for argument’s sake), which is then programmed to haul freight from that parking spot to another parking spot outside of San Francisco. At that point, a driver will need to take over the vehicle and make the deliveries manually-as explained in a previous article (link)-this is because AI is not at the point where it can handle the nuance of driving around cities. The second initiative, platooning, is actually a more realistic initiative: it involves the driver of one truck leading a fleet of automated vehicles trailing behind him, which allows for more hauling capacity per driver. The next iteration is making the entire fleet automated, with the driver manipulating controls remotely. From a technology perspective, some form of these initiatives could be implemented into a freight company’s business model on a short-term horizon. However, hurdles exist outside of corporate control. First, certain restrictions apply depending on the value and sensitivity of the haul-think gold, diamond or expensive medicine. Second, legislation from state-to-state is extremely complicated. There may be a reality where an auto-truck can drive through Indiana, but regulations require the truck to take the long route through Kentucky and Missouri because Illinois doesn’t permit automated vehicles. Drones, Aircrafts, and autopilot technologies have gone (or are going through) similar regulatory hurdles.

While automation as a corporate initiative in the freight industry is encouraging, I still didn’t have an answer to what I wanted to find out most: when will I be able to go to a dealership, buy a car, and nap in the back seat on my way home?

Deepak Verma is a partner at Innospark Ventures, a VC firm dedicated to investing in early-stage artificial intelligence companies. As a disclosure, the firm is not yet investing in autonomous vehicles, but their expertise in AI provides a great perspective and voice on when we should expect the requisite technology to be ready to power self-driving cars. In his opinion, it will be a while before this becomes a reality. Deepak states that the technology is “only as good as the data it’s been provided. Cars can get really good driving in one town, but placed in a different town it won’t do well.” This echoes the sentiment I made in my previous article which is the difference between narrow and general AI. Narrow is the technology that is available at present, defined as machine learning that equals or exceeds human intelligence or efficiency in one specific area (e.g. assembly line robots). General AI is a computer that is as smart as a human across the board and can perform any intellectual task that a human being can. “If conditions call for rain, snow or black ice, AI today could not process that” says Deepak. Processing these sorts of nuances would be general AI. As for how the general population will be introduced to self-driving vehicles, Deepak relays that it will not be in one fell swoop. Automation will gradually be implemented into existing vehicles feature-by-feature, and this is already in motion today with automatic parking and adaptive cruise control (if there is suddenly a car in front of you, the vehicle slows down automatically). Deepak reiterated the point that those features show progress, but getting to 100% is far away: “Most of AI has been focused on building machine learning to process more data, faster, but we haven’t figured out how to incorporate common sense.”

Though my dream of napping at the wheel seems like a distant reality, the focus of incremental incorporation of automated features is both encouraging and reassuring. Encouraging because incremental gains in vehicle automation are a central initiative for the biggest players in the automotive industry. Reassuring in the sense that the Silicon Valley ethos of “move fast and break stuff” is checked at the door since the consequence of technology failure would lead to disastrous outcomes. Tesla’s recent, high-profile mishaps (which has led to lost life) illustrates this point.

I’m still uncertain as to when my dream will become a reality, but thankfully many of the world’s brightest minds share the same dream. Though, I imagine for less self-serving reasons.