Six months ago, I was lucky enough to join Cruise's Beta program, and experience their autonomous taxis in San Francisco. Last week after a long-awaited decision by SF regulators, Cruise along with Waymo, have been given permission to scale up their activities in the city. I've used their self-driving taxis on many occasions, and the journey has been fascinating.
My Four Takeaways:
1. Machine Trust:
Within minutes, being driven by a machine felt surprisingly routine and relaxing with the in-car controls heralding at a new era of mobility UX. Today, it's choosing music or pondering why the map shows a longer route home (hint: it avoids riskier manoeuvres like ‘unprotected left turns’), but future services using the large screen are clearly in the wings.
2. Public Pushback:
The resistance to autonomous vehicles in general and the expansion in SF in particular was eye-opening. SF residents can be surprisingly conservative I’ve come to learn. While 'Week of Cone’ – was amusing in that the high-tech vehicles could be halted with a humble traffic cone on the hood (an admirably low tech way to defeat some incredibly sophisticated hardware and software), it does talk to deeper, legitimate concerns about gig worker job security, privacy, and unique safety challenges facing AI drivers – there have been isolated cases of driving over firehoses, downed power lines or through ‘do not cross’ tape.
3. Comparative Standards:
It's fascinating to observe the tendency to hold AI-driven vehicles to far higher standards than human drivers. In the US, human error leads to something like 13,000 accidents daily and evidence suggests a high tolerance to preventable deaths (e.g. gun control). However, one autonomous car's hiccup can become front page news. Why do we demand AI perfection in driving and other applications when it is clear humans are far from perfect?
4. The Future is already here, it’s just not evenly distributed yet:
Most of us have seen the artists impressions of futuristic cars with no steering wheel capable of driving in either direction and probably dismissed it as many years away. Well, Cruise has been developing its own version of this called ‘Origin’ which is waiting regulatory approval. Once you’ve been successfully driven in a retro-fitted Chevy Bolt, then the mental leap to driving in a vehicle like the Origin is easy.
Of course an important part of understanding the innovation lifecycle of any new product or technology is appreciating that many groups will be reluctant to embrace ‘the new’ even when the data shows it has the potential to improve our lives. It seems clear to me now, having sat in the back of multiple autonomous taxis, that we must tackle the challenges head on whether they be ethical quandaries, cyber threats or concerns around social dislocation. As Marc Andreessen eloquently wrote in his recent essay ‘Why AI will save the world’ we mustn't unconsciously default to a ‘Precautionary Stance’ for all things AI.
I believe the SF Public Utilities Commission should be commended for voting 3-1 to expand autonomous vehicle activity in San Francisco while accepting the road ahead contains obstacles - this week for example saw a slow speed collision between a fire truck and a Cruise taxi. Clearly a maximalist software development approach of 'go fast and break things' is not appropriate.
I'm optimistic about the future of autonomous ride hailing and I hope a forward looking mindset can unite developers, regulators, and the public allowing us to harness technologies like this for our collective good sooner rather than later.
Thanks to Marc Amblard at Orsay Consulting for inspiration and Amar Rathor for getting me onto the Cruise Beta!
Kommentare