I remember as a young PwC consultant, a manager who used to walk around and joke “what do it do?” as he reviewed our specs and our code. It was his way of reminding us to deliver what the customer wanted, not a “solution looking for a problem”.
As I review the inflated expectations people have around today’s automation technologies (AI, robotics etc.) that is very useful advice to bear in mind.
Take autonomous cars. When someone spots Google testing a driverless concept or when Elon Musk says “ you’ll be able to summon your car from across the country” people jump to driverless Uber scenarios. Within a year or two, no less. If displacing a few Uber drivers is the end goal we have missed the mark by a wide margin. The promise of driverless cars is much nobler — to reduce over 30,000 car accident deaths a year in the US, and over a million around the world.
I, for one, do not think we will see autonomous cars en masse for decades — and describe in Silicon Collar selected scenarios where we will see them. While the technology is evolving nicely, our road infrastructure, our driving laws, ethical discussions and incumbent interests will slow down progress toward broader deployment.
You know what? So long as the technology keeps evolving, we should still celebrate. In a decade, if most of our cars have blind spot sensors, auto-braking, distracted driving alarms and many of the 30+ Audi Q7 safety features we will be so much safer.
Indeed, Honda’s newest Civic LX, equipped with an advanced driver assistance system, recently navigated a 25-mile Detroit area road course nearly hands-free. With price points starting at a little over $20,000, it may be “driverless enough” for many customers.
Or when McKinsey says “about 60 percent of occupations could have 30 percent or more of their constituent activities automated.” again people jump to conclusions. Stop and think about the technologies needed to cover the breadth of the 2,000 activities across 800 occupations that are mentioned by McKinsey “from fine motor skills and navigating in the physical world, to sensing human emotion and producing natural language.” Think of the specialized sensors, robotic limbs, image capture and other technologies that would need to coalesce and allow for the vast sweep of capabilities.
While you consider that wide palette of technologies, enjoy this joke: An accountant, a lawyer, and an automation engineer are led to the guillotine, having been convicted of grievous crimes. The accountant is first in line, but the falling blade jams halfway down. The executioner tells him, “You may be unethical but my professional ethics require me to spare your life.” The lawyer is second in line, and again the falling blade sticks and the man is spared. Finally, the automation engineer is walked up, and he confidently addresses the executioner, “I know what your problem is. It’s the actuator.”
With the range of technologies described above, it is highly unlikely any single engineer could be that specific. You also hope he is less exuberant:)
Contract manufacturers and systems integrators will need to work together to develop customized machines. Similarly, on the customer side, process and automation engineers will need to collaborate closely. None of this will be cheap.
I look at my county and how much planning and investment it took to put a robotic arm on our garbage trucks. Every household had to adopt sturdy bins which the county subsidized. The drivers had to undergo significant re-training. All that to eliminate maybe 100 jobs of which used to pick up garbage bags at each house and toss them into the compactor on the truck. Not a complex job by any stretch of the imagination.
So, before you go and plan for such “Frankensoft” automation solutions, check if customers are truly asking for it and at what cost? Best I can tell customers want automation that can target what I call 3D — dull, dirty and dangerous — tasks, not complete jobs. Deliver that and we will have happy customers and grateful workers.
Moral of the story — don’t over-engineer. Keep asking the wise question “What do it do?”
As for those who confidently project jobless societies, perhaps they should go talk to practitioners, regulators and others with first hand experience with automation. The expectations are fairly modest after years of hype cycles around AI, robotics and other technologies.
Cross-posted at Medium.