How to make machine learning simple for dads

My dad is an agile-minded septuagenarian. If he is not telling a story, then he is most likely developing one. He wraps his experiences in stories, catalogs these stories and tags them well — all in his brain aiding rapid recall at a later time.

Brian Christian and Tom Griffiths explain in their Algorithms to Live By:

“The effort of retrieval is a testament to how well you’ve arranged it: keeping the most important things closest to hand”.

The authors attribute “sporadic latencies” in retrieving information to the inefficiencies that creep into the human brain with age in indexing, sorting and storing vast amounts of information collected over time. As for my dad, he is exceptionally efficient in indexing, sorting and storing well formed stories about his profession (Pharmacology), Sachin Tendulkar and of late, #NarendraModi.

Finally, the long awaited #Model3 arrived a few weeks ago. We got the car through a long and pre-dominantly online buying process. On Sunday, we decided to drive to Wente. My dad wanted to add Wine Tasting to his catalog of experiences. And I wanted to experience the autonomous features of the 3.

With ~25 years of driving experience under my faux-leather belt, I drive with intent and nuance. I love nudging the steering that precise wee bit and then providing the exact resistance to let it recoil as the car glides into the adjacent lane. I also love making digital speedometers count down 10,9,8,7,6,5,4,3,2,1 before the car stops. The car stops without the slightest palpable jerk as I release the accelerator and apply the brakes in careful synchrony. I can go on and on, but you get the point — I try to drive with intent and nuance.

With the 3, now the balance has shifted towards eventual control from primal intention. Now I simply but firmly guide the machine as it tries to perform with intent and nuance to find the right nudge, acceleration and timing. So yeah, its a big change to sit back and guide the machine.

I was marveling at this and thinking about how to put this experience-shift into words. That’s when my dad punctuated my thought process. He popped a question — one of those septuagenarian questions we all dread:

“… so I listened to Narendra Modi’s who spoke at IIT Bombay’s 2018 convocation and  he said something about “machine learning”, what the hell is that?”

Where do I start? Whenever I can, I try to be a first principles thinker trying to back-connect concepts to their origin. Doing so takes some time and asking around. But it does give me a kick when I do make the back-connection. My predicament here was unique as I had to help my dad forward-connect. I had to help him make that leap of faith from his current vantage point. If I could get him even a ball-park understanding I’d be happy, I thought.

From his then current vantage point, he was complaining about the “distracting” giant “iPad” on the dash. He was impatiently tracking the moving renditions of cars passing on the dash “iPad”. That was it! My dad had made a connection that was the basis of the foward-connect I wanted to help him make.

I asked him if he would believe me if I told him that the car was collecting a ton of data. Data about the cars that passed us, the cars that were in-front of us, and the ones behind us, and what time of day it was, and our destination and where we were coming from and much more.

My dad said, “Yes, so what?”.

When I told him that this data is sent back to #Tesla, he answered his own question — “… and they use this data to train cars to steer themselves?”.

That was the forward-connect he needed to make to be able to step-back and step-up. From that point on, it was so easy for him to realize (realize, not understand) that everything around him was generating data. And that this data was being piped back through the internet to corporations like Tesla who were using that data to help machines learn. He got it.

His next question was:

… does this car communicate with other cars on the road? Would’nt that make sense?

I knew then that he was in the right ball-park and that he’ll do a great job in regaling his future audiences with his first tryst with #machine-learning.

Cross posted on Medium: https://medium.com/@setu_kulkarni/how-to-make-machine-learning-simple-for-dads-2cc6cd9acecd