If Machines Were People, They would Look This Stupid

by on Jan 13, 2018
machine learning

Imagine you appoint a chauffeur to drive you around. You tell him he has to report by 9 a.m. and drive you to work by 9.30 a.m. every day.

Day 1:

It’s 9.10 a.m. and the chauffeur has not reported yet.

You pick up the phone and call him. He tells you he has arrived. You are puzzled as you don’t see him. You ask him and realize he has gone into the wrong street.

You explain, “No…no…no…it is street number 21, the street adjacent to the little pharmacy.”

9.15 a.m. and yet no sight of him. And then you get a call, “Madam I am came to street 21. Which exactly is your house?”

You tell him, “Okay come down to that white color house on the right, the one with a large black gate and two coconut trees overlooking the street.

Finally he arrives and you ask him to drive you to work and explain to him the location.

There are different ways you can reach work, but you have this preferred route. So you give him directions along the way, and finally reach work.

Day 2:

It’s 9 a.m. and the chauffeur has not reported yet.

You call him. He tells you he has arrived. You are again puzzled as you don’t see him. You ask him and realize he has gone into another wrong street, though not the same one as the previous day.

You explain, “Dude, it is street number 21, the one next to the little pharmacy.”

9.05 a.m. and yet not sight of him.

And then you get a call, “Madam I am came to street 21. Where is your house?”

You get mad and scream, “Don’t you remember, the white color house on the right, the one with a large black gate and two coconut trees overlooking the street.

Finally he arrives. and asks “Madam, where should I drive you to?” You ask him to drive you to work.

He asks for the location. You wonder what is wrong with him. Nonetheless you tell him it is the same place that you went to the previous day.

He starts driving, but does not remember your preferred route. You have to repeat every single direction that you gave him the previous day

Day 3:

It’s 9 a.m. and the chauffeur has not reported yet.

You call him, he has again gone to the wrong street. You again explain every landmark and every detail, and he finally reaches your location and again asks where he should drive you to. You again snarl at him, “Work”.

He again does not remember the location or the route and you have to go over it again directing him about every left turn and every right turn and every short cut.

Day 4:

The same story repeats all over again.

Day 5:

The same story repeats all over again.

Days, weeks, months, and now years have passed, but this continues to happen. Forever.

Though a personal chauffeur is not dumb to give you the kind of experience mentioned above, just imagine if this were to happen, how would you feel?

Well, this is exactly what happens every day if you use taxi apps!

You have to enter the drop location each morning, call the driver to give directions to reach the pick up location and then give driving directions along the way to the drop location.

Should they not be saving landmarks of pickup location to help the cabbie reach there without the customer having to call and verbally explain things?

Should they not be learning that the person goes to work at that time every day and thereby preload the drop location with an option to edit, just in case the rider is bunking work and going to a cinema?

Should they not be remembering the preferred route that gets you there faster, so the rider does not have to do back seat driving every single day? In case you are distracted by a message on your phone or something else, and miss giving the next direction, the driver will follow the default GPS navigation, which means you will reach work ten minutes late.

Maybe shopping apps are more intelligent?

When you go to a brick and mortar shoe store, the first thing the salesman does is checks the size of your foot and then gets back to you with various models in your size.

What if the salesman shows you all the shoes in the store? After painstakingly sifting through them, you tell him what you like, and the salesman at this point tells you to refer to a flawed size guide to check your size. He then informs you that the ones you liked are not available in that size.

Sounds ridiculous, doesn’t it? Well, that is exactly what happens in online stores.

You are shown products that may not be available in your size. Apart that, many products don’t fit true to size.

Shouldn’t they be asking for size information at the outset when users create an account with the website, and then display products according to that when a user searches to buy stuff? They should be asking actually body and foot measurements and not labels such as Large, Medium, 38 UK, 7 US, 9 EU, 10 India, etc. which are misleading and not uniform across different manufacturers and retailers.

Shouldn’t they be learning about color and style preferences of individual users and then showing products based on the intelligence?

We know by now that the machines behind ecommerce stores don’t learn. Even that online grocery store does not preload your basket with your regular weekly purchases, nor remember the choice of your milk brand.

Perhaps food apps are more sensitive?

If you frequently visit a restaurant and order something repeatedly with some custom instructions, executives will likely at least try to remember and follow the instructions in the future. They may forget, because they have only so much memory as opposed to the large volume of customers and their varied preferences.

Food apps, however, neither try to learn nor remember.

When I repeat the same order, be it from the same place or a different one, I have to specify my preferences each time. I wonder, didn’t the machine learn about my preferenes yet? It is not limited by memory issues like human sales reps are. It should be suggesting, “Hey Anuradha, you like your bhel puri less hot, don’t you?” And when I confirm, it should be saying, “Don’t worry, we’ll make it less hot!”

Sorry, no! Food apps are no more intelligent than than taxi apps or ecom stores.

What is all the hype around machine learning and artificial intelligence about?

Machines can’t see your face and read your emotions, so they should not be expected to have a high level of emotional intelligence. But in some aspects, such as the examples mentioned above, they can fare better than humans, thanks to big data and the learning capabilities of machines.

Almost all software claim they use artificial intelligence for better performance. Yet most machines are devoid of even basic intelligence and come across as amoeba with a lot of functionality.

If machines were people, you would roll on the floor and laugh till your tummy hurts, if  these experiences happened to your best friend, but fire the stupid machines if they happened to you.

A departmentalized approach to product development does not help one bit in this regard. It is emotionally intelligent people who can find opportunities for training machines. Those with all-round experience on product, training, customer experience, data, etc, along with a basic understanding of human psychology/feelings and a grip on the art of communication, will be able to look at things holistically and optimize the intelligence quotient of machines.