Swami Gulagulaananda said:
“Learning from experiences is supervised learning… Guessing by yourself and being right is pure bliss"
In an eCommerce company, a digital marketer typically creates ads on social media websites such as Facebook. These ads are paid, and the charges are based on the number of times ads are displayed to visitors (called impressions) or by the number of times that a user clicked on these ads (called clicks). Eventually, if I have to understand the performance of an ad, I will measure it against the total sales and revenue generated by that ad. While the social media website gives me metrics such as impressions and clicks, it cannot give me the sales and revenues realised because that is happening outside their realm (in my realm, my website). So, if I have to measure the true efficacy of a campaign, I have to collate data such as number of impressions and clicks from the social media website with data that I collect on my eCommerce website such as views of target page (product page), add to carts and checkouts. This allows me to measure conversion rates.
Now, the digital marketer’s job is to observe this data and take some decisions. If an ad is performing well - that is, if it is sending a large number of people to my website, then, I will push more money into that ad so that it reaches even more people. On the other hand, I’ll probably reduce money into an ad that’s tanking. A software developer who looks at this scenario can quickly observe that there is an opening for some automation here. Since a program can read all of this data, a program can also be coded to arrive at decisions. Simplistically, you can add certain thresholds to add money or remove ads. However, in today’s world which is replete with buzz words such as big data and machine learning, one can build a truly sophisticated system that considers a host of external data such as determine the quality of ad based on the graphic (by performing an image analysis), text (by performing text analysis) and external factors such as time, geography, interests, past performance etc.
Cutting down on the technobabble, it is perfectly possible to replace that digital marketer with a program. This seems fantastic, but is definitely doable. This program that you use will be significantly better than the person because it has a huge amount of data based on which decisions are arrived at. Let’s look at a few more examples where machine learning (ML) and artificial intelligence (AI) could be used. Self driving cars are already making news all over the world. Doctors could potentially be replaced because most doctors, at least in the Western world, rely heavily on tests and scans (watch House MD if you don’t believe me). We could have a machine which can scan our body, vital statistics such as temperature, BP etc., extract a few drops of blood and run a bunch of scans and then run the numbers against a huge database that it has built from data worldwide and arrive at the diagnosis. This diagnosis will be more reliable, perhaps, than doctors because of sheer volumes of data that is used.
Taking it a step further, ploughing a farm and planting seeds is not that complicated once you are able to build self driving cars.
You see, most of the jobs can be automated in the long run. The question that we need to be thinking about is - Should we let this happen?
On the one hand, it seems obvious that this should be the way forward. The technology will be top notch, cheaper, far less error prone, the experience will be significantly higher than any individual, etc. All of this makes it seem like this is the future. However, there are cons. Think about all the people who are going to be losing jobs. Let’s take one example of drivers. In India, professional drivers are the ones who drive buses, trucks, minivans to transport smaller quantities of goods, autorickshaws, taxis like Ola and Uber, school vans etc. If we automate these jobs, what are they going to do? Surely find other jobs, you say?
When I had been to the US, a toy drone that I wanted to purchase cost me one third on Amazon than in a brick & mortar store. When I asked the salesman why it was priced three times higher, he told me about electricity, salary and other overheads. It immediately made sense why people bought on Amazon and why their flywheel model works so well in the US. This salesman will be fired eventually because his company cannot compete with Amazon. What will he do?
It appears that most of the jobs that they can pick can eventually be automated… If automation and companies like Amazon & Flipkart rule the world, we get a lot of advantages. But there are also a large number of corresponding job cuts. If all the mom and pop stores, our friendly local department stores get closed, we will have a large number of jobless people.
A counter argument is - Well, when these companies grow, they will hire more people, won’t they? Automation is simply transferring jobs from one place to another. While this is true, the number of jobs created is different. 500 people may lose their jobs for 50 new jobs created for fulfilment.
What if all these jobless people start resorting to crime to fill their stomachs? We have already been reading news about disgruntled drivers kidnapping people to take revenge on ride aggregation companies and others following Silicon Valley employees and heckling at them… Will our society spiral towards its doom?
A good thing about eCommerce companies like Amazon and Flipkart, in India at least, is that they are playing nice with third party sellers who are benefiting a lot from an additional channel of sale. Similarly, Ola and Uber are giving a lot of structure and business to cab drivers. Our country and society can grow only when we all grow well together. Automation should not be shunned and technological progress should not be stopped - But before implementing them blindly, we need to consider a lot of things - such as their impact on society.
I thought about this recently because of a news item about the government’s decision of not having driverless cars in India. I remembered this video that I had watched long back titled “Humans need not apply”. I have attached it here for your viewing pleasure: