Visit blogadda.com to discover Indian blogs

Thursday, January 25, 2018

What should be the focus of Artificial Intelligence?

I've been reading and watching a lot of content related to AI and it's widespread impact on almost everything that matters to human beings. Let's get started with a video I watched last year: it showed that the priestly duties in a funeral was being conducted by a robot somewhere in Japan. (definitely subject to my memory being serving right) The reason being people were reluctant to take that job as it was very low paid. Now this is what I would call "Only humans can, at least now, make appropriate judgement as to what should AI do". Definitely AI would self learn and would automate such judgments as well.

This blog post has been triggered after reading the blog by revered SAP's CEO on AI here: https://news.sap.com/impact-of-artificial-intelligence-machines-cant-dream/?url_id=banner-us-homepage-row7-3blog1-mcdermott-machine-learning-jan18r2
- an extremely well articulated blog. My key takeaways from the blog has been the cautioning advice given to humans and the skills needed in the years ahead. However, one would infer that the blog ultimately hints that no job would remain untouched by AI and suggests collaboration between AI and humans. This would ultimately mean saying "Good Morning" to a robot when we go to workplace and expect it to return the wish. In my view the blog indirectly says that "Machines can't dream, at least for now" as even that is that not unconquerable.

Let's take a step back and understand who is controlling AI now? My mind would suggest it to be humans and technology. So, it's derivable that focus of AI too could be streamlined but only when humans can collaborate among themselves, the biggest challenge of the century, in my view.
It's not the competition between AI and humans, it's more among humans themselves.
Bill Gates, no wonder, suggested to introduce tax reforms on robots considering this.

Let's come to the point that I've in my mind. I live in the capital of India and live in a society having security. Who is the security? They are couple of old men whose job is to open the gate every time a car needs to go out of the gate or a car enters the gate and check if there is an unknown person entering the premise. How many of those in US would know of such job?
This comes to my mind because when I visit US, I stay in an apartment where the gates open on entering the apartment number. The point is who would be benefited if the job of the security guard in Delhi is automated? Would the initial investment cost along with maintenance charges be same as that paid to the guard? Can the old guard take up another not-yet-automated job? Can we even talk about re-skill options here?
Yet another example from the same apartment in Delhi, it has a gardener (again an old man) whose first activity each day is to water the plants? How many in US have seen a person watering the lawns? In the evening, when I go for a walk in US, sometimes the sprinklers don't hesitate to sprinkle water on the pavements and people. How visionary and impressive would it be to automate the old gardener's job by giving him enlightening mantras of up-skilling and re-skilling?

The intention is to highlight the need of a balanced approach for things to stay in control by automation. This would be making policies to combat the vociferous force which at this moment seems more viable that collaboration among humans to decide the right focus of AI.
Just because  a job can be fully-automated should not be the only criteria to start with. The impact, replacement strategy and most importantly the need should be the criteria.

Let's go to the other part of our daily lives - the corporate. Early last year, the client I was working with complained of high backlogs of issues to be fixed in SAP systems. The approach followed by management was to first segregate the issues based on different technical areas and then to see how many of them could be automated. Was there a need for automation? I would have to admit a "Yes". The reason being dependency on humans to work in night shifts when a great chunk done during that time could be automated, should really be done. Most of the time, employees are also not happy by working on activities which does not give them continuous learning. Such areas which involve human dissatisfaction are a great use case for automation. Increased backlog also led management doubt the productivity of the employees. Many would know of tools available which would track and report the activity done in computers and laptops. So, who did this job of tracking the employees before?
So, definitely automation has come into our daily lives.

Let's take the case of human dissatisfaction outside the corporate walls and into general world. In India, it's pretty often that we hear cases of  workers dying during cleaning of drainage systems. It's pretty unbelievable to imagine that a human has to get inside the drainage to clean it. Similarly, there are other perilous job activities, which could even be applicable to only few regions. AI could try to fix problems which are location-specific and need driven.

To sum it up, the Japanese example highlighted in the beginning, serves as the right mindset as to what should be the focus of AI.