Hey!

Welcome, and you can check out my posts. But, I don't write here anymore. So, if you are interested, come on over to https://sandhyavaradh.com/
Showing posts with label AI. Show all posts
Showing posts with label AI. Show all posts

Wednesday, January 15, 2020

The Human Factor In A Future Of AI

Sometime back I had to write this essay from a given prompt titled "The Human Factor In A Future Of AI", and sharing it here for your reading. Comments are welcome!


The world that we see ahead is one where the line between humankind and technology is thinning down further as every day goes by. The rate of development of technology is happening at such an unprecedented, break-neck speed that it has become nearly impossible for the other complementary fields to keep up. The new talk is all about the world of artificial intelligence(AI)- a realm that even the men who have discovered it are not entirely sure about. 

Artificial Intelligence, to be simply put, is the theory and development of computer systems able to perform tasks that normally require human intelligence. The disruption by AI is already happening around us, and it is believed that this is just a microscopic part of an unexplored universe of self-programming intelligence. It is also expected that AI would successfully replace most of the jobs today, thanks to its high efficiency and almost impeccable accuracy. 

So, in this future of artificial intelligence replacing all that humans can do, in a much more efficient and successful manner, is there any human factor left for us to hold on to? Well, it all depends on objectives and perspectives. 

There are still things, definitely, that a machine cannot achieve by replacing a human being. A machine, whether programmed to gain intelligence or not, operates on pure statistics and algorithms. It takes its decisions and makes networks purely on the basis of logical arrangement of probabilities and mathematical models. What a machine lacks is instinct. And, this is human factor in the future of Artificial Intelligence. 

Since time immemorial, we have always known that evolution has been possible and life exists only because of instincts, and so there is no question on the power of it. AI can make machines intelligent enough to understand our instincts and the human emotional quotient, but they could never have one of their own. To simplify: it might gauge our anger and love, but cannot produce the same. That is where humans score over AI. As Australian scientist Alan Finkel puts it (Finkel’s Law): robots won’t replace us because we still need that human touch. Robots do not understand irrational thought and context. These things can be essentially dealt with only through instincts and empathy that AI will not provide the robots with. 

To sum it up: AI could make robots almost life-like but they would never actually have a life because humankind can be two things that AI-bots could never be: human and kind. And, that is essentially the human factor in the future of Artificial Intelligence. 

Sunday, December 22, 2019

What does the Legal field need to utilize AI?

Image from Timo Elliott
In the legal field, there are enough and more sets of data that can be generated for standard document clauses, online contract clauses, filing of delay petitions and finding out necessary documents to be attached to such petitions, to name a few. 

Here is something interesting that I learnt in the introductory course to Artificial Intelligence(AI) and Machine Learning(ML) by deeplearning.ai. Let’s look at companies in general. It is important to understand the difference between a company that uses AI, and an AI Company. The difference can be understood easily with an example. Let us say that a big store has an online website. Would you consider it the same as Amazon or any other internet company? Just like a company that uses the internet is different from an internet company, similarly a company that uses AI is different from an AI Company. 

To use and incorporate technologies of AI and ML for your business, there are two most basic requirements-
  1. It must be a simple concept.
  2. Lots of data sets must be available for the task. 
Without these two key requirements, ML works poorly where there are complex concepts that needs to be performed from a small set of data or new types of data. Therefore, to utilize the tools that AI provides for your business, it’s wise to create a huge data set, and then execute pilot projects. 

All these are simple tasks, and generating a huge data set to feed for the effective use of ML is not difficult. It would be worthy to consider giving these aspects a thought, in order to efficiently handle the volume of cases, and the delay in filing them, in our country.

Monday, December 9, 2019

The Next Sci-fi Plot?

Image from Clipart

I have a habit of sitting in class, and connecting the laws I learn to weave a story, a fantastic legal thriller. Perry Mason and John Grisham have fed in too much imagination. I remember sitting in the Contracts class in my second semester in law school listening to the lecture. Our subject of discussion that day was s. 10 of the Indian Contract Act which talks about the validity of a contract or an agreement. Our professor casually threw at us a statement- what if in the future intelligent computers make contracts? Will the same validity criteria hold good then, too?

Of all that he said that day, my mind took fancy to this single statement. A simple, amusing thought: fast forward into the future and see through the lens, to observe two computers placed royally in a five-star meeting hall negotiating and entering into a contract for their respective parties. Good plot for a sci-fi novel, with a villain and a few twists and turns?

Jargon like AI, Machine Learning, Deep Learning, and over-exaggerated threats of an army of computers taking control and wreaking havoc on humans, are all very popular. But how many of us actually know what they mean? I did not, though I was well aware of all the jargon in use. To further explore and understand if this is even possible, I did a small introductory course on AI.

AI can be broken into two parts Artificial Narrow Intelligence (ANI) and Artificial General Intelligence (AGI). While there has been tons of progress in ANI, with all the self-driving cars, speech recognition, visual inspection devices, spam filtering applications, etc., there has been next to no progress in AGI.

So, here’s an imperfect rule I learnt during my most basic quest to understand these terms: anything, any job that we do under a second of thought, is automated or will soon be automated. So, where there is a series of tasks one after another which takes an hour to complete, where each task only requires a second of our thought, it can be automated to be efficient in time, resources and produce an almost-error free result. For example, in speech recognition, though translation of a 15 minute podcast may take us sometime, every word we hear takes us less than a fraction of a second to be understood and written. So, it has been successfully automated considering other factors such as language, accent, etc.

Now, connecting these aspects to law, we need to ponder on two important things.

- What can technology do and what can’t it do in the legal field?
- What proportion of the lawyer’s roles does it take over?

Most clerical jobs such as calculation of stamp duties, searching of case laws, etc. has already been automated and there are a numerous menial tasks through the day of a lawyer that can be further automated.

But can two computers make a contract? The general clauses that exist in a format in every deed of a particular type may be automated, but the negotiation of clauses cannot be done under a second of human thought. Also, there are several unknown extraneous situations that come up between the two parties. And, since there is no progress in AGI, it may be next to impossible for computers to gauge such aspects as of now or in the near future.

But, in a hypothetical situation, if the computers were able to make contracts in representation of their human parties, what would be the criteria of a valid contract? Seems like a question that I’ve thought real less about, as I set around understanding if that is even possible. I can at least say that I do know what the terms mean and their most basic logic. But some good food for thought, isn’t it?

Endnote: I think it would be an interesting sci-fi plot. Should I unleash the imaginative writer in me and give the world its next bestseller?