Is AI going to kill us?

This is not a new topic or a new discussion. There are multiple arguments towards both sides and some pretty high profile people who have strong inclinations on this subject.

The idea around writing this article is to pen down what I, personally, feel about the same. I am not an expert in the field but as the topic has more and more people speaking on both sides, you tend to listen and you tend to question. What I am going to write below is to make you think and see if you come to the same conclusions as me or if there are any other conclusions that you arrive at.

I would love to hear what your thoughts are after reading this and would request you to pen down your comments, as brief or long as they may be in the comments below.

LET'S DEFINE AI

Lets get down to the topic at hand. Firstly, lets talk about AI and what it is. Let us agree on a definition so that we can a basis for the discussion.

Machine Learning or ML, as it is often referred to, is largely considered to be the first step of the journey towards AI. In fact, the marketing geniuses that we have today, tend to sell and market ML solutions as AI. In my view, ML is the basis for AI.

Taking it to the next step, and to address the headline topic, for AI to pose a threat, it has to achieve "consciousness". This is also referred to as AI or Technology Singularity. In lay man's terms, it means that technology/AI has grown to a place where there is no turning back. It is also the time when AI becomes "Self-Aware". This is the topic of discussion.

WHAT DOES "SELF-AWARE" MEAN?

Do you remember when you started speaking? Do you remember the first word you spoke? Do you remember when you learnt to balance and walk on your own two feet? For almost every one of us, these are things that we do not remember inherently. But if you ask people who were around us, they would. So, as a human, if you ask someone if they are self-aware, it is a difficult question for people to answer, let alone answering when they became self-aware.

WHAT IS DIFFERENT ABOUT US?

There is a popular saying that no two human beings are alike. While you may meet people who think along the same lines as you, it is neigh impossible to find someone who is exact replica of you behaviorally.

Why do we not find humans who are exactly similar? Our biologically built is almost always the same for everyone. People may say, it is how you are "brought up" and how you were taught while you were in your formative stages.

ML TRAINING MODELS AND BIAS

If you are a student of machine learning, you will be taught that you need to avoid bias in your models. This is to ensure that the machine you are training is able to learn and make decisions devoid of bias. But is it possible to truly eliminate all types of bias? The most common bias types that get into ML training models are -

• Automation Bias
• Confirmation Bias
• Group Attribution Bias
• Homogeneity Bias
• Selection Bias
• Reporting Bias

Advocates of ML are constantly refining models to ensure that bias is appropriate. This is similar to how humans are brought up - we are constantly feed bias as part of our upbringing. Simple example, what may be construed as right and acceptable in one part of the world, can be seen to be wrong and unacceptable in another part of the world.

So, similarly, each ML algorithm has a certain amount of bias and this can change depending on who is refining the algorithm.

CONCLUSION

There are several people all over the world working on AI models. Depending on the pace and advancement currently, Self Aware AI may be a few years away or a few decades away. But one thing that also seems to be true is that once AI becomes self aware, it will have a certain bias based on who and where it was modelled. Similar to humans, no two AI models will be the same.

While one human may decide that he wants to kill his neighbour for partying loudly and making too much noise, another may want to go and join the party.

What will the first Self Aware AI do? Join the party? I hope so, what do you think?