You can fool all the people some of the time and some of the people all the time, but you cannot fool all of the people all the time.
While many historians continue to debate whether Abraham Lincoln was really the original author of this well-known adage, there couldn’t be a more pertinent expression to describe the most advanced AI (Artificial Intelligence) bots that you interact with today. They are designed and trained to address a finite number of specific use-cases as effectively as an intelligent human would, or at least should, but they cannot always trick everyone into believing that it is a human on the other side. Technologists call it Weak AI or Applied AI, and it covers the whole spectrum, from your grandfather’s pocket calculator to your most trusted personal assistant Siri.
The AI (displacement) wave
If we have been arrogant enough to name the present AI as Weak, then why do we fear it? Well, that paradox arises from a difference of perspective.
In 2010, a few months into my first startup which provided recruitment services for niche sectors, we had our first Eureka moment! We came up with the idea of automating ourselves, the complete cycle from understanding the job profile, finding the relevant candidates online, ascertaining their interest, and scheduling their interviews with a client. We were super excited by the thought of building the first fully automated recruiter bot. It took us a few hours to realise what it meant for our whole team of recruiters. We were going to build a system that automated the very skills that we had trained them for, and it was really the only way for us to create disproportionate value using our technology background. Until then, we had only read case studies about the workforce displacement caused by industrialisation, but doing it first-hand was much tougher than the lines of code we wrote in the months that followed.
Being a kid of the 20th century and an innovator of the 21st century, it feels conflicting to see my peers trying to build bots for stuff that were professional aspirations of my childhood friends. From travel planning to tax filing, we are now in a race to automate everything of necessity. The underlying logic is frighteningly simple: if you can train a human to do it, you can build a machine to learn it.
A better world for whom?
The cocktail of robotics and artificial neural network is likely to replace 40 percent of the current workforce by 2030. For whom would that be a better world to live in? A simple cold-hearted response would be the remaining 60 percent, but that would be a myopic viewpoint. Industrialisation almost completely eliminated the need for physical labour within a matter of a few decades, yet our entire generation is thankful as it has drastically reduced the per capita human effort required to survive comfortably on our planet. Similarly, while applied AI will continue to rapidly replace the analytical workforce in the years to come, our children would inherit the world where creativity and imagination would form the core of any work done by a human, and almost everyone on the planet will have the security of affordable healthcare.
If you are arguing against AI, then you are arguing against safer cars that are not going to have accidents, and you are arguing against being able to better diagnose people when they are sick.
– Mark Zuckerberg
The AI bias: a reflection of ourselves
So, will Applied AI eventually deliver perfect results for any task that it has been built for, and will it be completely free of cognitive, psychological, and informational biases? The short answer is NO. This is not because the AI being built is faulty in nature, but because bias is inherent in the data that is used in training the AI, and even machine learning can only help it optimise itself to the extent permitted by the behavioural data of users. That is why your personalised news feed tends to mostly confirm your existing beliefs rather than challenging them to enlighten you. A chatbot continuously bombarded with racial slurs soon starts exhibiting racist tendencies itself, and even the most advanced trading algorithms are never able to time the next crash in the stock market.
AGI: the birth of a new species
All the AI that you have seen until now, from receptionist robots to personalised chatbots, are neither self-aware of their existence, nor have the capacity to experience pain or pleasure. AGI or Artificial General Intelligence is the pursuit to create an entity that can sense, feel, think, reason, communicate and act like a human. However, it will not be subject to the same biological limitations of survival, reproduction, physical strength, and mental capacity. It would only be bounded by the energy that it can extract from nature to power itself and the matter around it, and hence it is expected to be far more potent than any human being. It will be the first-ever species that will be created by a process of intelligent design, rather than natural selection. Science fiction movie writers have envisaged numerous scenarios that might arise if and when we create this new species. While it is near impossible to predict whether any of these scenarios would eventually pan out, the real clues might lie in the books of human evolution and in the prevailing state of our civilisation.
AI is the rare case where I think we need to be proactive in regulation instead of reactive. Because by the time we are reactive in AI regulation, it’ll be too late. AI is a fundamental risk to the existence of human civilisation. – Elon Musk
A utopian dream or a dystopian nightmare?
In the couple of million years of human existence on this planet, a vast majority of the time for a vast majority of the population has been dominated by the primal struggle for survival and mating. The last few millennium have seen the emergence of a variety of cultures and philosophies, which have defined the modern psychosocial pursuits that presently dominate our civilisation’s collective psyche. Our minds are now systematically subject to multiple schools of thought with each persuasively trying to instill in us a unique purpose of existence, and there is even a liberating existential school of thought that leaves it to us to give life any meaning of our own choice. Will the AGI that we create objectively choose between these human established schools of thought, or will it be primarily influenced by its creator? Will it further evolve newer philosophies of its own, and will this new super-intelligent species ultimately retain any form of empathy towards the weaker species it originated from, we humans?