The Scary Era of Artificial Intelligence
Disclaimer: Luring but sarcastic headline
Humanity has been plagued with thoughts of machines taking over the earth. The ruling species could hardly get past insecurities of getting ruled out. I live in a developing country which is not at the forefront of the cutting edge technologies we keep hearing around, and I am not in any way more than just a rookie observer. But I do read up a bit on how hypes get carried on for decades, only to be forgotten quickly. Hollywood cashed a fortune selling stories on alien invasions, world-ending catastrophes, imaginative pandemics, and now, evil humanoids. Thanks to Pixar, we got to see a romantic and happier version of robots in Wall-E, but the likes of Ex-Machina are just too many to ignore. But practically, is it a realistic fear or just food for thought. I won’t (and can’t, to be honest) get into details of how the technology landscape is evolving, but I can surely throw in some surface knowledge to back my arguments.
Before getting into the advent of AI in our regular lives, I wanted to counter a common statement of how machines are making humans lazy and dumb. This also forms a base for fears of being outwitted by cold and calculating androids.
Let’s start from the start! If we rely on the Darwinian Theory of Evolution, Homo Sapiens made it to today because, perhaps, a long time back, some of us folks decided to prefer channelizing energy towards mental development over physical. We don’t stand a chance in the wild if not for the tools we devise. Yet, we are the primary danger to the entire ecosystem. How did we ever do that? At the risk of speculating, because of the decisions made (mind over matter), humans evolved at a much faster rate than what nature could handle. And we are nowhere near to slowing down. We may find a solution to counter this (electric vehicles, outside-earth sustenance, etc.) or we may not. But that is how we have been for millennia now. Good or bad, it flows in our blood and it defines us. We became lazy when we stopped hunting and discovered agriculture; we became lazier when we invented wheels and put animals to work. Carbohydrates used to be a rare commodity for early men, and they found a liking to it. Who would have thought that liking would become an addiction as we started pandering to our palate, and end up inventing diseases like obesity?
How is then that we are doing anything new? The rate of development will keep compounding. It took us tens of centuries to get from eating raw fruits and meat, to start consuming cooked food. It took us centuries to move from candles to lightbulbs. It took us decades to replace Turing Machine with a Supercomputer. It took India hardly a decade from getting comfortable in texting to start catching fantasy animals on the streets (read Pokemon Go). And the velocity of innovation will keep on growing exponentially, now that the whole world is connected through a giant web.
It is a piece of common knowledge that the techniques to teach a computer in making decisions existed way back, in theory. We didn’t have the hardware to process them back then. Soon, conventional Machine Learning models started to evolve in different domains, spurring talks of automation without explicit rule definitions. But it wasn’t until 2012 (when Geoffrey Hinton made a breakthrough in Deep Learning) that AI moved from computer labs, eventually to our bedrooms now. Alexa started becoming a household name and we became acquainted with chatbots. No wonder mankind started getting a bit scared.
So a machine learns with data. And we are generating copious amounts using the “free” apps that we love so much (read Google, Instagram, Whatsapp, Facebook, etc.). We started developing Edge computing so the brain sits where the body stays. But why is then that the state-of-the-art, ready to use ML tools on the cloud offered by the tech giants (AWS, Azure, etc.) focus only on a particular type of data, which is production-ready, and not all. I highly doubt even the best food recognition API in the world could always tell the difference between a plain Donut vis-a-vis a South India Vada, something a child could be trained in perhaps a few minutes. Talk to Siri about your significant other and then ask if he/she is the one for you; not a surprise as to what could ensue there. Nothing. We are constantly on the move to being better at technology, but still far from perfection. If we are going towards an apocalypse or paradise is something I doubt I can know in the near future.
One can argue that it takes a lot of years for a human to get trained to take mature decisions, but is it just that? Or does it have something to do with what we are not even daring to replicate, leave aside experiment with? Reproduction, of course!
We carry traits of our parents, something which trails back to our age-old ancestors. Humans still take years to grow up but look at a street puppy, and you will know it didn’t take much training for it to learn how to bark, walk, swim or bite. Animals have it in their genes and it gives them a jump start to keep evolving if at all they evolve further. Humans have a DIY kit ready to learn and build. DNA as a structure still fascinates us, so it is safe to assume we are still quite distant from understanding how to replicate a human brain in totality. Artificial Neural Network as a name is just that, a name. We are more than just a network of nerves. We are a product of wisdom evolved and passed through ages. We devised myths to teach good vs bad. It will take more than just an unbiased dataset to teach a computer to be unbiased.
Just like force-feeding vitamin tablets won’t make an Einstein out of anyone, just the fabrication of a faster system could not be the only parameter for an Artificial General Intelligence system’s development. We, humans, forged a sustainable intelligence in a cost-effective manner. I suppose there might be something still better in AI methodology than what we have just now.
The applications of AI nowadays is to automate tasks that a human does, mostly without invoking a lot of cognitive muscles. AI is trying to take the mundane tasks out of our hands so we can focus on bigger questions and maybe evolve further. Maybe it’s time we start finding answers to bigger questions of life. A lot of jobs may be disappearing, but many interesting profiles are getting created as well. It is all about the perspective after all. In my personal opinion, hitting the gym for half an hour should still be enough to not let the body become rotten; not necessarily do we need to all physical labour just because our ancestors did so. We can still do activities we deem fit, but that shouldn’t be a compulsion in the fear of becoming weaker.
