The AI era, a call to arms
Talking about AI is a difficult task for me, as it is an area that I really love but frightens me at the same time. In the last decade the growth on this tecnhology was amazing. What might have seemed like a distant future is no longer. Let it be customer support with simple chat-bots or something more complex like an automated car, Artificial Intelligence is already taking a big part in our lives, and we’re still finding new useful ways of applying it.
This growth raises some problems that I want to talk about, but before I bring that up let me give you some context.
Why (only) now?
The researches on AI started in 1950s, and even though there were some advancements it was only in this last decade that the “AI era” started. Why? well, there are a good number of reasons I’ll just point out some.
Lets begin with the Algorithms — First of all, Machine Learning (a subset of AI) appeared and solved one of AI biggest problems as it offloaded its optimisation problems. Its goal is to develop a prediction engine for a particular use case. The algorithm is given a set of information about a certain domain, and weighs the inputs to make a useful prediction. Machine learning algorithms learn through training, the initial algorithm receives examples whose outputs are known, and observes the difference between its predictions and the correct outputs, and then tunes itself to improve the accuracy by adjusting the weightings of the inputs. The more data its provided, the better the prediction becomes.
Then deep learning entered the stage. The first approach to deep learning was in the early stages of AI (at 1965) where it was created an effective multi-layer neural network, but only recent advancements on the algorithm made it revolutionary. For example, they can now recognise objects in images more effectively than a human. Loads of applications started to pop-out, some being very creepy — Adobe announced VoCo, which is pretty much a Photoshop for speech, it’s an AI-powered tool that can replicate human voices, all you need is to give the software a sample of a 20 minutes long audio recording of someone talking and afterwards everything you type will sound just like that voice — magic uh?!
Hardware — Well, it seems kind of obvious that the hardware is important. Advancements on GPUs (Graphical Processing Units) are making the time needed to train the neural networks used on deep learning way smaller than it used to be. Why GPUs you might ask? GPUs are typically used for 3D gaming/modelling, as they use matrix computations which is the same kind of computation you use to train neural networks.
Data — As I previously stated “the more data its provided, the better the prediction becomes”. Neural networks require big chunks of data for training, and data creation and availability has grown exponentially over the past years.
Hype — Both from investors/entrepreneurs and general public, public interest in AI has increased ALOT in the past few years.
The problem(s)
There are a lot of problems surrounding Artificial Intelligence. Starting with skynet, but I’m not getting into that today!
The problem I want to raise awareness at, is how it will affect our workforce.
The most common opinion is that technology “steals” humans jobs. I read somewhere on medium a different way of looking at it. A more positive way. It explained how automation forces people to better jobs. The example they gave was the following -“when sewing machines were created, the number of seamstresses/seamsters might have decreased, but now those individuals have the possibility to become something else”. This theory has two big problems though. First of all, this transition only works on the long-run, as in the close-run this change is so abrupt that there will be many people losing their jobs to machines without having the possibility to get a better one not having the “know-how”. But this leads to something worse.
In the endgame, what happens when machines surpass medics or lawyers, what would be the better jobs then? This is a big problem to solve, and it needs solving now. Big names are talking about it already, Bill Gates suggested there should be robot taxing, which caused a bit of a controversy. Stephen Hawking and Elon Musk also shared some concerns about it:
“The automation of factories has already decimated jobs in traditional manufacturing, and the rise of artificial intelligence is likely to extend this job destruction deep into the middle classes, with only the most caring, creative or supervisory roles remaining.” — Stephen Hawking
“What to do about mass unemployment? This is going to be a massive social challenge. There will be fewer and fewer jobs that a robot cannot do better [than a human]. These are not things that I wish will happen. These are simply things that I think probably will happen.” — Elon Musk
Jobs will vanish fast. We are getting to a point where we won’t need taxi/bus/truck drivers (transportation and warehousing employees 5million americans), amazon is testing a store with virtually no employees and there are tons of examples where humans won’t be needed and will, sooner or later, be replaced. Don’t get me wrong I’m by no means saying we should fight against it, as I think that automation is pretty much inevitable and can be a good thing, but we need to take action and find solutions to help displaced workers while we have time.