Close to the edge
Versão em Portugues: https://medium.com/@adsantos/close-to-the-edge-c886baab6a2e
Dr. Toby Walsh, the leading scientist in artificial intelligence (AI) in Australia, predicts that in 2050, among other things, humans will not be allowed to drive. That makes a lot of sense, in the light that it is an activity that demands constant attention and discipline, things in which humans are not very consistent.
Machines, on the other hand, generate results very consistently: they don’t get distracted by text messages and can anticipate problems, interacting with other machines and reacting in a fraction of a second, much beyond the human capability (already discounting the use of cellphones while driving).
But, in his predictions, Dr. Walsh doesn’t mention what will happen to the people that program those machines. We, humans, are subject to making mistakes, even if our processes are good (scrum, bdd, devops, etc.) and our engineers qualified.
Will humans, in the future, be allowed to program machines?
Since the Industrial Revolution, we have multiplied our physical strength, we do it faster, with more precision and the predictions that we would be out of a job by now haven’t come to fruition. It’s true, jobs have transformed themselves, at least for the middle class. What we see today in machines is a capability that was innate to humans: they couldn’t write, they couldn’t paint, they couldn’t compose, and they couldn’t write their own algorithms, until now.
There is no space for an exhausting list, but let’s take AIVA, an AI capable of writing music (good or bad, judge it yourself), as an example. Its compositions are used for games, events, and videos, by companies like NVIDIA, Vodafone and TED.
In bot or not, you must tell if a piece of poetry was written by a computer or a person.
A painting, generated by an AI, was sold for four thousand and thirty-two million dollars.
The service AWS Textract is capable of reading any type of documents and extract content from forms and tables.
Maybe something closer to the development of software, the tests. Practically every company involved in testing is including some capability of AI, be it in learning the use of an app, creating unit tests or helping the developer to increase the code coverage of their tests.
Most of the solutions are still limited, like self-driving vehicles, but with time and proper investment, they will achieve the appropriate level. Extrapolating, once an algorithm of AI can write it’s own tests, basing it on its possible inputs and outputs, what will stop it from writing the code that satisfies these tests?
As of now, an AI increases the capability of a software engineer, providing tools, tips, and suggestions, but in a few years, it will be able to write it’s own algorithms like it already does in image, natural language, and shopping recognition.
It’s almost poetic justice. We, software developers, are responsible for, directly or indirectly, eliminating thousands of millions of jobs throughout our history, we gained society’s respect like the scribes of the past. We know how to talk to machines, to tell them how to grant our wishes. We have this “gift”, nothing happens without the involvement of one of us. We are modern scribes with the power of altering the rules in our favor.
But we taught the machines to talk, to understand what we want and the users won’t need us anymore. The machines won’t need us to teach them anymore, they will learn by themselves, they will evolve independently and will continue where we left off (Dr. Walsh’s number one prediction).
I talk to many software developer colleagues that don’t believe that there will be a day in which we will be replaced. Maybe they are right, maybe our conversation with machines will achieve a higher level of abstraction. After all, we stopped flipping keys to writing zeros and ones, then a mnemonic language (assembly), structured language, object-oriented, and frameworks. However, I believe that the next level will affect the way we develop software radically, in the same way, self-driving vehicles will affect professional drivers.
How will the jobs in the future be? How will we get paid? Those questions are still unanswered. Maybe the creative idleness that Domenico de Masi talks about in his book is the right path, or maybe a guaranteed basic income, or the 24th century Star Trek economy. But as long as the right questions are being made, we will be closer to the answers. And yes, we will see the edge and a safe way down.