The Rise of Artificial Intelligence
Computers are quickly learning to think, and we may well be at the cusp of one of the most important technological watershed moments since the invention of the wheel. Artificial intelligence is a term that’s been thrown around since the genesis of computers, and is associated with science fiction by many, but the advent of hyper-intelligent, thinking software may be closer than we think. It will change everything.
Consider the tractor. The tractor is a perfect example of artificial muscle. It accomplishes what used to be a physical, human task, and does it far more efficiently. It replaces and outperforms human beings, just as cars have replaced and outperformed horses, and mechanized assembly lines have replaced and outperformed run-of-the-mill laborers. Artificial intelligence (AI) works the same way, but with intelligent thought.
Intelligence is defined as the computational portion of the ability to accomplish goals. AI seeks to create physical technology and non-physical software that can do this efficiently. Narrow AI, designed to perform and adapt to specific tasks, already exists today. Self-driving cars, facial recognition software, and the Spotify algorithm that recommends new music to you are all artificially intelligent.
What sets AI apart from an assembly line robot is an ability to learn. An assembly line robot is programmed to perform a specific task again and again, regardless of changing circumstances. Its goal is to perform its programmed action. The result, or any data outside of this pre-programmed action, is irrelevant. Artificial intelligence can not only collect data, but also draw deductions from it. AI adapts its methods in order to better meet its goals, and can do so repeatedly, quickly, and autonomously. It has the potential to learn and adapt at an exponential and mind-boggling rate, meeting and outperforming human beings in a massive array of tasks. Give a piece of artificially intelligent software internet access and it will be able to pull in an endless stream of data to constantly (literally constantly) study and change.
So what’s next?
If automation is possible and cost-effective, it will happen. The transportation industry employs tens of millions of people worldwide. Artificial intelligence is going to replace every single one of those jobs. Self-driving cars are just the beginning. It is only a matter of time before every form of transportation is carried out by artificial intelligence because it is cheaper, statistically safer, and more efficient.
Artificial intelligence is going to replace white-collar, high-skilled work as well. Computer programs are already capable of sifting through stacks of legal briefs and drawing effective conclusions at a pace that crushes any team of lawyers. There’s no reason software capable of diagnosing patients with illnesses, and then prescribing treatments, won’t continue to improve and inescapably alter the medical field. The stock market is already largely automated, and positions in finance and banking are going to quickly go the same route.
Unnervingly, AI is capable of writing, painting, and composing pieces of art at a level that is identical to that of human beings. Aiva Technologies has developed Aiva (Artificial Intelligence Visual Artist), an AI capable of creating music that has been used for soundtracks in film, advertising and video games. Aiva has an album, entitled “Genesis,” out now, and is recognized as a composer. Heliograf, an AI developed by the Washington Post, has written and published hundreds of articles for the newspaper. A team of researchers from Rutgers University, Facebook’s AI lab in California, and the College of Charleston have developed an art AI able to develop new styles of painting. Questions of artificial art’s value compared to a human’s aside, the rise of artistic AI will change the creative landscape. Artificial Intelligence will have an impact on nearly every part of human civilization. It’s only a question of when.
At the 2015 Puerto Rico AI conference, a poll of attending researchers was conducted to determine just when AI will be capable of meeting and out-performing human beings. The answers varied. Some researchers guessed hundreds of years or more, but the poll’s average answer was by the year 2045.
Beyond its sweeping effect on the job market, the dawn of super-intelligent AI poses a great deal of possibilities and dangers. Contrary to popular myth, the risks associated with super-intelligent AI do not stem from the possibility of such technology becoming conscious and then evil. The primary dangers surrounding artificial intelligence are related to its goals.
AI can be programmed to do almost anything, and then do it better than anyone. In certain instances, that’s amazing. In other respects, it’s horrifying. The rise of AI will lead to an arms race in autonomous weapons, which will have an ever-expanding and increasingly independent capacity to destroy and incite chaos. What happens when AI can be deployed to topple governments, launch missiles, sabotage infrastructure, or purposefully annihilate natural resources? What happens when these weapons become too intelligent for even us to understand, capable of creating their own weapons, imagining their own strategies without our consent? Autonomous weapons might well place us in a position similar to that of when we invented nuclear weapons: at the helm of a technology with a destructive radius that cannot be fathomed or responsibly wielded.
Far more chilling are the outcomes resulting from a misalignment of goals between humans and AI. If you program a self-driving car to take you from point A to point B as fast as possible, you might well arrive at point B injured or dead. The AI has misunderstood you, doing what you asked, but not what you truly meant.
The dangers of AI stem not from malevolence, but misunderstanding. We may reach a point where the intellectual divide between human beings and artificial intelligence is proportional to that of an ant and Stephen Hawking, and we don’t consider anthills when constructing skyscrapers. Our relationship with AI might become similarly and insurmountably alien.
It is also worth noting that while AI learns from cold data, data is not free from bias. If an AI is being fed misleading, or racist statistics, then the outcome of its goals will become inadvertently and effectively destructive for marginalized people. The data being provided to artificially intelligent policing systems will need to be heavily vetted and scrutinized to prevent such an outcome.
None of this is to say that AI will lead to an apocalypse. When carefully put to good use, artificial intelligence technology has the potential to perform near miraculous feats for humanity. But such a surge in hyper-intelligent tech has the potential to inflict whiplash on our world economy, infrastructure, and mass-consciousness. It would be unwise not to tread lightly into this new frontier. The AI wave is looming.
Learn to code.