Computers are rapidly getting smarter, and it’s sometimes hard to believe the things they can do now that we couldn’t imagine even five years ago. As computers get more intelligent, they can do more things automatically and help us solve problems in ways that were previously impossible. Here are just a few examples of what computers can do today
1) Compose music
While it might not be Mozart, computers can now compose music based on certain parameters, and most of us would have a hard time telling what was computer-generated and what wasn’t. Researchers at Sony CSL Research Lab in Paris created a machine-learning system called Flow Machines that can create musical pieces based on instructions input by humans. The AI then uses its self-developed emotions to develop its own unique style while staying true to your parameters. It took 10 years for engineers to code an algorithm capable of composing original melodies; Flow Machines started off as an algorithmic genome able to generate sounds with various styles and moods depending on how you ask it to perform. With many more years of work needed for improvements, we probably won’t see concerts featuring entirely AI-composed songs anytime soon—but it does provide insight into how machines can take over from human composers given enough computing power.
2) Read Your Emotions
It’s no secret that computers and robots are taking over many human tasks, but there are still plenty of things we can do better than they. One of those is identifying human emotions. Emotional intelligence may not be something that you’ve heard much about in computer technology, but it’s a big part of what makes us special, and it’s one thing that even the most advanced machines will probably never master. Using software to understand or respond to emotions has become more practical with each passing year. A few examples? Artificial Intelligence (AI) might detect your frustration if your device keeps freezing up or prevent a salesperson from selling you something if they sense sadness in your voice during a phone call; it could also evaluate if someone feels guilty when they don’t pick up their kids on time after being late several times before.
3) Diagnose diseases
One of Watson’s first big splashy hits (for lack of a better word) was its success in helping doctors diagnose patients. This isn’t exactly breaking news—IBM has been promoting Watson for years as a way to help human doctors sort through masses of data and make more informed decisions. However, there are a few important advances that have made computers like Watson much more effective than they used to be. The first is cloud computing: rather than storing all their data on their own servers, hospitals can now share enormous amounts of information with each other, which dramatically increases their collective ability to analyze huge amounts of records; similarly, they can send large numbers of highly specific questions back across these databases in search of answers and diagnoses.
4) Carry out surgery
Earlier in 2016, Google’s DeepMind AI proved it could play Go—and beat a human grandmaster at it. But now it’s set its sights on a more ambitious goal: operating an actual hospital. The team has been working with UCLH NHS Foundation Trust in London to train an AI that can diagnose eye diseases using retina scans and predict patient outcomes based on past records. Doctors are still monitoring its diagnosis accuracy, but they say that their expectations have been exceeded so far. This is just one example of how AI is becoming increasingly efficient as it learns from data sets; algorithms will likely be taking over menial tasks at work or school in no time. If you need another reason to fear technology overtaking your job, consider Tesla CEO Elon Musk’s recent warning that automation software is the single biggest risk we face as a civilization. Fear not (too much), however—AI developers tend to disagree with Musk’s assertion and believe technologies like automation will create jobs for humans.
5) Pass the Turing test
The Turing test is one of several methods that can be used to determine whether or not a computer has achieved human-level intelligence. It’s named after Alan Turing, who proposed it in 1950. The test involves a conversation between two individuals (in our case, a human and a computer). One of these individuals is isolated from both parties, and during their interaction with each other, they must attempt to identify which of them is actually human. In order for AI to pass as human, we have to struggle to tell them apart from actual humans. There are five signs that you may soon encounter true AI: machines that can think creatively; machines capable of learning abstract knowledge; machines able to speak autonomously; machines smart enough to solve problems on their own; and computers capable of solving major world issues like disease and famine.