Aarhus University Seal / Aarhus Universitets segl

Machine learning: solving a specific problem extremely well

Machine learning:

solving a specific problem extremely well

Artificial intelligence is not about building a new race. Hollywood robotics are a misconception of what the abilities of artificial intelligence is. Because in reality it’s all about solving problems in the best way possible, says Assistant Professor Alexandros Iosifidis. Photo: Lars Kruse.

Artificial intelligence is applied almost everywhere, and is already changing the way we live our lives. Not as the robots of science fiction movies though, but as computer technology that simply helps us solve problems.

In October 2017, Saudi Arabia became the first country ever to give citizenship to a robot. The robot, named Sophia, was built by the Hong Kong-based engineering and robotics company Hanson Robotics Limited. A month later, a Japanese artificial intelligence was the first ever virtual to be granted residency in a country.

It would seem the future of the various science fiction movies of the past has come to be. However, that is not the case. Both moves were attempts to promote the countries as places to develop artificial intelligence. And even though the visual impact of a robot giving a self-thought-out thank-you-speech at a grand event lingers, the actual level of artificial intelligence is nowhere near where you might fear.

“The public announcements and demonstrations work, but the idea of a “living” robotic citizen does not correspond to the reality of machine learning. We don’t have generic artificial intelligence that can combine multiple tasks and have superhuman power, and the abilities to destroy our world. That discussion – whether artificial intelligence will be good or bad for humanity – is due to a misconception of what the abilities of artificial intelligence is. Because artificial intelligence is not about building a new race. It’s about solving a specific problem extremely well,” says Assistant Professor Alexandros Iosifidis.

Machine learning is everywhere
Alexandros Iosifidis, who has a PhD in computer science, is doing research in machine learning, computational intelligence, computer vision and pattern recognition – all related to the topic of artificial intelligence and all in an effort to better the standards of today. Because actually – although we now have robotic citizens – artificial intelligence is still in its infancy, he says:

“We need to solve machine learning problems in a better, more efficient and more reliable way than now, and we need computers to be able to do so with very limited computational power.”

The thing is, that although we’re constantly being presented to artificial intelligence as the futuristic idea of a superhuman robotic a ‘la iRobot or Robocop, machine learning is actually going on virtually everywhere. In reality, you have your own artificial intelligence solutions in your pocket. At your desk. In your refrigerator.

Restructuring our lives
Machine learning is spreading to encompass virtually every gadget from household sensors to monitor temperature and humidity to kitchenware. That requires applications that do not need the latest and most expensive CPU on the market to be able to run.

“There are specific areas nowadays, where machine learning can outperform humans. Image processing for instance. What we try to do now is make the systems more reliable in their output, so we try to make better methodologies for training better models, such as powerful neural networks,” says Alexandros Iosifidis.

He continues:

“Today it is possible to have a genuine discussion with a machine, and you will not be able to discern whether you are actually talking to a human. Machine learning solutions are everywhere, and they are so well integrated in our lives that we might not understand what actually constitutes as machine learning. Take spam filters for instance. Those are actually machine learning solutions that work to make our day easier. In 5-10-15 years I think machine learning will have totally restructured the way humans think. The technology will be able to provide us with assistance in virtually every aspect of our lives, and even though we, as humans, are usually afraid of the future, and try to imagine the worst-case scenario to protect ourselves from what might happen, we already live side-by-side with machine learning, reaping its benefits every day.”