Artificial Intelligence is constantly pushing the boundaries of what machines are capable of. But could machines ever become better than us? The answer is of course ‘yes’, at least in many things where our intellect used to be the unchallenged champion of creativity and intelligence. Here are 5 tough intellectual areas where AI is already performing better than humans.
Search the web quicker
RankBrain is a machine learning AI that handles the toughest web queries in Google’s search engine. It understands the meaning of words and phrases, and can therefore guess what should be to the top ranking pages in never-seen-before searches. And it is better than its biological creators. When tested, humans could guess 70% of the time, while RankBrain’s success rate was 80%. It has not replaced Google’s brute force Hummingbird Search algorithm yet, but works synergistically with it; a sign of things to come as AI is embedded into existing information systems in order to enhance their performance.
Work in deadly environments
Robots can survive where no human can, in places like deep space, the oceanic benthos, or inside a radioactive reactor. The trouble has been that they could not perform at the dexterity and intelligence level of humans. As robotics pioneer Hans Moravec has famously noted, although high-level reasoning is relatively cheap to implement when it comes to low-level sensorimotor skills AI needs enormous computational resources. In other words, human babies can do more complex things with their bodies than the most sophisticated robots. But not any more: a UC Berkley team used deep learning to teach robots fine motor skills, such as screw caps on bottles, or use the back of a hammer to remove a nail from wood. The technique simulates eye-hand coordination in humans and the research results show that robots can now match human dexterity and speed.
Translate in many languages
In the “Hitchhiker’s Guide to the Galaxy” the babelfish was a fictitious alien fish that performed instant translation once inserted into one’s ear. AI is catching up quickly with science fiction, as so-called “machine translation” is finally coming of age. TheGoogle Translate app can instantly translate text in 27 languages. And Skype is using neural network technology that mimics the human brain in order to understand human speech and instantly translate from English to Spanish. At Microsoft, who own Skype, are beta testing the method with a view to expand it in any language, and thus facilitate face-to-face communication between humans with no knowledge of each other’s language.
Get a PhD quickly
Critics of AI nauseatingly argue that machines could never be creative, or curious, or discover anything of significance – because they lack consciousness. Nevertheless, a team at Tufts have proved naysayers wrong. Intelligence does not need consciousness to discover new knowledge. By combining genetic algorithms with genetic pathway simulation the researchers created a system that was able to make the first scientific theory to be discovered by an AI: of how flatworms (or the species “planaria” to the initiated) regenerate body parts. The AI-generated theory will have a significant impact in human regenerative medicine.
Deliver a correct medical diagnosis
Ever since that day in 2011 when it beat the human champions of Jeopardy!, IBM Watson has been growing its capabilities with leaps and bounds. One of its focus areas has been oncology and the diagnosis of cancers. For human physicians the challenge of making correct diagnosis is huge. It is estimated that in order to be at top of medical knowledge human doctors must spend 160 hours per week reading new research papers. IBM Watson’s AI does that at a fraction of the time. On top of this it has the ability to search through millions of patient records, learn from previous diagnoses, and improve the reasoning links between symptoms and diagnosis. The result? IBM Watson’s accuracy rate for lung cancer is 90%, compared to a mere 50% of human physicians.
And yet, although AI has conquered many of the high castles of human intellect it is still limited because it lacks our ability for general reasoning. AI systems can do any the above 5 things better than any human, but there is not a single AI that can do all 5 things together, or more. “General intelligence” remains the Holy Grail for AI research. Once achieved, we will have arrived at the beginning of a truly intelligent mechanical mind. Nevertheless, DeepMind’s seminal paper last year in Naturedemonstrated how AI could develop general intelligence; in the example presented in the paper a deep learning algorithm was able to play many different Atari games by reasoning from first principles. So watch that space in the next two to five years, as researchers build on top of DeepMind’s DQN (“Deep Q-networks”) algorithm, and AI enters a new phase of accelerated evolution.