For years people have talked about the risks of playing computer games, but now there’s a great reason to keep on gaming. With recent advances in AI technology, video games now provide the perfect setting for artificial intelligence to train.
Since AI relies on vast amounts of data to be successful, the more you play, the more opportunities you create for computer science to break new ground in the quest to understand how the human mind works.
It may seem surprising, but gameplay importance within AI has always been central. Although it might not seem important to learn how to make a computer intelligent enough to beat a human at a game, it is an easy way to quantify the effectiveness of AI algorithms and can lead to breakthroughs beyond gaming.
The first game researchers sought to teach AI to master was chess using a brute-force method of learning whereby millions of moves and sequences were analysed before every move. When this AI method finally beat a human world chess champion in 1996 many believed it would take a century for AI to master more complex games, such as the Chinese game ‘Go’ due to the fact it has more possible moves than atoms in the universe.
This didn’t put Google off trying though, and in just 20 years they had designed ‘AlphaGo that beat the world Go champion 4-1 in a best of five contest. However, this AI was designed entirely differently and focused its energy learning the way humans play, rather than analysing every possible combination to play. This has also opened the doors to the next frontier of teaching AI to deal with real-time decisions and incomplete information.
Lessons from gaming
Lessons from gaming have helped computer scientists take AI to the new levels beyond the boundaries of gaming. A pivotal branch to develop has been the development of new ways to work with language. As AI has become increasingly intelligent and more and more data has been collected and tagged online Natural Language Processing (NLP) and even Natural Language Generating has become possible.
This has implications for almost every industry as AI reaches the stage of a neural network that is capable of learning on its own and make connections without relying on pre-defined algorithms. AI has begun to be used in the legal world, to assist doctors with diagnoses, improve cybersecurity, and even improve weather forecasting to make blaming it on the weatherman a thing of the past.
How has AI learned so quickly?
With the explosion of the internet and the advances in smart-phones, we are almost all feeding into AI learning every single day. One example is the “Captcha” prompt that often appears when filling in forms requesting you to prove your humanity. However, Google Captcha is using humans to train by helping computers learn the answers to questions it can’t answer alone and distributing the effort across the globe.
The “reCAPTCHA” programme was designed in 2007 and has already been able to digitise the entire archive of books in Googles Books records and has now moved on to digitizing imagery by using snippets from Google street view. As more and more images are tagged and fed to Googles AI, the more effective it becomes at understanding what’s what from imagery as well as words. This understanding is directly feeding into the advances in driverless cars.
How is AI being used today?
So if the written word has already been mastered and imagery is on the cards, what about the spoken word? How else are apps and the internet making use of data to change how we all experience today? Here are some of the top apps using AI in innovative ways that you are probably already using:
- Siri & Alexa Both Siri and Alexa are laying the groundwork for understanding natural language in the spoken form while also offering us immediate assistance by providing directions and organising our calendars as well as making it easier to listen to our favourite songs and turn our homes into smart ones. Machine-learning technology used by these AI’s means they get more intelligent, at least in the ‘knowing more’ sense, the more we interact with them.
- Netflix The rise of Netflix has been nothing short of meteoric, and this is in no small part due to its creative use of Artificial Intelligence. Using customer reactions to films and analysing billions of records it can suggest movies based on the reactions of previous viewers. The downside of Netflix’s AI is that it leads to clusters around bigger movies and does not adequately balance this in its methodology.
- Tesla These amazing cars are no longer just a way to get from a to b while looking great. All Tesla’s now send data directly to the cloud so if you are wondering ‘can voice assistants in cars gather human input?’ The answer is well and truly yes. However, it is also picking up on driver movements and external data to help build the data required to bring safe driving driverless cars to the roads.
- Pandora & Spotify Just like Netflix, these apps are changing the way we connect with entertainment by enabling past human experiences to guide future expectations. Pandora’s AI is particularly interesting as they first use manual analysis of songs by professional musicians to understand the DNA of songs and help make their recommendations.
- Amazon Amazon has gone from Jeff Bezos’s basement in 1994 to him being richer than Bill gates in 2019 with a net worth estimated at over $110 billion. A key component of this growth has been its use of AI in its transactions. This has helped it to learn what we are interested in based our previous purchases and online activity and they even hope to be able to start preparing to ship products to us before we’ve realised we want them soon.
Playing games has not always been something we are encouraged to spend more time on, but this is an erroneous decision. The more we play games and interact with AI in different ways, the quicker we will see the many advance AI offers, from saving lives in healthcare to driverless cars on the road, AI is bringing the future into the present day.