Further to the last episode on mahjong crossover mascot, I want to animate my mahjong mascot to become my avatar in videos, or in the Youtube community’s terminology, the mascot will become my V-tuber or Png-tuber. This leads me to start researching automatic lip-sync technology which involves artificial intelligence of relating lips movement and facial expressions with vocal sound patterns. While I am still picking up the knowledge on this, I would like to share some of my initial learnings on artificial Intelligence by doing this episode of Mahjong crossing over with AI.
Many people find artificial intelligence quite threatening as a lot of jobs may be replaced by machines and a lot of skill sets become obsolete in near future. Autonomous driving eliminates the need of drivers while walkout shops like Amazon Go defy the need for any salespersons. Machines diagnose patients without doctors and streets are patrolled without police. More alarming to many people, not only these more mechanical and routine works are replaced by machines, many works we considered creative can now be done with AI.
Analytic reports can be written directly from online data and decisions reacting to situations can be made quickly. Pieces of writing and music in diverse styles can be created by responding to commands. Paintings can be generated from chains of words through apps like Midjourney while caricatures can be created from portrait photos which artists or illustrators take hours or even days to do.
Game AI is another beachhead of AI since its dawn. The famous Alpha Go beat human people in the game of Go. Gradually from these simple games with two players and no-hidden information like Go, chess and shogi, to more complex games with partially hidden information like multi-player Texas hold’em poker and finally complex game like Mahjong was commanded.
Mahjong was found challenging for machines to learn not only because it is played by 4 persons and many hidden tiles in the hands of other players and in the walls of tiles to be drawn, but its number and types of tiles, its winning hands and scoring rules are pretty complex.
In 2019, for the first time, Suphx from the words Super Phoneix, a mahjong AI designed by Microsoft designed, outperforms the best human players in Mahjong. Suphx has demonstrated stronger performance and a more stable ranking than the top human player on the Tenhou platform. Suphx involved deep reinforcement learning with some newly introduced techniques including global reward prediction, oracle guiding, and run-time policy adaptation.
Avoiding these technical terms, maybe we can use Suphx to explain the concept of AI in simple languages. Machines are feds with tons of data and results, and they are then told when they have guessed right, or in tech terms, “rewarded”. Gradually the machines pick up the relationships between the two through mathematical models like neural network, a bit like training kids or animals. Through many exercises of doing something and getting results, they start to recognise what input or actions will lead to what kinds of results through experience. ||
But while we are fascinated by these opportunities, when there are insufficient pre-existing data, AI cannot perform well. Just the word mahjong should be pretty new in the Midjourney database. End up, I cannot produce anything too relevant when I try to generate paintings about mahjong. That is why people say data will be the new currency of the future. In this aspect, China has considerable advantages with its huge population. Also, the collection of information from the Chinese may be easier as the Chinese have much less concern for privacy or being under surveillance as explained in our earlier episode on Mahjong crossover Architecture in which we explain the long history of co-living in big families or neighbourhoods.
One may then ask, will mahjong become obsolete? My answer is no.
To the contrary, I believe mahjong will become more important than ever as we need to exercise our brain and in the process, we enjoy the challenge.
I always use this example to tell people that despite many old jobs have gone, new jobs will come. I asked people to imagine what it will be like when you see a farmer seeing someone running on a track mill. He will ask what are you doing, why pay others to run on the same spot. These people do not do this just to keep fit and healthy, but also to enjoy running and to find happiness and a sense of achievement. To that farmer from the old days, sweating works were not choices, but a way to make a living. But when people no longer need to do these farming works, they need to work out to keep their physical health and to pass their time. These create new jobs of making and maintaining these track mills, running business of gyms, designing and selling sportswear, etc.
Imagine one day, we do not need to do too much work with our brains, we may need mental exercises like mahjong to challenge our minds and keep our brains in good shape. We will also need that to enjoy the passing of time and get connected with others.