The most Personalised Shopping Experiences yet and a World Champion GO Player
Last week I summarised a couple exciting developments in the world of AI, focusing on AI transport and movie making. In this second part I have been looking into how AI will be impacting our shopping experiences and looking at the monumental developments that are going on in the Video Game and Board Game industry.
While we are a long way off personal robot stylists and shop attendants, AI has already been having useful impacts in retail.
The consumer of the 21st Century have become used to highly personalised shopping experiences, which has come about from intelligent algorithms programmed into our shopping apps and websites to track what we like and buy. However, AI takes this to the next level. AI can produce specifically targeted product campaigns for users and simultaneously deploy dynamic vouchers and discounts to maximise our engagements. It can track your shopping trends, correlate them with global fashion trends, work out what is popular each season, then show you the clothes it thinks you’ll like and it works. It works well.
AI in retail is making the entire process far more personalised and we aren’t far away from in-store personalisation either. The AI could easily be implemented into in-store kiosks or the apps recognising when you are in store and directing you to the section you are most likely to buy from. Further driving customer engagement and increasing revenue simultaneously.
From the very early stages of video games they have been using computers for practice and to control certain characters in games. Developers work tirelessly to develop realistic and challenging ‘AIs’ for players to fight however they have always had predictable elements too them and set paths that can be learnt easily. However, this has recently been changing.
For quite a few years now, AI systems have been playing and beating chess champions around the world but in 2016 there was a massive development.An Artificial Intelligence was developed in the US called AlphaGO to beat the worlds best GO player. GO is regarded as one of the oldest and hardest board games in the world dating back 2,500 years in which players attempt to control as much territory on the board to win. AI attempts to beat GO players have failed in prior years due to the sheer complexity of the game with at any given point there are 361 legal possible moves and the total possible number of moves in a game is 2.08×10170, which is 2 followed by 170 zeros. In other words, there are more possible moves in a game of GO than there are atoms in the universe. However, in 2016 AlphaGO achieved greatness and beat the world champion, and possibly the greatest GO player ever, Lee Sedol. AlphaGO even played moves that overturned hundreds of years of understanding of the game.
Following this, AI game playing skyrocketed. Only a few days ago another AI called OpenAI managed to beat a team of 5 Dota 2 players. 5 AI bots worked together to beat 5 people playing the game, which is a top down real time strategy fantasy game, in real time for the first time ever. While this is an impressive headline, the truth behind it sheds light on how far AI still must go in terms of games that rely on teamwork. The OpenAI was run through 180 years’ worth of games each day and by the time it was ready to beat the players it had played 3500 years’ worth of Dota 2, and the Dota players weren’t even professionals. If it took humans 3500 years to learn a game, we wouldn’t be very far. But, it just goes to show that one on one AI is incredibly impressive, but in terms of AI working together, there is still work to be done.
AI can develop in leaps and bounds and its learning patterns are clearly exponential. It learns everything possible from its mistakes and given the fact that it doesn’t tire, need rest, food or drink, or gets lazy means that many companies will start to develop AI to take over the mundane, repetitive jobs that are still integral to a company. AI is still very much in its infancy and the future for AI is exciting, however I don’t think we are going to have any naked Arnold Schwarzeneggers appearing any time soon.