Tencent – Brute force.

Reply to this post

Tencent has time as most gaming AIs are relatively simple.

• Tencent is finally jumping into artificial intelligence (AI) and I think that it is fortunate in that it is not very difficult to create good algorithms for the vast majority of the games that Tencent offers.
• Tencent has created AI Lab which now has more than 250 employees whose task it is be to create algorithms that create more sophisticated game opponents as well as chat bots for companies that use WeChat and QQ to talk to their customers.
• The general impression of AI seems to be that as soon as one has created an AI group, superb, hyper-intelligent algorithms will come rolling out of the door but in reality, this is very far from the truth.
• RFM research has found (see here) that single biggest determinate of AI excellence to date is time and those that have been doing it the longest tend to have the best AI.
• This is why RFM has found that it is the search engines thatare the most advanced even though some of the biggest brains in the field are employed elsewhere.
• With Tencent just getting into this field, I think it will be a very long time before it will be in a position to roll out algorithms that are capable of making its services meaningfully more intelligent.
• In the long run this will be crucial to maintaining its dominance in the Digital Life segments where it is present as I think competition will become much tougher as the market matures.
• The good news is that it is unlikely to prove very difficult to create algorithms that are more than good enough to play the games that it offers to a very high standard.
• This is because most games are either based on hand eye co-ordination or can be solved by an algorithm using a brute-force approach.
• Brute force involves evaluating every possible outcome from a given position and choosing the best one.
• With today’s improvements in memory and processing power, this is not very difficult to achieve.
• The highest profile exception is Go which has so many possible combinations that brute force becomes impossible.
• This is why DeepMind’s AlphaGo was such a breakthrough, as it uses AI to work out which options should be searched much like a human would.
• Tencent has produced an AI Go player called Jueyi which has been able to play to a very high standard but I think that the design has been copied from AlphaGo.
• AI is a co-operative field and DeepMind has published most of its methodology and results for the creation of AlphaGo in the scientific magazine Nature.
• Consequently, I do not view this as a good example of Tencent coming up with an innovation of its own and I think we will not be seeing hyper-intelligent AIs appearing in Tencent’s services anytime soon.
• However, I think that Tencent has time as its core markets are still seeing steady growth and it should be reasonably easy to improve the AI opponents in its core segment: gaming.
• I still like Tencent as there remains substantial upside should it really begin to monetise the ecosystem that it has created but it has a very long way to go before it can be considered a force in AI.