Artificial intelligence is not clever start-ups maze of the decision

(compile: small white)

cloud network hunting note: “concept maze” is just like a real version of the map, followed the instructions of treasure map, the enterprise finally found their best decisions and trade-offs. And a good leader, is to be good at using the hands of the “treasure map”, leading his team to find rich treasure. But different from the legend, our “treasure map” is not ready-made, different areas, has a different path to success.

I’m very happy to in my familiar field, artificial intelligence (AI) startup – for example to explain the concept of the labyrinth “, it is a very interesting thing for me. This is what we call sketch of the labyrinth. Next, I’ll explain each step in detail.

accuracy as high as 80% – 80% of the MVP (most simplified may apply products). in the field of machine learning these words are common people,” machine learning are indeed part of the solution with a very clear advantage.” For most of the problems, we can more easily build a solution model, and in most cases can guarantee 80% 90% of the time. After that, however, before long, time, money, mental, and data, etc., these began to sharply reduce the revenue. According to past experience, the first thing you need to spend a few months to reach 80% accuracy, and then in the next few years, or even a lifetime to fight for the remaining 20% of the room for improvement.

this time, as we, as shown in the legend of the first decisions you will face. You have two choices: 1) try to increase the accuracy to 100%, the closer the better; Or 2) a product design even in some cases there will be a mistake also do not affect the normal use of the product. This product design, I call it “fault-tolerant user experience”.

fault-tolerant user experience. iOS autocorrect and fuzzy search in the Google search features, etc., is a great example of fault-tolerant user experience. You might say, Google search itself is a “fault tolerant user experience”