Something from Andrew Ng
I. Introduction
AI has been booming in recent years, and nothing will make it grows slower. Although Many applications such as voice cognition and computer vision have been applied in industry maturely, the AI industry has just entered its early stage of developing. Getting more people involving the AI by knowing how to code is no longer a comparatively urgent issue. Instead, the process of how AI product should be managed by communicating with engineers is a brand new challenge that every leader in AI industry is trying to develop.
Supervised Learning is the most popular method which is broadly adopted in many industries nowadays, whereas the other methods are still in the research phase. According to Wikipedia, “the Supervised learning is the machine learning task of inferring a function from labeled training data. The training data consist of a set of training examples. In supervised learning, each example is a pair consisting of an input object (typically a vector) and a desired output value (also called the supervisory signal).” https://en.wikipedia.org/wiki/Supervised_learning
The sweet spot for AI problem is in the case of if a human can process less than 1 second, that is good for AI to solve.
II. What is the typical PM role in Silicon Valley?
This picture below well-defined in a team, particularly in Silicon Vally, that how a product is designed and launched. It is PM’s job to figure out what users’ love, whereas the engineer’s job to find figure out what is feasible. The bilateral bar on the top of the picture connects between PMs and Engineers means they must communicate to make the product meets both requirement to be great. Needless to say, they are many failed products in Silicon Valley that are only settled in the right half of the picture, which is only existed in engineers’ dream.
III. What are the current challenges for AI Product Managers?
- No well-defined wireframes.
- Such as UI/UX user interface design wireframes, there are no such things in AI product.
- Agile and Code review techniques don’t work for AI PM, yet!
- There are no such techniques like Agile development or Coding Reviews in AI product development yet, that can be efficiently used to communicate between PMs and engineers. On the other side, however, I think it is a great opportunity to become a product manager.
- One of the most efficient processes for PM to communicate with engineers, for example in voice cognition, is PM provide the datasets from customers to engineers.
IV. Current Concerns
- job displacement.
- People are worried about job displacement, but it happens all the time since ancient time. Instead of worrying about your is going to be replaced, let’s focusing on what are the facts, and which of those are positive facts. First of all, It is still the early phase of job displacement for AI and job displacement is inevitable. Secondly, AI will gradually create massive new jobs, just as semiconductor had done in a few decades ago, as well as the electric industry had done in about one hundred fifty years ago. Lastly, but the most important is, education. The most challenge for education is motivation. Despite the facts of how our jobs will be displaced, we need a new education system to re-skill ourselves to fit new jobs. The government should provide the safety to its people by paying the un-employees to study so they can contribute to the society by paying the tax as the return. The circulatory system can be then healthy operated to benefit the whole society.
- How long does the AI era will last?
- AI will last decades. Thinking about silicon industry, such as transistors and all the application of using silicons. Right now is the eternal spring of the silicon technology, whereas AI hasn’t reached this phase yet. And when it hits the point it will last even longer than silicon industry did, because AI has created so much value already and there is a clear map to transfer the industry, even only with the ideas we have.
V. Reference
"Andrew Ng: Artificial Intelligence is the New Electricity", https://www.youtube.com/watch?v=21EiKfQYZXc.Andrew Ng is one of the leading thinkers of AI industry, his research focusing on Deep learning. He taught machine learning on 100,000+ students through his online course on Coursera. He also found Google project, coherent VP and chief scientist at Baidu. He is the co-chairman and co-founder of Coursera. Last but not least, he is also the adjunct Professor at Stanford University. All the credits of this content goes to this lecture.
