The state has already used artificial intelligence as a strategic reminder. In this context, the gap between AI product managers is huge, and talents who know artificial intelligence and understand products in the market are very scarce. Many students have also begun to consider the transformation of AI product managers, what capabilities are needed to successfully transform? I hope that students will have a general understanding of AI through this article.
Introduced today in the following four directions
- AI preliminary understanding
- AI Product Manager Classification
- AI PM's capability model
- AI application field
- How to transform from 0?
01. Initial understanding of AI
AI market situation
Large-scale technology companies quickly turned to the field of artificial intelligence: Google, Microsoft, facebook, Amazon, Baidu, Keda Xunfei, Tencent, Alibaba, etc., all entered the field of artificial intelligence, the previous Internet cloud habitat conference has now basically become artificial intelligence conference ;
Employment Salary for Artificial Intelligence Product Managers (AI PM) in the market: Domestic AI-related products or technical salaries are basically above 30 per year, and the annual salary of AIPM by foreign first-line Internet companies such as Google Facebook has reached one million per year.
Artificial Intelligence Product Manager (AI PM) Market demand: At present, there are only two digits in the domestic AI PM with more than 1 years of experience. In the future, there will be a blowout phenomenon in the market.
According to the industrial structure, AI is currently divided into three types of companies. During the transition, students should first define their own points of interest and advantages, so as to carry out targeted learning and improvement.
01 Industry + AI
Such companiesFocus on the "industry", to provide users with products or services after AI empowerment. For example: smart home, smart car, etc. In the future, the industry will be more and more segmented and applied. Therefore, the requirements of such product managers focus on the understanding of the industry, and they need market insight and scene analysis capabilities.
This is also the opportunity for product managers.Because the current 2C product scene is too open, and 2B's work scene range is relatively limited; weak artificial intelligence can further work staff's work efficiency and reduce enterprise cost in a limited scenario.
Such companiesFocus on "AI technology"It belongs to AI service company, customers can improve their products through corresponding services or solutions, so as to quickly improve product value, such as: intelligent customer service, face recognition and so on.
Most of these companies are mainly 2B, and product managers need strong communication skills to quickly tap the real needs of customers.
There are also companies that provide basic AI technology platforms, intelligent service platforms, intelligent terminal platforms, etc., such as Baidu, Science and Technology News, to help companies reduce the cost of investment in artificial intelligence research and development. It is more important to the product manager's understanding of the underlying technical framework, if you have research and development experience, it is very advantageous.
02.AI Product Manager Classification
The essence of AI product manager work is to build wheels. Need PM to mine customer needs, to solve problems for users in a productized way, for the Internet, for example, headlines, the previous content framework has been finalized, and the product manager who went in is a tinkering job, not a step from 0-1. In the face of the highly competitive AI industry, it is even more difficult to be a product manager in the AI industry. It is necessary to have the productization capability from 0-1.
01 Narrow AI Product Manager
Product manager who completes the design, research and development, promotion, and product lifecycle management of related AI products through AI technology. In recent years, it has been basically developed in the fields of semantics, speech, computer vision and machine learning. It can be subdivided into semantic AI product manager, voice AI product manager, visual AI product manager, machine learning AI product manager, and terminal application AI product manager.
02 Generalized AI Product Manager
Indirectly involved in the 4 domain of AI technology in semantics, speech, computer vision and machine learning, or directly applied other sub-domain AI technologies (such as brain-computer interface, quantum computing, etc.) that are not mature enough to complete related AI. Product manager for product work. In the future, general AI product managers will slowly develop into narrow-minded AI product managers.
03.AI PM's capability model
The development of each industry has to undergo heavy technology, heavy products, and heavy operations. Nowadays, the AI industry has entered the heavy product stage. Therefore, the industry's requirements for AI product managers are as follows:
01 finds the business realization mode and closed loop
Lu Qi once said that the most important thing about artificial intelligence landing is to find the scene and business model, make the ultimate experience, and quickly iterate.
There are two main ways to achieve commercial realization in the AI market:
▲ One is the direct output value of AI, which replaces some manpower through AI, improves production efficiency and saves labor costs, such as intelligent customer service system;
▲ Another is that AI empowers humans and supports human decision-making, such as the application of AI in the medical field, assisting doctors in diagnosis and treatment, and AI is an assistant role to help humans.
These require the AI product manager to actually participate in the business process, and need to have sufficient understanding of the relevant industry.
For example, the current industries with better commercialization are security (smart cameras for portrait data, vehicle data, background analysis systems), finance (higher commercialization of technical applications for intelligent risk control and quantitative investment), and Internet services. (Translation, P-picture, smart recommendation, voice transfer, etc.), enterprise services (smart marketing and intelligent customer service), to B's scenario is mainly from the perspective of improving labor efficiency, reducing labor costs, helping decision-making, to C The scene is more focused on improving convenience.
02 controls product requirements
The industry generally thinksThe development of artificial intelligence is inseparable from three major elements: data, algorithms, and computing power.However, the application scenario of artificial intelligence landing is also the key to the success of a product.
The core skill of AI product managers is to redefine scenarios and requirements through artificial intelligence technology to provide a viable artificial intelligence solution.. After clarifying the specific requirements scenario, it is necessary to consider where the customer of the product will use it in the current process, and what the existing solution is, and where our product solution is better than the existing solution. AI product managers need to quickly verify that products that can solve pain problems quickly in the rapidly changing AI field are more challenging than Internet product managers.
03Mutual achievement with technology
Product design should start from the perspective of commercial profitability and the effective resolution of user pain points rather than technology.Therefore, in this sense, the AI product manager can reverse the technical optimization according to the needs of business and products.
In addition, AI product managers also needBroaden your cognitive limits and understand technical boundariesMore communication with the AI engineers in the team, always pay attention to the latest developments and changes in the AI industry, read the cutting-edge paper.
04.AI application area
AI's current main technical application areas are 3 directions, including:Computer vision, deep learning, natural language processing
After browsing through the major recruitment websites, it is found that the definitions of artificial intelligence product managers (AI PM) are different and vary greatly. Basically can be understood as two types:
Smart direction product manager
It is the PC or mobile product manager, but needs to know more about other smart competing products, or capture and mine the intelligent demand of the target customers. For example, the 叮咚 speaker is a platform that can add applications in a variety of specific scenarios, provide applications for children's teaching, and provide conversational learning. The service layer is the specific software service provided by the specific application.
▲ The most likely and most easily transitionable to the Artificial Intelligence Product Manager (AI PM) is the intelligent application-level product manager, which can be understood as a product manager who is transformed into a software service provider; only the original operating system has changed. The user interaction terminal has changed, and the input and output methods for the service layer system have also changed.
Partial algorithmic AI product manager
This type of product manager is very different from traditional PC/mobile products, and basically requires an in-depth understanding of machine learning, deep learning,CNNThe principle of the technology and the simple implementation method have higher requirements for mathematics and statistics, and there are requirements for school and education. The difficulty is relatively large.
▲PM needs to learn a simple summary of the basic theory:
- Mathematical foundation: advanced mathematics, discrete mathematics, probability theory and mathematical statistics;
- Basic science: biology, physics, sociology, etc.;
- Computer Science: Machine Learning, Deep Learning, CNN,RNN, DNN, etc.;
- Programming and framework: python, C++, Java, tensorflow, Caffe, ros programming, etc.
- Hardware: Raspberry Pi, sensors, controllers, brakes, etc.;
How does 05. transform from 0?
The white white as the basis of 0 should start with the concept and application layer of AI, and gradually deepen and find the technical boundaries that AIPM needs to understand. Therefore, for the time being, you don't have to go directly to the mathematical level of Python programming, but more of the theory, you can learn the following directions ▼
- Understand various AI technology features and input and output methods: voice processing, word processing, video / image processing, deep learning and other application technologies;
- Understand the various related terminal devices and application scenarios that AI is currently market-oriented;
- Understand market-oriented sensor components and steerable modules;
Books: "Artificial Intelligence: A Modern Approach" "The Ultimate Algorithm: How Machine Learning and Artificial Intelligence Reshape the World" "Sensor Practical Raiders", etc.
AI product manager and Internet product manager, in comparison, soft power is the same, just because the market is competitive, in the booming period, the requirements for AI product managers are relatively high, you need to know the industry knowledge, but there are a large number of people Already on the road to transformation, as long as you keep learning, you can do it.
This article is transferred from the public number to interview INTERVIEW.Original address