This is the first article in the series of articles on "How to Learn AI in the Workplace", and I will combine my workplace experience in the past ten years with my insights on using and teaching AI in the past year to provide you with a set of AI learning maps from 0 to 1 from the perspective of job and industry. The purpose of learning AI is not for product managers to use it themselves, but to help others use it well.
This article starts with the product manager.
- Everywhere is shouting for everyone to learn AI, what exactly?
- Everyone says they want to master AI, but what is the standard for mastering it?
- Where should professionals in different industries, occupations, and business needs start when learning to use and master AI?
It has been exactly 10 years since I first did a project as a product manager, and that year I took a programmer and a business push, and used WeChat service account + pocket pass (now called Youzan) to build a "Changping local version of Didi": after paying attention to the official account, send the geographical location, you can get the list of drivers who are "lying on the job" within 1 kilometer, and the master will contact you to pick you up when you place an order.
Today's AI era is very similar to the era of WeChat service accounts more than ten years ago: many product ideas that used to need to write a website and develop an app can be easily handled on the service account.
That's when I taught myself how to get started with PHP programming on the blog, and while it's still an entry-level level, it made me much more efficient when I worked with programmers over the next decade.
At that time, it seemed that after learning the PHP language, I would develop WeChat service account products.
Obviously not, because the fundamental driving force of the fission products based on the explosion of WeChat service accounts is not the PHP language, but the understanding of the WeChat ecosystem.
So, when we are in the era of AI, how should we understand AI if we want to make our products AI or develop an AI product?
What is AI today?
When we talk about AI today, it's a generative large language model, an AI tool like ChatGPT, or a product with intelligent interactions.
But for product managers who use or leverage AI, none of these claims are conducive to learning to master it.
We can learn about the Transformer Attention Neural Network mechanism, the MoE architecture, and the Embedding embedding vectorization, but these things can make you communicate with programmers more efficiently just like PHP, but they can't help you design a good product.
AI is a new technology, but for product managers, it's not the technology itself, it's the interactive innovation it brings.
Taking WeChat service account as an example, the reason why it appears to make product development easy is that "the base and interaction mode of user interaction have been designed, and you only need to pay attention to the implementation logic behind the interaction." ”
Today's AI is a stronger innovation in interaction: even if users don't have to provide much information and the product doesn't have a strong reserve of knowledge, the interactive experience can still be good.
Because AI "generates" something that fills in what information is missing between the user and the product.
So, for product managers, when you want to understand what learning AI is, you should focus on "generative AI".
Here are some of the questions I recommend you prioritize when learning AI:
- What is generative AI?
- What are the differences, strengths and weaknesses of generative AI versus previous AI?
- Why can generative AI revolutionize interactions?
- What are the advantages and disadvantages of AI based on "generative" characteristics when applied to product design?
After building these basic cognitions, it makes sense to learn how to use AI, how to write prompts, how to introduce AI into products, and even develop AI-native products.
How to make good use of AI?
Unlike other workplace roles, product managers don't just "use" AI, they need to be able to "use" AI to get started.
No product manager can be proud of "using the Internet", and only those who can design and develop products that take advantage of the characteristics of the Internet are just qualified.
The prerequisite for using AI must be to be very proficient in using AI.
Let's first understand the matter of "tuning AI", so that the follow-up LangChain, RAG, and Agents can have a solid foundation.
Here are some of the questions I recommend you must understand when learning how to use AI:
- What usage patterns can AI be categorized into, and what scenarios do they address?
- How do you "tune" AI to generate specific content based on how it works?
- How to tightly control AI-generated content: output information, format, structure......
- How do you turn your process of using and tuning AI into a reusable "product"?
- Use process canvas tools with AI to productize your own AI use cases
Advanced "utilization" of AI
When you "use" AI, it's basically based on off-the-shelf AI products like Wenxin Yiyan, Kimi, and ChatGLM, but when you're trying to "use" AI, you need to have something of your own.
Almost all large models provide API call channels, that is, you can introduce their capabilities into the interaction in your own product by calling API interfaces.
Even many large models even open-source their own model files, which you can deploy and call on your own server.
In my opinion, a product manager should at least know how to "front-end and back-end interactions", "API requests", and "basic scripting language understanding".
If you're not familiar with AI and want to "leverage" AI, we recommend that you understand the following questions:
- What are markup languages and scripting languages? Write a web tool with interactivity
- How do I call an API request to use AI outside of Wenxin Yiyan and Kimi?
- Follow the README of Github and other platforms to complete the deployment of an AI open source project
- Be able to understand the structure of the other party's open source code from the code, and know where to adjust the other party's prompt words
- Learn why a "plug-in vector knowledge base" allows AI to answer private knowledge questions
- Understand how and how the basic decision-making process for Agents to sense, plan, and act is implemented
The application of AI in specific scenarios
The purpose of learning AI is not for product managers to use it themselves, but to help others use it well.
Understanding the principles and value of AI and integrating it into product design is only one dimension, and the other dimension is that most users have a large number of stuck points in the process of using AI, and these stuck points can also be productized.
Take the user's copywriting as an example: we can provide a product that users provide a theme to help them write a good copy with one click, or we can provide a product that helps users generate a prompt word that can write a good copy.
Some users want to "sit back and relax", while others want to be in control.
A good product manager shouldn't just focus on the solution to the problem, but should spend a lot of time looking at the "diversity of things" to see what problems your target users are facing and how they perceive them.
Whether you want to learn AI or observe the "diversity of living beings", I highly recommend that you join my "AI Learning Action Circle".
This is a circle initiated and maintained by me and the community where everyone is a product manager, and in just 80 days, 2400+ AI leaders have joined.
The circle includes a knowledge planet that precipitates AI information reports, practical prompts and AI tools, an exchange community, an AIGC knowledge base and irregular live broadcasts.
The original price of the circle is 299/year, and now you can join for 49 yuan as an early bird.
Note: The circle triggers an immediate price increase (every 500 additional people), and there are still 85 spots left to reach 2,500 people.
If you want to learn all the AI-related questions mentioned in the article in one time and systematically, you can sign up for the "AI Reshaping Workplace Competitiveness Action Camp". Chapters 1 and 8 are about what you need to learn as a product manager in the AI era, and the other chapters are about the practical application of AI in various business and job scenarios.