Being an AI Product Manager: Everything You Need to Know

AI-driven features for products have burst into view, birthed by the humble AI product manager. While there have been many products with, or at least marketed as having, Artificial Intelligence (AI) features, those have tended to be niche and not general.

Since the launch of ChatGPT and the popularization of Large Language Models (LLM) and Machine Learning (ML), we’ve seen the rise of the general inclusion of AI-driven features in all sorts of products.

But how do you product manage AI-driven/powered features? Is it different from managing non-AI-driven features?

The short answer is: yes, it is different! And, because of this difference, we are seeing the swift rise of the sub-discipline of AI Product Management and the AI product manager.

In this article, I’ll explain just what AI product management entails, why it’s quickly becoming so important to PMs, and offer some advice on how to start specializing in AI/ML-based product management. I’ll then finish with my top tips for nailing that all-important interview.

What is AI product management?

AI Product Management is the practice of product management tailored to support the incorporation of the unique behavior of AI technologies into products in order to solve customer problems. It requires greater familiarity with the underlying concepts of various Artificial Intelligence technologies in order to make informed decisions about their application to solve customer problems.

AI product management brings some unique user experience challenges. These challenges lie in the trade-off between AI-driven features working like “magic” and the level of customer control.

“Magic” here refers to how much of the process is automated, and how many things are just done without any user input or control. While it can seem attractive to make everything very “magical”, you have to weigh that up against data protection, privacy, and user expectations.

What is an AI product manager?

The AI product manager is a PM with the appropriate understanding of the concepts of AI technologies, who can use this knowledge along with the practice of product management to apply AI to products in a way that ensures it’s not merely a gimmick, but in fact addresses a genuine customer problem.

What’s the difference between a traditional product manager and an AI product manager?

Unlike regular product managers, an AI product manager has to contend with the basic fact that the majority of AI technologies are non-deterministic. Clicking the AI button won’t always produce the same result, unlike when you use deterministic technologies. A simple example would be a calculator – you get the same answer from the same inputs every time.

The challenge for the AI product manager during the product development process is constraining the AI technology to produce bounded responses to input, and that those bounded responses remain useful and effective to customers.

The other challenge that an AI product manager faces, particularly with Generative AI technologies, is that those can often hallucinate (which is a polite way of saying that they sometimes “make sh*t up”).

AI product managers have to be able to train or constrain the AI to reduce the amount of sh*t it makes up, and to find ways for customers using these technologies to spot and feed back when the AI is making sh*t up.

This isn’t to say that the AI product managers will be developing the model or AI technologies themselves. However, they should be comfortable with being able to test and prototype prompts and other AI inputs with or without specialists, to be able to characterize how they will be used in a product.

The growth of AI product management

A quick scroll through Product Hunt’s list of the best AI apps and products of 2023 reveals that there are already hundreds of 5-star-rated products on the market right now that incorporate, or entirely focus on, providing AI-powered answers to users’ needs.

There are over 800 listings in the US alone when you search for ‘AI product manager’ on Indeed.com, another 100 in the UK, and that’s not even looking at burgeoning Tech centers like China and India.

Staying ahead of the curve is vital in product design when game-changing tech like the recent wave of AI comes along. That’s why I’ve put together this primer to help you catch up, so you can then get ahead.

Why is AI and ML so important for software product managers?

At the core, AI and Machine Learning (ML) are productivity enhancers. They enable a user to do things faster, or even to do things that they couldn’t effectively do before.

Some AI technologies allow products to process large amounts of data (log files, for example) to spot anomalies, and others are able to compare protein shapes at scale. These AI tools simplify tasks that are prohibitively expensive or impractical for humans to do.

Other AI technologies apply directly to individuals’ day-to-day work and allow them to be more productive by focusing on higher-value activities. On a side note, these are the applications that many feel will destroy jobs in a similar vein to the Industrial Revolution.

Just as product managers will apply AI to their products, they should be looking at how AI can help them with the practice of product management. The biggest win is overcoming the blank page syndrome and getting a starting point going – like generating a first draft of user stories for an idea, or filling out the idea canvas for a new idea.

Making the AI work for you

The right product with the right AI toolkit (like ProdPad, for example) will help manage the deluge of feedback, summarizing and grouping it into useful insights.

While we’re still at the relatively early stages of the AI Revolution, the productivity increases and quality-of-life upgrades these tools can provide will soon move from being delighters to being a basic expectation. Not just for your users, but also for you as an AI product manager as part of your daily workflow.

If you’re familiar with the Kano model, you’ll know how you can categorize features based on customer perception and satisfaction, with those categories going from Dissatisfied and Indifferent and Basic, to Performance and Delighter.

It’s standard for new functionality, services, and features to start life as Delighters, and then to gradually move down through the categories until they become a Basic requirement – a minimum expectation for customers.

Have a look at our Glossary page on the Kano model for another example to help illustrate this. But in the case of AI, we can expect the trajectory of AI features to travel that same path.

That’s why is so important to get your head around these changes now and to start making use of them yourself as soon as possible. You don’t want to be playing catch-up with your competitors, you need to be at the bleeding edge. You can be damn sure everyone else worth talking about is trying to be.

Moving into AI product management

As a still new and evolving sub-discipline of product management, moving into AI product management doesn’t have many hurdles. The largest is being able to understand the concepts behind the various AI technologies.

With that in mind, here are my suggestions for how to find the knowledge you need to make your move into an AI specialism.

As the majority of AI technology that is generally applicable will be Generative AI, being able to prototype and define prompts for the developers to implement is a strong must. Luckily, OpenAI has an interactive playground and good documentation on writing prompts. It’s a great way of getting started with learning how to write prompts, even if you don’t end up using OpenAI’s generative AI models in your product.

Other good learning resources are Hugging Face, FastML, and Microsoft Azure AI services. In addition, there are many (free as well as paid) courses on AI/Machine Learning such as those offered by Udemy, Coursera, and MIT.

In terms of people to follow, a few to get you started are Andrew Ng, Fei-Fei Li, and Yann LeCun. There are so many people talking, researching, and writing well about AI/ML at the moment, so use the ones above as a leaping-off point to find more expert insights.

And finally, head here for a really helpful list of even more AI/ML resources.

Interview tips for an AI product manager role

With the proliferation of AI products, you’re likely to see more and more AI product manager roles being advertised – heck, it might even become the most common product role in the market! Another reason why you’d be smart to start specializing.

There’s a lot you’ll need to get your head around if you’re going to be able to convince your interviewers that you know what you’re talking about.

So here are my top tips for impressing at any AI product manager job interview:

• Understand the underlying concepts of AI, including:
  • Neural networks
  • Large Language Models
  • Deep learning
  • Natural Language Processing
• Know the difference between embeddings, prompts, and LLMs (particularly if going for a Generative AI role).
• Have a few examples of problems AI can solve, which are the best AI techniques to solve the problem, and why.
• Understand and be able to speak to the ethical considerations in the use of AI technologies – e.g. how bias in the data set can bias a model, and result in human harm.
• Be aware of the emerging legal implications of using AI.

With all of this in mind, and by staying on top of the ongoing developments in the field, you’ll be sure to impress in your next interview, and you’ll have the knowledge necessary to guide the successful development of your products through the AI revolution.