Three ways for product managers to think about AI
As a product manager, I think about AI in three ways: AI in the market, AI as a tool, AI in products.
AI in the market
Understanding how AI is affecting the market you operate in is essential. It’s a must. As with an emerging tech, product managers need to be watching how the tech is developing, what use cases come out and how organisations use it to meet those needs.
The introduction of AI is mostly ‘technology push’ which means it’s a solution in search of a problem, but also means that it’s widely applicable to lots of problems. If we believe even a small part of the hype, we can expect AI to change every industry and so affect every person’s life in some way. It’s more of a question of when and by how much, rather than if. So as a product manager, even if you don’t work in ecommerce or healthcare or fintech, your users are being affected by AI used in those sectors, and it will change their behaviour and expectations (just like the internet and mobile devices did) for your sector and product. Understanding and getting ahead of the trends is part of the job.
AI as a tool
Using AI as a tool means using it to be more efficient in what product managers already do. I’ve heard it said that “AI won’t replace you, but someone using AI will”. Again, when, not if. Can you imagine working as a product manager and not using that new-fangled email for communicating? It won’t be long until we think of AI in the same way.
Product managers can (and increasingly, should) use AI to write user stories or create prototypes or summarise workshop notes. Current thinking seems to be that AI isn’t reliable enough to be used without a human-in-the-loop, so as that human, products managers have to exercise the kind of critical thinking they would is other areas of their work. Its AI as a tool, not AI as an avatar/assistant (yet, but it’s coming), and a bad worker blames their tools.
AI in products
Using AI in products relies on product managers asking, “does it solve a worthwhile problem?” In the rush to find the problems technology push allows, adding AI into a product without that problem being worth solving could be wasteful but there are advantages to exploring ambiguous spaces and building an organisation’s capability for the future.
Product managers could be discovering worthwhile problems in two complimentary ways at the same time.
Firstly, testing hypotheses about user adoption, even if that is done quite bluntly by plugging an LLM into the product to see if anyone uses it is a completely viable approach given the state of the market mentioned above. There are other ways to understand user adoption, of course, but it’s important to understand that what doesn’t work today might work next year because of how user expectations will be quickly changed by other organisations using AI.
At the same, story-telling about the possibilities, building the case for investment, and creating longer-term technical capability for using AI as part of the product tech stack are all good things for productbmanagers to be doing. As well as, or instead of, surfacing AI to the user as part of the interface, it can include using AI to make predictions, analyse images, automate proceses with uncertain outputs, etc., etc.
It is also for product managers to balance the opportunity with the risks.
Product managers could be doing supplier due diligence, data protection and information rights assessment, figuring out an IP protection stance, analysing the total cost to run, assessing the risks, understanding ethical concerns about privacy and bias, and environmental impact of AI’s use of energy and water, and all the other things that come with introducing an emerging technology to an organisation, their product and their users.
That’s my current thinking on how product managers can think about AI. It’ll probably change, but for now I’d say product managers must be understanding AI in the market, should be using AI as a tool, and could be introducing AI to their products.