In the rush to embrace the AI revolution, I've watched countless product managers dive headfirst into using GenAI tools without fully understanding their limitations.
Let's be honest – we've all been guilty of this at some point. After all, who wants to be left behind in the AI gold rush?
But as with any shiny new technology, the true value emerges only when we understand precisely when to use it – and perhaps more importantly, when not to.
The Noise around GenAI in Product Management
If your LinkedIn feed looks anything like mine, you've been bombarded with posts about how ChatGPT and Claude are "revolutionizing" product management. Product managers are using these tools to draft PRDs, create user stories, analyze feedback, design wireframes, and even craft product strategies – often with mixed results.
Recent product management studies found that while generative AI is reshaping the role of PMs and has potential to fundamentally rewire the product development lifecycle, there are significant gaps in how the technology is being applied.
While GenAI brings massive promises to product development, it can also cause a regression in product culture and ability to think critically. My prediction is that we could loose sight of first principle thinking. Not too far from now, we will find PMs proudly showing a product roadmap generated entirely by ChatGPT. When you ask about the strategic rationale behind specific feature prioritization, they wouldn't articulate a convincing answer besides “ hey, but I had a good prompt that included…”. This is exactly the kind of trap we need to avoid.
Hence today, I want to explore 5 specific areas where product managers should be cautious about over-relying on GenAI tools, along with better alternatives and suggestions.
(Note that all citations/sources are shared at the end of the post)
1/5 Customer Research and Empathy Building
GenAI tools may excel at summarizing customer feedback or creating synthetic personas, but they fundamentally cannot replace direct customer interaction and the empathy that comes from it.
Many major retailers are utilizing GenAI to build recommendation services. While the GenAI-powered tool could make product recommendations based on customer preferences, the company discovered that relying solely on the AI for customer understanding led to generic insights. Their product team had to complement the AI with traditional ethnographic research to truly understand the emotional aspects of wine selection.
Similarly, many e-commerce marketplaces have experimented with using GenAI to analyze buyer feedback, but quickly recognized the limitations. They've since adopted a hybrid approach where AI helps categorize feedback, but human product managers still conduct direct interviews and usability studies to capture the nuanced emotional reactions that AI simply cannot detect.
What I would rather suggest is:
Reserve direct customer conversations for yourself. Use tools like Dovetail or Userbit to organize and analyze research, but always conduct the primary research yourself. Build a regular cadence of customer interviews where you can ask follow-up questions, notice body language, and develop genuine empathy. The core responsibilities of the PM will double around identifying the strategic roadmap of 'what' to build 'when', as well as identifying how to market it.
2/5 Feature Prioritization and Trade-off Decisions
While GenAI can help gather data points for prioritization, the actual decision-making process requires nuanced judgment that algorithms still struggle with.
Research shows that a staggering 49% of product managers don't know how to prioritize without valuable customer feedback. This challenge becomes even more pronounced when using GenAI tools, which often lack the context to make meaningful trade-off decisions.
Consider how many companies approaches prioritization. While they've developed a scoring framework based on Reach, Impact, Confidence, and Effort, their product teams still rely on human judgment for the final decisions. When they experimented with automating this process through AI, they found that the AI couldn't adequately account for organizational politics, strategic alignment, and the subtle interdependencies between features.
Alternatively, use structured frameworks like RICE, weighted scoring, or value vs. effort matrices as a starting point, but recognize that these are tools to inform your judgment, not replace it. The most successful product teams emphasize that without clean and structured data from diverse sources, generative AI cannot be successfully integrated. Build a collaborative prioritization process that involves key stakeholders and incorporates both quantitative data and qualitative insights that GenAI simply cannot capture.