AI and product management
Takeaways from my talk at Vanderbilt
On Wednesday this week, I had a chance to talk about how I use AI tools in my career as a product manager for Pex. (Special shout out to Vanderbilt and Aditya for giving me the opportunity!)
For those who missed it or want a comprehensive recap, here's an expanded summary of key points. You can also check out the full slides here.
What AI looks like now
AI terminology engulfs the machine learning world, blurring lines between traditional ML and newer AI techniques
Large Language Models (LLMs) dominate current usage among Product Managers
Chatbot interfaces democratize advanced ML outputs, often available for free
LLMs provide pre-trained foundations, reducing the need for massive datasets in machine learning applications
Technical plateau reached with current LLM technology, signaling need for significant advancement
What AI may soon look like
Many small AI startups face extinction or acquisition, especially those primarily offering LLM wrappers
High-value, low-cost AI tools like Perplexity and ChatGPT will survive and thrive
Big tech companies continue risky strategy of scaling up models
AI hype decreases as opportunists move to next trend
Research beyond LLMs gains momentum
Trends to watch
Bigger, more expensive LLMs
TinyML for edge devices
On-device models for privacy and reduced latency
Neurosymbolic AI combining neural networks with symbolic reasoning
Q* and similar approaches to enhance reasoning capabilities
Efforts to reduce training parameters while maintaining performance
A Product Manager's guide to AI usage
Do use it for
Rapid ideation and brainstorming sessions
Iterative requirements creation through back-and-forth refinement
Generating quick scripts for rapid prototyping
Advanced UI wireframing and initial design concepts
Writing documentation in a specific voice or style
Summarizing meetings and user interviews for quick insights
Maybe use it for (with caution)
Blog posts or long-format documents (requires heavy editing)
User story generation (needs human validation)
Initial market research (supplement with human research)
Internal communication drafts (feature updates, etc.)
Email drafting (personalize and review carefully)
Basic data analysis and visualization (verify results)
Don’t use it for
Production code (security and reliability concerns)
Final documentation or requirements without thorough human review
Direct communication with users (lacks necessary context and empathy)
Complex data analysis (potential for misleading results)
How product management is changing
Traditional asks are shifting
From: Having to ask your team and peers for every little thing — UI wireframes, design drafts, technical documentation
To: Handling most of this yourself, and only asking your team for core value proposition elements
Emerging critical focus
Deep understanding of user needs
Alignment with business requirements
Identification and prioritization of essential product elements
My presentation aimed to demystify AI's role in product management, guiding PMs to leverage AI tools effectively while avoiding potential pitfalls. As AI and the industry continues to shift, staying informed and adaptable will be key to successful product management. Make it happen!