Product managers are being asked to deploy AI features they don’t fully understand. Here’s the security and ethics checklist nobody gave you, with real examples of when to say yes and when to say no.
AI PMs are struggling to strike a balance between senior management expectations and data-scientists' obsession with MMLU scores.
What they really need is to know where to use AI (and where not to), align stakeholders, and build products people actually need. These mental frameworks and checklists will be a super addition to their toolkit.
This framework is essential. Embedding ethical, security, and data considerations into every AI feature is the difference between costly failures and responsible innovation.
I see a lot of "AI" slapped on features that aren't even AI. Same thing that happened with "crypto" back in the days... Labels always catch the hype, but you always need the substance.
This is excellent! I was pulled in from the start.
I was just talking with someone on Zoom this morning saying that building the AI part is no problem, anyone can use Claude or Cursor to build the feature. What the client is more concerned about (particularly in healthcare) is how the user data is being stored and processed.
And I think in that mad race of "adding AI to the product" most people ignore the ethics, safety, and compliance part.
Great collaboration! This is very practical, it can be a side process during these days of integrating AI wherever it fits. I liked the reduction to checklist to quickly redirect any conversation or awkward email request… thanks for sharing!
Ideally, AI would be treated like any other feature. But product teams, executives, and CEOs see it as a strange monolith that they don’t understand but want to benefit from.
This is a checklist that should be in every product team in the next years for anything!
This is great insight, Ilia! My latest articles contain a lot of data, and I was wondering if it would be overwhelming for the audience. But looks like it’s resonating well with everyone.
This is really well put together, great work to both of you! I like that you specified what should and shouldn't be done and where things usually fall apart.
Love this article. What depth in value!
AI PMs are struggling to strike a balance between senior management expectations and data-scientists' obsession with MMLU scores.
What they really need is to know where to use AI (and where not to), align stakeholders, and build products people actually need. These mental frameworks and checklists will be a super addition to their toolkit.
Thank you, Vishal. So much of the challenge isn’t the technology itself. Frameworks like these are exactly what we need. Kudos to Elena!
This framework is essential. Embedding ethical, security, and data considerations into every AI feature is the difference between costly failures and responsible innovation.
I’m with you, Suhrab! 💯
I agree! Elena created a very useful framework that should guide everyone thinking about integrating AI tools.
Absolutely banger framework.
I see a lot of "AI" slapped on features that aren't even AI. Same thing that happened with "crypto" back in the days... Labels always catch the hype, but you always need the substance.
Oh, the crypto wave! Thanks for the good memories right there. 😂
Thank you, Mia! ❤️
This is excellent! I was pulled in from the start.
I was just talking with someone on Zoom this morning saying that building the AI part is no problem, anyone can use Claude or Cursor to build the feature. What the client is more concerned about (particularly in healthcare) is how the user data is being stored and processed.
And I think in that mad race of "adding AI to the product" most people ignore the ethics, safety, and compliance part.
Thank you, Juan, I agree. It seems like ethics and safety are afterthoughts when in reality should be at the forefront.
Yeah exactly that.
Security and accessibility have been long an afterthought in software dev. Looks like ethics and safety is having the same luck with AI....
And we haven't learned anything from that?
Well… unfortunately…
Specially in healthcare! Couldn’t agree more with you, Juan! 👏
Great collaboration! This is very practical, it can be a side process during these days of integrating AI wherever it fits. I liked the reduction to checklist to quickly redirect any conversation or awkward email request… thanks for sharing!
Thank you, Antonio, for reading and commenting. An actionable checklist can make a huge difference in making responsible decisions.
Thanks, Antonio!
Ideally, AI would be treated like any other feature. But product teams, executives, and CEOs see it as a strange monolith that they don’t understand but want to benefit from.
This is a checklist that should be in every product team in the next years for anything!
Extremely good article. I’m particularly interested in the data world, and you guys nailed it with the question that are in that part!
Thank you, Ilia, for taking the time. Elena did all the great work on data!
This is great insight, Ilia! My latest articles contain a lot of data, and I was wondering if it would be overwhelming for the audience. But looks like it’s resonating well with everyone.
Thanks for the input! 🙏🏻
This is really well put together, great work to both of you! I like that you specified what should and shouldn't be done and where things usually fall apart.
In the days to come, this will be so important!
Thank you, Adrian, I appreciate you. Clarity on what to do (and what not to do) is everything.