Marc Andreessen famously stated “software is eating the world“, and very few doubted him on that. But I think there is a new king in town: Artificial Intelligence.
Every day we are hearing of amazing new applications of AI (check out Business Insider’s cool list here) that are slowly changing the world, or sometimes quite rapidly. More recently, Google has been able to turn those low-quality Street View images from Google Maps and make them beautiful, all using AI with no human interaction.
As always with new, revolutionary technologies, we have the constant fear that our jobs we’ll be replaced by that technology. It’s no different with AI. AI will kill the need for humans in some blue- and white-collar jobs, and hopefully be starting with the dull and dangerous jobs. My sincere hope is that AI will do a number of things for the human race:
- Increase our life span by taking over the aforementioned dangerous jobs such as mining, adding more safety automation into our lives such as autonomous vehicles, and help provide more accurate diagnosis of diseases earlier and even assist in determining leading indicators of potential diseases.
- Lower the overall cost of living as machines are more efficient, and so the cost of creating goods and services should decrease. AI can also help us be effective agriculture-wise to help increase harvest sizes and decrease waste.
- Allow humans to spend more time not working in two ways: take over the mundane aspects of our jobs (the “dull” components) and assist us in our work helping us make better decisions, produce higher-quality work, and thus allowing us to go home sooner.
Out of those three above, I think #3 will directly show those worried about AI’s impact on our jobs that AI is more useful than dangerous, and it’s designed for your benefit. Eric Schmidt recently shared the same sentiment when he mentioned a McKinsey study stating that 90% of jobs are not fully automatable, meaning AI will be there to help, not replace.
(I couldn’t find the mentioned McKinsey study with that particular fact, but this could have been it)
So how can AI help me as a Product Manager go home sooner?
Automated Feature Testing
A recent trend I’ve noticed in my Product Management career is the lack of QA Analysts and Engineers, meaning ensuring quality has fallen mostly on PMs. This manifests itself by taking roughly 30-50% of my week testing individual features before I give it the stamp needed to go out into production.
There are three problems with this situation:
- 30-50% of a PM’s time is quite a lot; that time could be better spent on doing competitive research, user research, etc.
- The PM is a potential blocker of product: if I’m too busy to QA a feature due to meetings or whatever, then that feature is not going out.
- If I don’t thoroughly test the feature due to time constraints, and as a result and miss that one critical bug, I’ve disappointed my users.
I’m a strong believer that AI can solve the above problems by:
- Testing the feature, and the product, as an actual user based on thousands of user sessions, quickly finding bugs that users would have encountered that the PM or QA might have missed.
- Providing a degree of confidence on whether a regression could be introduced, and if it’s a high degree of confidence, then the PM or QA can step in to do manual testing.
Assisting in Design/UX
As a UXie PM, I almost always jump into UX with my designers as I feel two brains are always better than one…and to ensure the designer stays in scope.
What takes a lot of creative muscle during design is thinking about how users use the product, validating that thinking, thinking of edge cases that we should and shouldn’t support, and thinking about whether this design or that design would help us reach our business objectives and provide value sooner.
These are hard problems.
AI can help us solve these assumptions.
AI could inform us on how people use our product, as that context can help us produce a better design. It can also help us decide what edge cases to solve for as it uses that same user behavioural data to tell us how likely it is for somebody to encounter that situation.
What I’m most excited about though is that AI could help us determine at the design phase whether a particular design would be successful or not. Imagine having the AI go through the proposed user flow and actually generate the predicted results of whether a user would complete the desired action. Absolutely amazing.
With such knowledge, instead of building out a number of variations to A/B/N test, we can do that with design mocks, find the best design, and then have the dev team build the validated design and move on with life!
With these two problems solved, I feel my life as a Product Manager got immensely easier, meaning I don’t need to work 40+ hours to achieve my company’s goals, I can do it in 20 or less!
So sure, AI could one day replace me as a Product Manager, but until it does, I’m sure we can use AI more as an ally than a foe in product development. And by the time it does replace me, I’m probably be doing something that AI can’t do or design the use cases for AI to work with!