Typically, I struggle to teach basic literacy about AI/ML products. For example, people come to me saying: “I want to do this task – when is this product able to do this task?” We then tell them that the product will be able to do the task real soon, we are trying to do our best. In many verticals, what is important isn’t the machine learning – but the data itself. So, conversations around the fundamental and intrinsic value of the data are often more fruitful.
Right from the beginning, the PM needs to have “nerd cred.” In other words, the product manager must understand enough to talk about the capabilities possible. In addition, they need to bring the capability-centric view to users when discussing their needs. It’s important to have product managers get involved with how one measures the performance of AI logarithms. Typically, one can measure performance in a mathematical way or in terms of its business or product objectives. The latter is more important and valuable – but also more difficult to measure.
To protect the measurement you have to do a few things. Firstly, build in your head a holistic view of how the model fits into the larger business or product system. Second, you need to understand the things that might go wrong with the data as it interacts with the model. Third, you need to talk with data scientists about the pure mathematical scoring of the model and its decisions. Fourthly, get an understanding of the product and the user consequences about those decisions.
The biggest lesson, for me, is that data matters more than the algorithm. The other big lesson is understanding how to instrument early so that the process can get smarter through time. Finally, there are a few myths I would like to address concerning Machine Learning and AI. I’d like to bust the myth that it is easy. I will not bust the myth that it will change everything. However, I will bust the myth that it will change everything for the better. We need to be smart about how and where we apply this technology.
Learn about the best use cases for utilizing AI/ML tech to improve society-at-large from Credit Suisse Labs CEO Jacob Sisk.
Credit Suisse Labs CEO Jacob Sisk defines AI/ML attributes - similarities/differences and how they make an impact on product development.
Understand opportunities and limitations for AI solutions by evaluating how input data is organized and target product audiences