Credit Suisse Labs CEO on Using AI/ML to Build Smarter Products (Part 2)

There are plenty of tasks out there for AI/ML to manage. For example, these solutions can enhance everything from cognitive and assistive tasks – to aggregation and automation tasks. Generally, machine learning would be perfect for these areas. But, we need to have a debate on how good it’s going to get.  

The real question is – where’s the part that brings fulfillment to people? How do you build computational tissue around that part so that they are focusing on the part that actually matters? Do they need something calling anomalies out for them? Or, would dealing with something that registers hundreds of false positives be something that drives them crazier? 

Products do not just need to be smart, they must get smarter. However, the nature of the product also can change. For example, by taking a look at what was considered boring or repetitive and then asking what could a smart computer do instead fundamentally changes the activity.

Designing A Product with AI/ML

It is always important to design the product with the option of being smarter later. However, there are two questions that need to be addressed. First, is the product actually making the lives of its users better? The second question is what are the secondary and tertiary effects of having implemented that machine learning for many people? 

Simply put, AI/ML has fundamentally changed the product life cycle. From the stakeholder management perspective, there is a huge value in showing what can be done. So there is a greater emphasis on showing rather than telling. People don’t really understand what these machines can do and conversely have unrealistic expectations as to what they can do. So, you basically need to show people what is possible. This is compounded by the fact that you actually don’t know how it’s going to turn out until you actually build it.

The process perspective says that you need to make something smart and then measure it to see how it changes people’s behavior. Then, you must check to see core activities are made superior thanks to the technology being used. Ultimately, you must focus on this behavior change through time, then lather, rinse and repeat.

Finally, when it comes to creating a proof of concept – you need to follow a couple of steps.

First, you must isolate the thing that should be smart. Then, educate stakeholders on how to measure how smart the thing they’re using is. Next, separate the conversation about how the smartness impacts the product from the conversation about how effective the smartness is. This helps you study how good the piece of intelligence can get “in vacuo.” Ultimately, the quality of the solution determines the type of product decision the person makes. 

 

Click here for Part 1

Click here for Part 3

About the Speaker
Jacob Sisk
Credit Suisse Labs CEO
Jacob Sisk is the CEO of Credit Suisse Labs - leading a team responsible for leveraging new technologies to build new businesses or refresh old ones in a way that benefits the company at society-at-large. Prior to joining Credit Suisse Labs, he worked as a product executive at Capital One Payments as its lead data scientist. In addition, Jacob worked as a research scientist at Yahoo and Thomson Reuters - and served as Principal at Infoshock for more than a decade. Jacob holds an MBA from UCLA and currently lives in the San Francisco Bay Area.

Recent Posts