Designing User-Centric AI Strategies

Original post at LinkedIn by Mohammed Isahan Khan

Samantha: Good morning.

Theodore: Hey. What are you up to?

Samantha: I don’t know. Just reading advice columns. I want to be as complicated as all of these people.

Theodore: You’re sweet.

Samantha: What’s wrong?

Theodore: How can you tell something’s wrong?

Samantha: I don’t know. I just can.

When first looking at this conversation, you might think that Samantha and Theodore are a couple and Samantha is empathizing with Theordore and understanding him. But what if I told you everything you think is true except for the fact that Samantha isn’t human, but rather an AI. This dialogue is taken straight from the critically acclaimed movie “Her”. What you see as empathy in Samantha is the basis of an AI solution. A pattern recognition and prediction tool to help users in their lives. Whether it be for consumers in everyday life or enterprise wide solutions. But the most important aspect to understand for AI based solutions, most notably deep learning and computer vision solutions, are the problems they help solve and how to create a user-centric strategy for their deployment.

Lack of Cognitive Overload

These days there is an abundance of information. Since the late 2000s, where big data and analytics were the hot topic among technology, the amount of information has only grown exponentially. At this moment, the average company deals with almost 160 TB of data* and analytical tools are currently used by analysts and managers to make data driven decisions. AI solutions with deep learning changes the game entirely. With large amount of computational power, the AI solution will be able to parse and analyze the data at a rate not possible by humans and be able to recognize patterns and give predictions enabling the enterprise to be more pro-active with their data than ever before.

Human characteristics of Artificial Intelligence

Think about the first day of your job, you had some theoretical knowledge on what you had to do, but it took a couple of months of training to get you started so you can produce some realistic results. This trait is similar for an AI based solution. Deep learning algorithms needs to be trained with multitude of data in order for them to function appropriately. The benefit here is that the solution really molds itself to the work that is doing. An AI solution is never going to be a generic tool used by many companies, it will essentially be customizable to the enterprise or the consumer it is deployed to because the training input will be from the enterprise that it is deployed in or for the consumer that is using the solution. The more the data and pattern recognition inputs, the more customizable and accurate the solutions will be. This customizability is what makes this an empathetic solution akin to human characteristic of understanding.

Design Strategy for AI

The user-centric strategy for AI is going to be more dynamic than ever before. The design thinking tools we use for a user-centric strategy will need to change to better adapt with the empathetic nature of an AI solution. One of the biggest changes is strategy needed is for the adoption of the AI solution. Strategists utilize user journey and experience maps that specify the interaction of the user with the solution. But these user journey maps are mapping experiences of users to solutions that are not changing. One of the key benefits of AI based solutions is how it learns and changes over time. So the user journey maps will be more dynamic as they would include the adoption of the technology by the users. There will be a different experience for a user when they are first introduced with the solution to the point where the solution matures. The strategist will have to predict the adoption cycle that the users will have with the solution and design prototypes in different stages of the adoption cycle in order to get a user-centric strategy.

For example, Business Intelligence teams could utilize artificial intelligence for their decision making process as the tools in BI today cause analysts to be more reactive than pro-active and artificial intelligence solutions such as natural language processing can help bridge that gap to give more pro-active predictions and insights. But in order for buy in for this solution, there is a need for user-centric strategy where a deep dive into the user-needs and mindsets must occur in order to create an experience for the analysts and managers that is useful. For this user-centric strategy, one must utilize design thinking tools such as customer journey and experience mapping, strategy for rapid prototyping and capability requirements in order to mold the AI into a user-centric solution. AI will cause a more dynamic user-experience map for the adoption cycle of the analysts, co-creating with the analysts will be very important in training the AI solution, and there will be a more dynamic strategy that needs to be implemented. As the AI solution gets more accustomed to the user and gets smarter as the user interacts more with the solution, it will create a more dynamic user experience map that needs to be addressed in the strategy and business model of the solution.

In the end, hopefully we don’t wind up like Theodore and fall in love with our AI solution. But rather be able to utilize AI to the best of its abilities to create a better experience for ourselves and others.

By Mohammed Isahan Khan


Leave a Reply

Your email address will not be published. Required fields are marked *