Human Centered AI Design

When we talk about AI, it's about making humans better at their jobs. It helps us do things more precisely. AI isn't just about automation—it's about letting humans control things a lot while also using automation.

Introduction


When we talk about AI, it's about making humans better at their jobs. It helps us do things more precisely. AI isn't just about automation—it's about letting humans control things a lot while also using automation. The focus lies on empowering, amplifying, and augmenting human capabilities to execute tasks with remarkable human control with unparalleled precision.

Why Human Centered AI?


Human-Centered AI (HCAI) moves in this direction. It creates tools that make humans better at what they do. These tools focus on giving humans a lot of control while using AI and machine learning for automation. HCAI tools are like super tools—they make human performance better by amplifying, augmenting, empowering, and enhancing it. They ensure that humans stay in control while using a lot of automation through AI and machine learning. For instance, Netflix uses AI algorithms to recommend movies and TV shows to users based on their viewing history, ratings, and preferences.  Airbnb uses AI-driven recommendation systems to suggest listings based on past bookings, user preferences, and property characteristics. This recommendation engine helps users find relevant listings, streamlining the booking process and improving the travel experience.

The main aim of an AI system is to be reliable and trustworthy. A design approach that starts with understanding users and acknowledges the limitations of both machines and people is crucial for creating trustworthy systems. By being empathetic, designers can better understand the challenges and frustrations users might face, especially in critical situations where AI systems are involved. Human-Centered AI (HCAI) follows design methods like observing users, involving stakeholders, testing usability, making improvements based on feedback, and continually assessing how people use systems with AI and machine learning.

​

Applying two key principles of user experience design—1) preventing mistakes and 2) making it easy to undo actions—works well for AI systems too. This approach ensures that users have control to activate, operate, and override actions, resulting in systems that are both trustworthy and safe.

​

Automation: Automation is the term used to describe the ability to accomplish tasks without human intervention.

Augmentation: AI augmentation means making humans better at reaching their goals. It's about using AI systems or tools to help and enhance human skills and knowledge.

​

AI Powered Solutions and Benefits

• Personalization: AI makes personalized user experiences possible by studying user behaviour, preferences, and past activities. By knowing what each user likes, AI can customize content, suggestions, and interactions for a more personal and fitting experience.
   

• Predictive Analysis: AI algorithms can foresee user behaviour and predict user needs by analysing historical data. This allows designers to anticipate user requirements and offer more intuitive and seamless experiences in advance.
   

• Faster Processes through Automation: AI-driven systems automate repetitive tasks, saving time and effort. This frees up designers and developers to concentrate on more challenging and innovative aspects of UX design, leading to greater productivity.
   

• Natural Language Processing: AI-driven chatbots and voice assistants are capable of understanding and responding to natural language, creating more conversational and engaging user interactions. This improves usability and accessibility, making technology more user-friendly for everyone.
   

• Enhanced Decision Making: AI can analyse vast amounts of data and offer valuable insights to guide decision-making processes. By examining user feedback and actions, AI algorithms can uncover patterns, trends, and user preferences. This empowers designers to make informed decisions backed by data, leading to enhanced user experiences.

​

​

DESIGN WITH AI

​

We should follow the UX golden rules for creating user experience.

• Consistency

• Usability

• Informative feedback

• Design dialogs to yield closure

• Prevent errors

• Easy reversal of actions

• Keep users in control.

• Reduce short-term memory load

​

​

Designing for AI is a collaborative effort involving product, UX, and engineering teams. Therefore, understanding the capabilities of AI and how it operates is crucial for creating impactful user experiences.

Understand AI capabilities

​

• Machine Learning: A subset of AI that teaches algorithms to predict patterns in data without explicit programming.

​

• Natural Language Processing (NLP): Enables AI systems to understand human language, creating conversational interfaces and improving user interactions.

​

• Recommendation Systems: AI-driven systems that suggest personalized content or products to users, enhancing user experiences.

​

​

Identify the Problem

Before diving in, it's crucial to determine whether the product truly needs AI and how it can benefit the user. Will it improve the user experience by enhancing performance, or will automation streamline processes for efficiency?

​

• What user needs can AI address?
   

• Which features could benefit from AI integration in the product?
   

• How can we enhance human capabilities while automating tasks? For example, AlphaSense developed an AI-powered financial search engine to assist investment firms in gaining a competitive edge. By scanning millions of data points and generating actionable reports based on relevant financial data, the system saves analysts significant time and effort.

​

UX methods to help identify user needs

Conducting interviews with your target users is a crucial step in the design process. By engaging directly with your audience, you can gain valuable insights into their goals, pain points, and expectations. These insights will not only help you understand their needs but also provide valuable guidance on how AI can assist them in achieving their objectives.

​

In addition to interviews, consider using other research methods such as user testing, and observation to gather a comprehensive understanding of your users' needs and behaviours. By combining insights from multiple sources, you can develop a better understanding of how AI can be integrated to enhance their experience.

​

Finally, involve your target users in the design process by soliciting feedback on proposed AI solutions. By collaborating with them throughout the design process, you can ensure that the final product meets their needs and expectations effectively.

​

Technical Aspects

Now, let's briefly delve into the technical aspects of AI. This understanding is crucial for designers to effectively collaborate with developers, leading to improved outcomes. How does an AI system determine the accuracy of its predictions?

​

Binary Classifier: A binary classifier is a machine learning model designed to classify data points into one of two possible categories. Its purpose is to ascertain whether a given object or entity belongs to a specific class. For instance, in medical diagnosis it might determine whether a patient has a particular disease or not.

​

Understanding binary classification principles is vital for designers and developers alike. Designers can use this knowledge to create user interfaces that convey classification outcomes, while developers can refine machine learning algorithms to improve model performance and accuracy.

​

In binary classification, there are four possible outcomes:

• True Positives: The model correctly predicts that an event will occur, such as rain.

• True Negatives: The model correctly predicts that an event will not occur, such as no rain.

• False Positives: The model incorrectly predicts that an event will occur when it actually does not, such as predicting rain when it doesn't rain.

• False Negatives: The model incorrectly predicts that an event will not occur when it actually does, such as predicting no rain when it rains.

​

From User Needs to Data Need: A Crucial Step in AI System Design

One vital aspect of designing an AI system is ensuring that the data used to train it aligns well with the tasks and goals of the users.

User Needs
The tasks, goals, and requirements of users as they interact with the AI system play a pivotal role in shaping the design and functionality of the system. Understanding these aspects allows designers and developers to tailor the AI system to better meet user expectations and deliver a more intuitive and satisfying user experience.

Data Needs
In designing AI systems, understanding the data requirements is essential for ensuring they meet user needs effectively.

Labels
In AI systems, a label is like a tag given to a piece of data. It shows what the AI is trying to figure out or predict. Labels are super important for teaching the AI, especially in supervised learning, where it uses labeled examples to learn and predict new ones.

Features
In AI systems, a feature is a piece of information about something we're studying. It's a specific detail that we use to help the AI make predictions or decisions. For example, in a weather prediction AI, features could include temperature, humidity, and wind speed.

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

Conclusion
Although AI can improve user experiences, it's crucial to find the right balance. Relying too much on AI might result in less personal interactions and a loss of human connection. Designers need to think carefully about when and how to use AI to support, rather than replace, human interaction.

​