Artificial intelligence (AI) and human-computer interaction (HCI) are two fields that aim to enhance the capabilities and experiences of humans using technology. However, not all AI and HCI designs are equally accessible, fair, and respectful of diverse users and contexts. In this article, you will learn some principles and practices that can help you create more inclusive AI and HCI designs.
The first step to creating inclusive AI and HCI designs is to understand who your users and stakeholders are, what their needs, preferences, and goals are, and how they interact with your technology. You can use various methods, such as interviews, surveys, personas, scenarios, and user testing, to gather and analyze user data. You should also consider the ethical, social, and cultural implications of your design choices, and how they might affect different groups of people.
I find the use if personal very useful. You have to think about your product and your potential cluents/ users. All users will be interested in your product in theory but they will want/ expect different things from your product. So understanding the profile and personas of your users will not only create inclusive environment, it will also help you with your marketing strategy.
It's important to talk with a wide range of different people. The narrow assumptions usually come after speaking to one type of person then building a solution around that conversation. This is true even if you speak to many people, but they are all essentially the from the same group.
Crafting a suitable user persona is the first step to building a user-friendly application. You can't shave a man's head in his absence. You have to know who you are building for, so well that the solution can interest him.
The second step to creating inclusive AI and HCI designs is to design for diversity and accessibility. This means that you should ensure that your technology can accommodate a wide range of user characteristics, such as age, gender, race, ethnicity, language, culture, disability, literacy, and digital skills. You should also follow the principles of universal design, which are: equitable, flexible, simple, intuitive, perceptible, tolerant, and low physical effort. You can use tools, such as accessibility checkers, guidelines, and standards, to evaluate and improve your design.
In the current scenario, most Gen AI algorithms lack transparency, posing significant risks of bias and exclusion that can cause substantial harm to humanity. It is challenging to hold technology solely accountable for these issues. So, what steps can we take now? Firstly, we should focus on educating developers and deployers about equity and inclusion. This awareness is crucial to ensure that AI systems are developed and implemented with fairness in mind. Secondly, the AI leadership team should be diverse, representing a variety of backgrounds and perspectives. Inclusion should be embedded in every step of the process, from people involved in the development to the final product itself.
Creating inclusive AI and HCI designs involves several key principles and practices. To start, it's important to design with diversity and accessibility in mind, considering a broad range of user characteristics such as age, gender, race, ethnicity, language, culture, disabilities, literacy levels, and digital skills. Following the principles of universal design, which prioritize equity, flexibility, simplicity, intuitiveness, perceptibility, tolerance, and low physical effort, can help create more inclusive designs. Utilizing accessibility checkers, guidelines, and standards can aid in evaluating and improving the design to ensure accessibility for all users.
HCI thrives in providing an opportunity for every user to be served. A technology solution should be such that accommodates the usability preference of even the disadvantaged user. Every design should always answer the question - Who would not be a able to use my solution? From that point, you know what the problem is and ensure your design addresses it to great extent.
The third step to creating inclusive AI and HCI designs is to involve users and stakeholders in the design process. This means that you should engage them in co-design, participatory design, or user-centered design activities, such as brainstorming, prototyping, feedback, and evaluation. By involving users and stakeholders, you can gain insights, ideas, and perspectives that can enrich your design, as well as build trust, rapport, and empowerment among them.
In my recent work with mind medicine, which is available in preprint, we were reviews the methodology behind building an open label research study in order to detect symptoms of anxiety using artificial intelligence in the context of psychotherapy and medication management. I think the most important part of this research, which Todd Solomon is the lead author on, is the fact that we use informed consent in order to do the work. I would argue that the informed consent process that research provides is the best and most ethical scaffold on which to build the AI future of Healthcare, at least.
The fourth step to creating inclusive AI and HCI designs is to explain and communicate your design decisions. This means that you should provide clear and transparent information about how your technology works, what data it uses, how it makes decisions, and what outcomes it produces. You should also use appropriate language, visuals, and feedback mechanisms to convey your design intentions, logic, and values. By explaining and communicating your design decisions, you can increase the understandability, accountability, and trustworthiness of your technology.
Explaining and communicating your design decisions is crucial. Explaining how diverse your data is important. Explaining the source of the data is important. Explaining how carefully equity was kept in mind and how the algorithms were decided is important. Explaining how extra effort is added to improve the algorithm and reduce bias is super important. By doing so, we are are not only creating a more inclusive algorithm which does not harm anyone but also brings trust to the AI Services and Products.
The fifth step to creating inclusive AI and HCI designs is to evaluate and iterate your design. This means that you should test your technology with real users and stakeholders in realistic settings, and measure its usability, functionality, performance, and impact. You should also collect and analyze feedback, data, and metrics, and use them to identify and address any issues, problems, or gaps in your design. By evaluating and iterating your design, you can improve its quality, effectiveness, and inclusiveness.
When getting deep into the technical aspects of AI development, it can be difficult to take the perspective of a new user. Don’t be to hard on yourself, it’s not the way our brains are designed! To combat this sit with your end users, especially the ones that think in a different way to you. Shadow them and see how they use your AI solution. You may be surprised when what you thought was obvious is not so. Discuss with your counterpart and let them inject some of their value into your solution.
The sixth and final step to creating inclusive AI and HCI designs is to learn from others and share your learnings. This means that you should seek inspiration, guidance, and support from other designers, researchers, practitioners, and experts in the field of AI and HCI. You should also share your own experiences, challenges, successes, and lessons learned with the broader community, and contribute to the advancement of knowledge and practice in the field. By learning from others and sharing your learnings, you can foster a culture of collaboration, innovation, and inclusion in AI and HCI.
Learning from others and sharing expertise are key to innovation. I've participated in numerous panel discussions on inclusive AI. These discussions allowed me to learn from fellow panellists and share my perspective. They led to collaborations, and exchanging expertise helps me implement the gained knowledge for the greater good.
Generative AI is a new computing paradigm that enables us to have more natural, conversational interactions. Rather than forcing users to adapt to technology, we should push ourselves to reimagine interfaces to meet evolving expectations for intelligent and intuitive experiences. Just as the best physical tools feel like extensions of our bodies, the best AI tools should feel like extensions of our minds.