5 things to consider when designing AI products
5 mins read
×
November 24, 2024
Responsible AI is not just about liability, it's about ensuring what you are building is enabling human flourishing.- Rumman Chowdhury, CEO at Parity AI, 2023
Make it meaningful
Artificial intelligence has been an exciting concept since it gained popularity, and it’s no surprise that so many businesses want to jump on the bandwagon. Honestly, it’s not a bad move, it’s exciting and positions your product as innovative and current. This strategy can be effective as long as the AI is useful, serves a purpose, and solves a real problem. However, it’s crucial to avoid adding AI to a product that doesn’t genuinely need it. Like any other concept or feature, AI must be thoughtfully designed to address user needs and fulfill a meaningful purpose.
The first step is to evaluate your product to identify areas where AI can be helpful. This could involve solving basic needs, like speeding up a process to save users time, or addressing more complex challenges, such as incorporating a totally new feature to meet an identified user need. One way to achieve this is by conducting heuristic evaluations of the product or leveraging data from analytical tools to pinpoint stressful or lengthy processes that AI could streamline.
Users come first
Just as I mentioned, companies are jumping on the AI bandwagon, and they’re moving fast. It’s easy to get caught up in the hype or fall victim to FOMO and lose sight of the real purpose behind launching an AI product: solving problems and meeting user needs.
You need to ensure the AI features address genuine user pain points and aren’t added just because they’re trendy. The product should help users achieve their goals in a way that leaves them satisfied. The key is to follow a user-centered design process; always putting the user first, not the AI.
This might have to be done in a fast-paced environment, but that’s just how it goes sometimes. As a designer, it’s also crucial to communicate the importance of your work and findings effectively. Your role often involves advising and advocating for the best product directions. If you’re good at this, you might even buy yourself some extra time to do the work properly.
One thing I always emphasize when designing new features/MVPs is that first impressions matter because, for some users, it might be the only impression. A product or feature needs to be exciting and useful enough to keep users engaged, so they don’t abandon it for the next shiny alternative.
Give it a personality
For many users, AI is still a strange or unfamiliar concept, and people don’t typically warm up to things they don’t understand. So, make it fun and relatable. Imagine it as someone they’d easily say “hi” to at an event, not someone who sits alone in the corner of the room, all wired up and looking weird. Infuse the AI with a personality that aligns with your brand, but keep it non-intrusive and avoid gimmicks.
One way to do this is by designing empathetic responses, especially in sensitive scenarios. For example, using and advocating for supportive language during frustration or cheerfully acknowledging progress can make the AI feel more "human". You could also give it a relatable name, something users have heard before or that sounds familiar. It doesn’t have to be something super simple. It can be a name inspired by a popular movie, cartoon or game character that resonates with your audience.
In branding, it’s highly recommended to make your brand feel as human as possible to improve relatability and connect with users. This is equally true for digital products and especially for somewhat unfamiliar concepts like AI. Advocate for human-like interactions, using non-controversial, non-abusive language or slang that your target audience uses and understands, while encouraging two-way conversations.
Build trust and give users a way out
One big concern for both familiar and unfamiliar users is usually: "Is it safe?" People won’t use a product or enter their details if they feel it's unsafe. They'll opt out as quickly as possible if they have any unresolved doubts while using the product. It's crucial to keep users informed by building a transparent system that keep them updated at every step of their journey. Don’t leave them wondering for too long; design effective error-handling systems, and if they decide to opt out of a process at any step, give them a clear pathway out.
Another way to build trust is by giving users control. It helps reduce user anxiety and can lead to a better user experience. Allow users to adjust AI behavior or opt-out of specific features.
Ensure the system informs users about how their data is being handled. Given the ethical concerns around how AI systems process user data, be upfront and transparent from the start. If their data will be used to further train the AI, let them know and obtain their consent. Provide ways to anonymize or delete user data.
Make it accessible and Inclusive
The importance of considering accessibility in the design process cannot be overemphasized, as millions of users may be unable to use your product if it doesn’t cater to their needs. Ensure the product is usable by people with diverse abilities by leveraging standards like WCAG (Web Content Accessibility Guidelines). To make it inclusive, the system should also support multiple languages and cultural contexts for broader accessibility.
TL;DR
When designing AI products, focus on solving real problems and meeting user needs rather than jumping on trends. Ensure the product is meaningful, user-centered, and provides a positive first impression. Give the AI a relatable personality that aligns with your brand, using empathetic responses and familiar, approachable names to humanize interactions. Build trust through transparency by clearly communicating data handling practices, allowing user control, and obtaining consent for AI training. Finally, prioritize accessibility and inclusivity by adhering to standards like WCAG and supporting diverse languages and cultural contexts to make the product usable for a broad audience.
