Photo by Jochen van Wylick on Unsplash

We’re in an age where artificial intelligence is rapidly evolving and more powerful and accessible than ever. This shift is reshaping how we design and develop digital products. As creators, the choices we make can have far-reaching consequence on individuals, communities, and societies.

As exciting and promising as these advancements are, they also come with an increased responsibility. It is on us to ensure that our products not only harness the power of AI to solve real-world problems, but to do so in a way that is ethical, responsible, and considerate of the diverse needs of the people who use them.

I recently had a chance, at the Mindstone AI Hackathon in London, to share a few tips of advice on how to create solutions that are both impactful and conscientious, mindful of both the opportunities and limitations AI brings to a solution.

Here are a few of my essentials to consider when designing and building products that use AI.

Move beyond possible and into purposeful.

Yes, it can be a thrilling, exciting sprint to build cutting-edge solutions using AI. However, we must be careful not to get caught up in the rush and forget why we’re here. Make sure you’ve identified a clear, well-defined problem to solve. This is your focus. Get to know your problem inside out and validate that it’s a real-world problem with real demand.

Make sure you don’t end up building a cool piece of tech that’s in search of a practical application.

Do’s & Don’ts

✅ Talk to people. Research the market.
✅ Prioritize features that directly address the problem.
✅ Be open to pivot if the problem, or your understanding of it, changes or evolves.

❌ Build it just because you can, or it’s “cool”. No tech for tech’s sake.
❌ Lose sight of the core problem.
❌ Make decisions based on unvalidated assumptions.

Make usable solutions, not only functional.

While you write those lines of code to do the heavy lifting, be sure never to lose sight of the user experience. Your product has to go beyond just working. It needs to be accessible, intuitive, and, ideally, a delight to interact with. It’s essential to consider the user journey, and provide guidance and feedback to users on how to use your product.

In the end, even the most technically impressive solution will struggle to gain traction if it’s not user-friendly. User experience can make all the difference.

Do’s & Don’ts

✅ Make things accessible to people with varying abilities and backgrounds.
✅ Use accessible language.
✅ Gather feedback from people on the user experience.

❌ Assume one size fits all.
❌ Overload your product with features.
❌ Make users go through too many steps to achieve a goal.

Watch out for the bias behind the algorithm.

AI is like a superpower. But to quote the cliche, “with great power comes great responsibility”. Be aware of the biases that are inadvertently built into algorithms, and understand the broader societal implications of your product. The biases present in AI can have far-reaching, unintended effects, especially on marginalized communities, and can perpetuate inequality and unfair practices. Therefore, don’t underestimate the impact of biases on real people’s lives, and plan in to regularly audit your models and provide channels for feedback, as well as watch your own biases.

Consider the ethical implications of your algorithm, or the algorithms you use, and help build AI-based systems that make the world fairer and not the other way around.

Do’s & Don’ts

✅ Use diverse training data and question the biases in historical data.
✅ Test features and algorithms with diverse groups of people.
✅ Educate people about the limitations and risks of AI.

❌ Assume your algorithm is neutral. It’s not.
❌ Ignore feedback and criticism.
❌ Treat the algorithm as a static entity.

Beware of echo chambers or outsourcing decisions.

AI-driven products can inadvertently create an over-dependence on technology as our source for information, analysis, and decisions. When people consistently rely on AI, like ChatGPT, for information and opinions, there’s a risk that the responses of the AI become the de-facto standard, narrowing the Overton Window. This can lead to the creation of echo chambers, where AI-generated perspectives are amplified while alternative viewpoints are marginalized. You can see the issue with that, especially when combined with the fact it’s biased. Be sure to monitor your system to see if it’s inadvertently reinforcing certain perspectives.

This all becomes especially risky as people start to uncritically accept and over-rely on the decision made by AI systems, or deference bias. Outsourcing of critical thinking and decision-making to machines can lead to problems in both the short and longer term. Especially, again, when it’s possibly based on a limited, biased window. Implementing human-in-the-loop systems can get you the best of both worlds: the powers of AI with the intuition, judgment, and ethics of human decision-makers.

Build systems that empower users with diverse perspectives and ensure they have the agency, and ideally the motivation, to challenge outputs and make informed decisions. Think more trusted sidekick than puppet master.

Do’s & Don’ts

✅ Introduce randomization or exploration mechanics to present diverse viewpoints and sources.
✅ Provide options for human intervention in AI decisions.
✅ Be transparent on when and how AI acts.

❌ Outsource the wrong decisions to AI (like ChatGPT) yourself.
❌ Rely solely on automated decision-making without any oversight.
❌ Create systems that make it difficult for users to exercise agency.

Take advantage of different perspectives.

It’s imperative to acknowledge that AI development is not only about the technological aspect. Crafting AI-driven products that are holistic, innovative, and impactful requires bringing together insights from different domains, including social sciences, psychology, humanities, and more.

Breakthrough innovations emerge from interdisciplinary collaborations rather than isolated efforts.

Do’s & Don’ts

✅ Embed ethical considerations from the outset.
✅ Seek input from people from diverse demographics and cultures.
✅ Foster a culture of open communication and sharing for fresh perspectives.

❌ Think only a technical perspective matters.
❌ Only get diverse insights from AI itself (like ChatGPT).
❌ Assume your own perspective is universally applicable.

Shape tomorrow with intentional design today.

As we stand at the precipice of an AI-driven future, we must acknowledge that the choices we make today will shape the world of tomorrow. If we continue ahead without mindful use of AI, we risk entrenching biases, amplifying inequities, and building a society that’s overly reliant on algorithms that lack human empathy, intuition, and discernment. This will lead to unintended harmful consequences and a very homogeneous world that’s less creative and unique, rather than more.

On the flip side, if we proceed with intentionality, inclusivity, and a commitment to ethical values, the possibilities are boundless. We can see a future where AI augments human capabilities, helps us address complex global challenges, and plays a role in moving us toward a more equitable and prosperous society.

Make no mistake, AI will give us the tools we need to redefine healthcare, education, environmental sustainability, and countless other domains. But the journey is as fulfilling as the paths we opt to tread; we must travel with considerate steps and be mindful of the way forward to be on the side of the positive transformative potential of AI.