Let’s embark on a journey towards a fair, inclusive, and equitable future. As entrepreneurs, creative thinkers, developers, and UX designers, we can shape the narrative of artificial intelligence (AI) to impact the world positively. Mo Gowdat from GoogleX encapsulated the challenge succinctly, stating, “In its infancy, AI is being assigned the mission of selling, spying, killing, and gambling.” Though discreetly camouflaged as advertising, recommendations, defence, and investment, these missions beckon us to contemplate a profound question: How can we transform this trajectory?
Just like fire – a tool for both warmth and destruction, and nuclear technology – a source of clean energy and devastating weapons, AI possesses the potential for both positive and negative impacts. As the torchbearers of the AI generation, we must comprehend and mitigate its potential pitfalls.
AI bias, a formidable challenge, manifests in tools like Midjourney, an AI-powered image generator. When prompted with phrases like “A lawyer presenting a case before a jury,” Midjourney reflects predominantly white male images, underscoring the inherent biases in its training data. The issue is exacerbated when, for example, prompted with “a Polish person,” Midjourney generates images that are likely to create inaccurate and harmful stereotypes that perpetuate Xenophobia.
Unfortunately, the implications of biased AI reach beyond visuals; political bias in Chat GPT and the racial inaccuracies generated by facial recognition technology underscore the far-reaching consequences. These biases can mould our worldview, potentially leading to discrimination and social division.
Our eagerness to embrace AI should be tempered with conscious consideration. This is why intO has developed the following best practices that have been designed to support informed choices.
In the dynamic landscape of artificial intelligence, adopting best practices is not a mere choice but a strategic imperative. Our mission to build an inclusive future through AI requires a robust framework. Let’s delve deeper into the key principles that pave the way for equitable AI, drawing inspiration from real-world examples and those insights shared earlier.
In the words of William Gibson, “The future is already here, but it’s not equally distributed.” Addressing this inequity is at the heart of our first principle. To make AI tools accessible to all, we advocate for adaptable pricing models. These models encompass special student pricing, location-based pricing, and free options to ensure that economic barriers do not impede access.
Moreover, inclusivity extends to individuals with disabilities. In our quest for equitable AI, we champion inclusive design. Features like voice activation, text subtitles for images, and screen readers empower users with varying abilities. By making AI tools user-friendly across diverse tech proficiency levels, we amplify their impact on a global scale.
Transparency is the cornerstone of user understanding and trust. As we’ve seen with Midjourney, biased outputs can arise from biased inputs. Therefore, AI tools must demystify their operations. Simplified explanations, open-source code for scrutiny, and accessible open data become paramount. These measures empower users to comprehend AI technology, fostering a sense of control and accountability.
By prioritising transparency, we pave the way for responsible AI development, ensuring the tools we create align with ethical standards.
AI should be a tool that reflects the rich tapestry of human experiences. To achieve this, diverse participation is non-negotiable. Our advocacy for a diverse AI workforce reflects our commitment to building inclusive solutions. We create AI models that transcend biases and work seamlessly across varied contexts by incorporating various experiences, cultural understandings, and cognitive approaches.
Furthermore, investing in User Experience (UX) research with a diversity focus becomes imperative. This approach ensures that the design process includes feedback from a representative sample of users, thereby reducing biases and creating AI systems that resonate with a broad audience.
Data control is not just a legal requirement; it’s a fundamental user right. The power to access, modify, and delete personal data should rest in the hands of users. Consent mechanisms, designed with clarity and simplicity, empower users with varying levels of tech-savviness to make informed choices.
By championing data control, we establish trust, comply with legal regulations, and enhance the quality of AI outcomes. Users who feel in control are more likely to provide accurate information, contributing to a more accurate and reliable AI ecosystem.
As AI becomes an integral part of our lives, monitoring its outputs is essential. Midjourney’s unexpected applications are a stark reminder of the need for vigilance. Continuous tracking of how AI systems are used provides valuable insights. By understanding user behaviours and the consequences of their interactions with AI tools, we can refine and improve these tools over time.
Output monitoring ensures that AI complements human intuition rather than overshadowing it. By staying curious and attentive, we safeguard against unintended consequences, ensuring that AI aligns with our ethical principles.
The journey towards equitable AI doesn’t conclude with understanding best practices; it begins with their meticulous integration. The imperative for teams integrating AI into their operations is clear: treat AI tools as potential employees. Just as you wouldn’t employ someone harbouring sexist, xenophobic, or racist attitudes, scrutinise AI tools for ethical considerations. Before onboarding, ask the pivotal question: Does the AI tool align with your values, ensuring it contributes positively to your brand and mission?
Developers working on AI initiatives within start-ups must embrace a proactive ethical approach. In environments where ethical considerations might be overlooked, consider changing jobs. The AI landscape is vast, and opportunities abound for those committed to fostering inclusivity, fairness, and equity. Developers must adapt equitable best practices as we usher in a new era of AI innovation.
Research teams exploring the frontier of AI have a pivotal role in shaping its trajectory. Researchers must start developing a specialised AI-UX research method that will integrate principles of equity. Integrating equitable principles into methodologies means we can create new opportunities by testing AI concepts from an equitable point of view.
“AI that is not fair isn’t advanced – it’s regressive.” Chat GPT, Open AI
As we stand at the intersection of technology and ethics, it is clear that the path forward is one of shared responsibility. Whether integrating AI into teams, developing it within start-ups, or researching its potential, the call to action remains unified. Equitable AI is not an abstract concept but a tangible commitment to shaping a future where innovation coexists with inclusivity. In the relentless pursuit of progress, let us forge ahead, mindful of the impact our choices today will have on the AI landscape of tomorrow.
In this one-hour intO DISCOVERY workshop, you’ll learn how to master prompts to support brainstorming and the development of future scenarios, and receive practical tips for evaluating ChatGPT output effectively and responsibly. The session will be led by intO’s Director of Research, Chloe Amos-Edkins, and Foresight Strategist, Thiara Cavadas.
Join intO CEO, Joanna Brassett, and Cultural & Foresight Strategist, Thiara Cavadas, at the AIBA Conference (October 25th, 2023, Katowice and Online). Sharing the stage with speakers from Meta, Microsoft, Miro and more, Joanna will introduce intO’s Equitable AI Framework, designed by intO to support ethical AI-empowered innovation journeys.
