Summary: Why does Generative AI have an Access Problem? For AI to be ready for the masses, it must pivot away from ‘prompt engineering’ toward point-and-click or smart auto-correcting strategies. Accessibility, inclusion, and user adoption at scale are at stake.
‘Prompt engineering’: clever but not a user adoption strategy for Generative AI
Generative AI has revolutionized how we think about productivity, problem-solving, and human-computer interaction. Its remarkable UX patterns are even taking us into the age of human-computer collaboration, where you “dance with a machine“.
Yes, but: I know many people (family, friends, and colleagues) who are not using Gen AI- many who are aware of it and have used it a few times. While powerful, this barrier to entry limits generative AI’s usability to those willing or able to master this skill. It’s akin to requiring every driver to be a mechanic – effective for some but unnecessarily complex for most. This complexity can stifle creativity and restrict the potential applications of generative AI across various industries.
Let’s face it, to get the best out of Gen AI you need to be a prompt master or be on top of your Improv for AI game. Don’t get me wrong; prompt strategy is great e.g., “Pretend you’re a UX Director and you want to improve the UX maturity of your org; what things should you do?” (Check the ChatGPT output, not a bad to-do list).
What directions can lead to more inclusive AI?
For the general population, being a prompt ninja isn’t going to cut it. To truly democratize access to generative AI, we must evolve beyond scripting AI interactions and embrace more intuitive interfaces like point-and-click filtering. Or better, perform smart auto-correct in which Gen AI interprets question construction limits combined with user nudges and offers more precise answers.
For example: “What are good Greek islands to visit?” A travel GPT’s answer: Santorini, Mykonos. (Bad answers if you know Greece; these islands are overrun by tourists– and prohibitive in the case of Mykonos, which is overrun by celebrities).
Instead: Prompt hint= What’s important to you? Filters= Relaxing spots, Less Crowds, Cultural activities, Panoramic views, Stunning Beaches. Results then tailor to your values, interests and needs.
These “filters” would take you to more precise (curated) results, which you can find on a travel-savvy blog in a simple Google search. Note: An extraordinary hat-tip to UX pioneer Ben Schneiderman, who advised this approach in his book Human-Centered AI (released a year before Gen AI emerged).
Samsung’s designers Scott Jeonggun Choi and Heri Naare are taking a similar approach to mobile AI in their Intelligence UX approach:
“It would be ideal if the output appeared immediately after pressing a button, but AI generation can take time. To make the process feel magical, we added visual and interactive effects as well as text. Designers from all fields — including user interface (UI), graphic user interface (GUI) and visual interaction (VI) — and UX writers collaborated to overcome any limitations encountered during the creation of Intelligence UX”.
Integrating point-and-click interfaces with generative AI represents a potential pivot towards inclusivity. By simplifying user interactions through graphical interfaces, filters, and sliders, we can make AI tools more accessible to a broader audience. This approach aligns with the fundamental principles of UX design – creating usable and beneficial systems to as many people as possible, regardless of their technical background.
Let’s face it: Gen AI is just a very dynamic command line interface. Command lines require you implicitly know how to “fetch” the best results with shortcode. This is not a user adoption model at scale.
Toward inclusive HCC: Human-Computer Collaboration
Microsoft recently outlined a set of AI Access Principles. Significant is this emphasis on safety, harm reduction, and inclusion. Pay attention to the language:
“We are applying a strong Responsible AI Standard to keep people at the center of AI design decisions and respect enduring values like fairness, reliability, safety, privacy, inclusiveness, transparency, and accountability.”
We’re seeing this in real-time: Google’s new image generator Gemini was released and pulled– in order to correct AI image generation that was representing historical inaccuracy. And that’s from a company that prioritizes Inclusive Design.
As the AI safety legislation rolls into your roadmap, AI will necessarily move toward a more inclusive and accessible interaction design.
A start. Several platforms have begun integrating these user-friendly interfaces, allowing users to manipulate AI-generated content through visual cues rather than textual prompts. For instance, image generation tools now often include sliders to adjust features like style and complexity or filters to refine outputs based on color schemes or moods. These innovations highlight how point-and-click interfaces can simplify complex AI functionalities, making them more approachable for the average user:
For AI interaction simplification, products like Runway ML, Lobe (by Microsoft), and Adobe Photoshop’s Neural Filters are examples. Runway ML offers user-friendly tools for creative projects, Lobe simplifies AI model creation with a visual interface, and Photoshop’s Neural Filters allow users to adjust images with AI-powered options. These platforms exemplify how visual interfaces can make AI more accessible.
The fusion of point-and-click filtering with generative AI does not just lower the barrier to entry; it opens the door to a new era of collaborative creativity. Imagine designers, educators, and creatives at all levels harnessing the power of AI without needing to master the nuances of prompt engineering. This collaboration could lead to more diverse, innovative applications of AI, as users from different backgrounds bring their unique perspectives and ideas to the table.
Call to Action
The journey to mainstream generative AI adoption is a collective one, requiring input from users, developers, and designers alike. Moving beyond exclusive prompt engineering as an interaction pattern is necessary. By advocating for more accessible interfaces and supporting platforms that prioritize inclusivity, we can all contribute to shaping a future where generative AI tools are as commonplace and easy to use as any other aspect of our digital lives.
Go deeper: On March 21st, 2024, attend Frank Spillers AI for UX Masterclass (or buy a recording)
😍 Get it FREE with UXIC membership (monthly subscription starting at $49)
