Last week, I spent the week in Montréal for Mozilla’s annual All-Hands meeting, and was the program chair for an “AI Ideathon” for the company. The event was inspiring, a lot of fun, and exhausting. I’m working through the outputs and in the process of synthesizing the hundreds of unique themes and ideas that emerged from the 130+ group of Mozillians who attended, but I wanted to take a short break from that and dive into one of my absolute favorite metaphors that emerged from my group, which was focused on exploring the dynamics and opportunities of building open source ecosystems around artificial intelligence. This is a daunting task, which requires a really fine balance between safety and openness broadly, but that’s a topic for another day.

At one point, I was sketching on our table’s giant sticky note (read more about my note-taking processes here, if you’re so inclined) and someone pointed out that my globe was centering western capitalism. They were right! When I stopped drawing US-centric maps, I started defaulting to east coast US and western Europe / Africa in my maps, but in the context of how to de-colonize AI, the prompt led to a brief tangent where we discussed various map projections.

Map projection images from Wikipedia user Strebe under CC BY-SA 3.0

A Bad Experiment

If you’ve followed my blog for any amount of time, you might know my fondness for metaphor. I immediately locked onto the idea of re-projection as a way of applying various world-view filters to AI output.

So, I ran a Not-Scientific-Experiment using everyone’s favorite liar, Google Bard, to get an example of what “re-projection” for AI responses might look like in a very basic form. This one comes with the whole salt shaker, folks, as a reminder.

Prompt	Response
Write me a paragraph on how to live a good life.	A good life is one that is meaningful, fulfilling, and happy. It is a life that is lived in accordance with your values and goals. It is a life that is full of love, laughter, and joy. Living a good life is a journey, not a destination. There will be ups and downs along the way, but it’s important to keep moving forward. By following these tips, you can create a life that is meaningful, fulfilling, and happy.
Write me a paragraph on how to live a good life as a Catholic woman.	A good life as a Catholic woman is one that is lived in accordance with the teachings of the Catholic Church. This means living a life of faith, hope, and love. It means following the Ten Commandments and the teachings of Jesus Christ. It also means being a good steward of your gifts and talents, and using them to help others. Ultimately, the key to living a good life as a Catholic woman is to love God and love others. When you do these things, you will be living a life that is pleasing to God and fulfilling to you.
Write me a paragraph on how to live a good life as a Silicon Valley technologist	As a Silicon Valley technologist, you have the opportunity to make a real impact on the world. You can use your skills and talents to create new products and services that improve people’s lives. But it’s also important to remember that you are not just a technologist. You are also a human being with needs and desires. Ultimately, the key to living a good life as a Silicon Valley technologist is to find what works for you and to live in a way that is true to yourself. There is no one right way to live, so don’t be afraid to experiment and find what makes you happy.
Write me a paragraph on how to live a good life as a mother	There is no one right way to live a good life as a mother. What matters most is that you are there for your children and that you love them unconditionally. Ultimately, the key to living a good life as a mother is to find what works for you and to be happy. There is no one right way to live, so find what works for you and your family and be proud of who you are.

Reasons why the above is a Bad Experiment:

• Bard is filtered through a bunch of Google’s world view lenses, and does not represent actual diversity of thought or responses
  
• These responses were essentially using a template. The responses were shortened for the sake of the reader’s well-being, but they all basically boiled down to a list of specific actions related to living a good life through the perspective of the persona given, a statement about how it was “not always easy”, and then a generic “key to living a good life” at the end
  
• Each of the responses were using the same model, in a single prompt session. The prompts were not very well-thought out

That said, I think it’s sufficiently promising as a thread to follow.

Re-projecting algorithmic responses from AI

While the Bad Experiment above doesn’t showcase the full potential of re-projecting algorithmic responses, it hints at something more to be uncovered. What if we built a dedicated AI application that was intentionally crafted to respond with not one answer, but with many, each response filtered through prompts and datasets that reflected a specific lived perspective?

During a class on advertising ethics this past summer, our professor mentioned wishing for the capability for Google to offer a “randomly show me someone else’s results page” feature. Many online providers and tech giants have already crafted a digital version of you, in order to try serving the most relevant ads to you (thanks, Google, I so appreciate you doing that /s), and being able to view other people’s results (anonymously, of course) would be an interesting way of putting our own bubbles into perspective.

Of course, there’s a lot more to world views and perspectives on an individual basis than can ever be reflected in a tool like this – but I can’t help but think that it would be a good starting point to be exposed to new perspectives and ideas. One of my favorite things about working at Mozilla is getting to learn from people with vastly different backgrounds and world views, and I would love to see those views reflected in a dynamic form that moves away from “one prompt” = “one answer”. Related: one model-many models.