Issue 2 · AI Risk & Governance | Bubble, Bubble, Toil and Trouble

Days are getting longer, spooky season is here, and lurking in the shadows is all the AI use our family interacts with on a daily basis. Not so transparent is the information we end up sharing with AI. So let’s confront this scariest question this season: 

We assume they do, but this stranger danger looks and moves differently. Meta just released a Pokémon looking AI assistant. So how our family moves, especially the kids and our parents, is what lets this Zeitgeist float right through the walls. Kids shared everyone’s full names and birthdays in ChatGPT to make a family tree. Dad didn’t understand how to fill out a medical document, googled an AI assistant, and found a lovely site to help handle his Social Security number, his name, your name, etc. Because what one family member does with AI can be risky for other members, it’s critical to have the Talk. I recommend this Talk, somewhere after stranger danger and before the Birds and the Bees. ;)

We all live inside information bubbles. A bubble is just the slice of the world you actually see, shaped by your choices, your people, communities your family is a part of, and the algorithms picking what shows up in your feed. A bubble can be useful. It brings you what feels relevant. It cuts the noise. It saves you time. That’s the whole appeal and comfort, and it’s by design.

BUT our children’s education, future workplaces, and information habits are hardening into hyper-isolated bubbles. Computer science researchers at the Illinois Institute of Technology analyzed news filtering systems and discovered that algorithms consistently trigger a homogenization effect, rapidly narrowing the scope of content shown to users. Bias in the Bubble 

AND if you only ever hear what already fits inside your bubble, you start losing the stuff an informed decision actually needs: context, a challenge, a perspective that isn’t yours. We used to just worry about the social bubble. There’s that saying you are most influenced and become like the 5 people you interact with the most.

SO makes you wonder:

The algorithmic bubble, what your feed decides to serve you, is a far more powerful machine than the slow, printed world most of us grew up in. We’ve always picked where we get our information, some of it on purpose, a lot of it on autopilot. What’s new is how much gets picked for us now, by systems built to keep us scrolling.

Parallel research from UC Berkeley economists looked at how recommendation engines actually behave and confirmed: they don’t just find your bubble, they nudge your feed toward narrower, more extreme stuff over time, because that’s what holds your attention. UC Berkeley Mingduo Zhao Summary.  A bubble gets dangerous the moment it starts deciding what you never even see.

Within organizations, this lack of collaboration bridges creates massive governance failures. Data scientists might deploy a tool to solve an immediate project need, while the risk and compliance teams operate in a separate bubble with zero visibility into what is happening. The risk remains entirely invisible until a crisis hits. Furthermore, the very AI algorithms our kids interact with everyday actively accelerate this cycle by feeding them increasingly narrow content, eroding the shared reality required to solve these massive, systemic problems.

And YES, this scales all the way up. Want to see a bubble gone fully unchecked? Look at the top policy makers in the US. NPR’s On Point spent a whole week on what they called America’s “gerontocracy,” a geriatric democracy, titled Senior Class: Gerontocracy in America. It’s leaders sealed so far inside their own bubble that the rules they write stop matching the world the rest of us live in. That’s the warning for all of us. Let a bubble go completely unchecked and the bridge to everyone outside it collapses.

We cannot simply wish the bubbles away; hyper-specialization is a permanent fixture of how modern information is structured. Nobody should feel bad for having a bubble, because actually we can’t live without one. As the bubbles get bigger and more specific, the bridges have to get bigger too. Therefore, our children’s future depends entirely on the strength of the bridges we help them build. 

A bridge is simpler than it sounds. Listen to something outside your lane. Translate an unfamiliar view into words your own side can actually hear. Meet people where they are, not where you wish they were. Some serious people are building this on purpose: right here in Germany, institutions like the University of Freiburg and the Technical University of Munich are pioneering interdisciplinary AI research centers, forcing computer scientists, legal minds, and philosophers into the same room. They have realized that the risk remains entirely invisible until you force different fields to communicate. That’s a bridge, institution-sized. The one you build at home is smaller. My family keeps 2-3 contacts with completely opposite views to keep perspective. This is more powerful on the daily and

AND here’s a bridge to more context for you to consider: the Reuters Institute for the Study of Journalism at University of Oxford runs ongoing global literature reviews on echo chambers, filter bubbles, and polarization. They found echo chambers are smaller than we assume, and that the real narrowing often comes from us self-selecting, not the algorithm. It means the bubble is a choice we keep making. Echo chambers, filter bubbles, and polarisation 

Which brings us right back to our homes. The strongest bridge you can build this week is one short family conversation to build your family philosophy about AI use. Here’s 3 questions that cover a lot of ground:

This Talk is about the Fam getting the why behind the decisions we want to make around AI to reduce our AI risk. This is your very own AI governance, homegrown. A kid who understands the why makes the better call when you’re not in the room. A kid who only memorized the rule doesn’t. What we feed these systems shapes what they feed back, how we get seen, and the choices we end up making. It’s all one loop.

Give this a try: Everyone’s together, each person names one kind of information they’d keep private, and why. Grown-ups included. Every answer is a bridge.

Start with one small bridge of conversation. Then keep it going, at home, at school, out in the communities you’re part of. The box only stays black if we agree not to look inside it together.

Build one small bridge and think outside the black box. 

Think outside the black box.
Lean. Learn. Lead.
- JV

Thank you for reading. Find the big picture here. Please subscribe (it’s free) to show support or give a monetary gift by joining the Collective to help me reach my Initial Legal & Compliance Goal.

You’ve put in the work to build an audience. But when the only way to reach your fans is through a social feed determined by algorithms, the platform decides who sees what you make next.

beehiiv gives you the tools to reach your audience directly. On their all-in-one platform, you can turn followers into subscribers, grow your newsletter, and connect it all to your community or podcast. Stop chasing virality on platforms you can’t control. Build a stronger, more sustainable business on beehiiv.

Get started for free.

Sources: recommender feedback loop — UC Berkeley (ls.berkeley.edu); filter-bubble homogenization — Illinois Institute of Technology (iit.edu); interdisciplinary AI centers — Univ. Freiburg and the Munich machine-learning centers; gerontocracy framing — NPR On Point, “Senior Class.”