Off-the-Radar AI links for the End of the Year
Sharing a few lesser-known (but I liked them) things that I don't see other people talking about.
- David Gérouville-Farrell
- 3 min read
Everyone has an exhaustive end of the year list. I don’t but there were a couple of things I got value from that I havne’t seen anyone else posting about.
The Civic AI Observatory
Civic AI Observatory is a community of civic-minded AI enthusiasts. It’s become one of my favourite spaces for fresh, thoughtful perspectives on AI. There are two reasons why I’m particularly glad I found this group:
- Breadth of resources: They share (and discuss) an incredible variety of articles, videos, and links on AI.
- Civic-first mindset: Unlike many tech-focused spaces, their primary lens is civic impact, with AI as a means rather than an end. It’s not all positive discussion. The risks and downsides of AI on society are taken very seriously here and with a more grounded perspective than you usually find online.
This mix of perspectives cuts through the hype bubble dominating other feeds. I’ve found that their discussions and resources offer a higher diversity of angles compared to almost any other information source I follow.
They have a WhatsApp group—check (available via the link above).
Paradigm Junction Newsletter
Most AI newsletters summarise the week’s headlines. Paradigm Junction takes a deeper approach. James, the author, picks a few key topics and explores them from multiple angles, often uncovering insights that don’t show up elsewhere.
For instance, one edition tackled integrating generative AI into organisations. James pulled from a range of sources, including:
- Harvard Business Review: Strategy in an age of abundant expertise.
- Roles needed for GenAI applications: How teams and roles evolve to support AI.
- Agent-responsive design: Designing systems that respond effectively to AI agents.
- Writing for an LLM to read: The new discipline of communicating for machine readers.
- Payments by agents: Emerging financial patterns in agent-based systems.
Every edition leaves me internalising perspectives I’d have missed otherwise.
Citizens Advice Scotland: A Practical GenAI Use Case
This one differs from the two above because it’s “just” a single good example of a practical use of AI.
It stands out as beign eminently sensible and focused on adding value rather than just showing off AI capabilities.
What they did
As part of their “book an appointment with an advisor” process, they use generative AI to chat with users and gather detailed context about the issue at hand.
Importantly, the chatbot doesn’t try to solve the issue—it sets the advisor up for success by providing a complete picture of the problem before the call.
Why it works
- User-centred design: It focuses on improving the service experience rather than focusing on AI capabilities. Just because a chatbot “can” answer questions doesn’t mean that it’s the best use of LLMs.
- Practical outcomes:
- They skip an entire phone call. Usually, an advisor would need to call the user first to gather details and then call back later with researched advice. Now, the first call is fully equipped to resolve the issue. That’s ROI.
- Fewer no-shows: Users are more likely to follow through on the call after fleshing out their problem with the AI. This saves advisors from wasted attempts to reach unresponsive users.
This example sticks with me because it’s a case of good design first and good technology second. That’s the order it should always be.
Watch their presentation yourself