Suddenly they were everywhere. The AIs. The Chatbots. The LLMs.
ChatGPT was the first Large Language Model that truly went public, unlike Siri and Alexa, ChatGPT didn’t just answer questions like it was Google Search, it had conversations and asked you questions. It felt so human it scared some people while some others fell, at least a little, in love with it.
At first, this relationship was consensual – you had to choose to use ChatGPT. Then things changed. Copilot appeared inside Windows without being invited. Meta AI turned up inside Facebook and WhatsApp. An AI joined your Zoom meeting, ready to take notes, unless you switched it off.
Chatbots — or LLMs, Large Language Models — are now the most familiar face of generative AI: systems that generate text, images, music, or video. They produce answers in a way that seems like thought, like consciousness, even though what they’re doing is something very different.
Meet Lumo and the Rest
Lumo is the in-house LLM of Proton, a privacy oriented email provider, and if a major concern is privacy then Lumo is to be recommended alongside Proton Mail.
Lumo is much less well known that other LLM’s like ChatGPT, Gemini and Claude. You can find links to and comments about these LLMs on the AI Gallery page of this site.
NAIC and Threats of AI
I hope that everyone who is on board with the Negotiating AI as a Community (NAIC) initiative has tried using one of the LLMs we’ve discussed. I recommend using ChatGPT or Claude as your main AI assistant and one other, perhaps Lumo.
It’s easy to see how an AI assistant can be really useful but as well as looking at the strengths and opportunities aspects of AI we must also consider the weakness and threats. We have to consider the whole SWOT – Strengths Weaknesses Opportunities and Threats – spectrum of AI with regard to the welfare of individuals and the community.
The following video illustrates one very alarming threat. We see LLM’s like ChatGPT as useful apps that we and our children can learn from but perhaps they are also the stranger we are allowing our children to talk to.
Workshop 2 Group Exercise 25 November 2025
In this workshop, I’d like the group to identify and agree on an issue you consider a high priority for the borough.
Once you’ve chosen the issue, use AI to explore ideas and possible strategies for addressing it.
As you work, please consider how you or your organisation might contribute to the solution. Please work in pairs or groups of three.
We will be drawing on three modes of intelligence:
Collective intelligence — what you can work out together through discussion, challenge, and collaboration..
Your own intelligence — your lived experience, judgement, and knowledge as individuals.
Augmented intelligence — insights that emerge through your dialogue with AI tools.
AI and the Actionist Workshops
Session 2: Cool Tools – Part 2
Workshop Notes
The second workshop session of the NAIC – AI and the Actionist season:
In the workshop at RDCS this session we looked at Proton Mail’s Lumo and other Large Language Models LLMs, ie chatbots, that I have linked to on the Newham Mosaic page: https://newhammosaic.org/ai-gallery/
- We asked ChatGPT to suggest ideas and an outline for Des’s upcoming Skills Sharing workshop. Des was impressed by the results and asked me to send this to him.
- We discussed the weaknesses of AI and the potential dangers including those of chatbots giving inappropriate responses to children.
- Some people were concerned about whether their email addresses could be compromised if I used ChatGPT to create a list that included their names and email addresses from the Luma site where they registered for the workshop. The simple answer is no. Using ChatGPT for this is no more likely to compromise their data than using any other common service such as Luma, Eventbrite, Google Docs, or similar platforms. If anyone would like a more detailed explanation about how the data is handled and what safeguards exist, I’d be very happy to explain further.
- In addition to answering questions, helping to plan events and helping to brainstorm ideas, ChatGPT can generate images and even code for games and applications. I asked ChatGPT to generate the code for an African and Asian traditional game called Mancala. It did this but I was only able to show the code. I later embedded that code in this webpage so that you can play it.
The next session of the AI and the Actionist Workshops is on Zoom at 6.00 pm on Tuesday 9th December. Zoom details to follow.
Attendees on 25th November:
Gavin, Kiran, Halima H, Desmond, Patricia, Shona, Paul B, Mojisola, Akik
Apologies:
Mavis