Hi, y’all!
If you noticed this newsletter was sent a bit later than usual, that is right. I had an extended weekend in Bruges and the Belgian coast to visit two art triennales (next to Bruges, the Beaufort). The quality of the work was very good, and some were very nice. Not related to anything human-AI so if you are interested, visit my Instagram for some impressions. It limited my time for collecting the notions of the news that is often part of the weekend and the Monday. Let me at least share a triggered thought and look back at last week quickly.
Last week was here in the Netherlands, a noticeable week, not in a positive sense. The new government started with a tumultuous kick-off that confirmed all worries. Stressed even more with the celebration of ending slavery Keti Koti just before, it feels like the drama of a political play you cannot write. Too bad it is for real… I am doing a political analysis here; I can only hope it will not last long and let the UK and France inspire the progressive and left movement to develop a sensible alternative.
In other news, the Civic Interaction Design research group completed the Charging the Commons research project with an insightful event presenting the research and tools for Be-commoning. Find all the details on their website. The day before, I attended the presentations of student work from Amsterdam UAS Master Digital Design and Communication and Multimedia Design and also work by research groups. Good stuff.
Triggered thought
Two posts discuss the quality of the new model, especially its tuned interaction with the model of Claude 3.5 Sonnet.
In the podcast Nathan Lambert “Switched to Claude 3.5” he shares why he loves the new model interface. He compares it with OpenAI.
Beyond the base metrics and throughput, Anthropic’s models consistently feel like the one with the strongest personality, and it happens to be a personality that I like. This type of style is likely due to focused and effective fine-tuning. The sort of thing where everyone on the team is in strong agreement with what the model should sound like.
Earlier in a post on VentureBeat, they discussed what they think is what is the most important AI feature now: Artifacts.
To understand why Artifacts matters, we need to look beyond the raw capabilities of large language models and consider the broader picture of AI integration in the workplace. The true challenge isn’t just creating smarter AI; it’s making that intelligence accessible, intuitive, and seamlessly woven into existing workflows
This could be seen as a way of unlocking the co-performance with the AI, which is more than just having a chat conversation with the model, really getting inside via a conversation beyond just being an interface.
The importance of the new interfaces and interactions with the AI for the impact in our digital life and beyond is clear and stated repeatedly. The ChatGPT-moment is not for nothing the ChatGPT-moment, the power of making the capabilities accessible and making conversations and duplex learning, both the user is learning how to prompt the AI, and at the same time, the AI is adapting and learning from new input by the human player. Reinforcement Learning from Human Feedback (RLHF).
What struck me in the considerations of Nathan, is how he evaluates the two models not on the intelligence and capabilities, but for their character, with what feel like the model is interacting. This feels like a step further in maturing these AI (or better LLMs) as part of our daily life. This nice notice by @bashford brings back memories of how the Internet was treated back in the mid-late 90s.
Read the full newsletter here, with
- Notions from last week’s news on Human-AI partnerships, Robotic performances, Immersive connectedness, and Tech societies
- Paper for the week
- Looking forward with events to visit
I started blogging ideas and observations back in 2005 via Targetisnew.com. Some years ago, I started a weekly update with links to the news and reflections. I always capture news on tech and societal impact from my perspective and interest. In the last few years, it has focused on the relationship between humans and tech, particularly AI, IoT, and robotics.
The notions from the news are distributed via the weekly newsletter, which is also archived online here. Every week, I reflect more on one topic, a triggered thought.
I share that thought here on LinkedIn and redirect it to my newsletter for an overview of news, events, and more.
target is new 