Stephen Downes

This is a longish article with some good examples showing a future state (and likely applications) of what the author calls our planetary nervous system. What is meant by that is the interconnected network of sensors and indicators that respond to what's happening in the natural world, with inputs ranging from waterflows to migration patterns to the spread of wildfires. "This is machine intelligence at its most vital," writes Rimma Boshernitsan, "not replacing judgment, but extending our senses." The objective is "to integrate so coherently with the biosphere that the whole can self-regulate rather than just react." And what we want, I would say, is for this integration to be available to everyone, the way the Global Biodiversity Information Facility (GBIF) "weaves millions of records from field notes, museum collections, citizen observations and satellite traces into a living archive" creating "a global network and open-access infrastructure funded by governments worldwide." If we don't require that this data be open access, someone will attempt to privatize it.

Martin Fowler reports, "I've heard a number of reports recently about people setting up LLM agents to work on their email and other communications. The LLM has access to the user's email account, reads all the emails, decides which emails to ignore, drafts some emails for the user to approve, and replies to some emails autonomously." As enthusiastic as I am about AI, I agree with him that it's far too early to trust agentic email with real access to my email, and not only because of the security risks. I mean, I can't even trust my anti-spam services to keep out all the spam and only the spam. I'm not ready to let it make statements on my behalf. And oh yeah, the security risk.

This report (26 page PDF) is based on a survey of 468 heads of learning and talent from 394 companies in Europe, the United 
States, the UK and the Middle East. The results are not surprising. The top priority continues to be leadership and executive development, as ever, and close on its heels is artificial intelligence. Also, "In last year's data, the phrase 'skills-based organisations' came up more than twice as often as in 2024.... That persistence reflects the 'skills based' approach becoming part of the 
mainstream as we head into 2026." This priority, and also the emphasis on 'learning culture', reflects the need to adapt to a rapidly changing skills landscape; for this reason as well it is difficult to link learning programs directly to return on investment (ROI). The conditions before and after learning are often completely different, and it's often a case of 'adapt or get left behind' for individuals and companies. The report is behind a spamwall, and you can give them your contact information if you're feeling nice, but this direct link should work as well.

I contributed to this paper (9 page PDF) - not a ton, but definitely not nothing. Here's the argument that came out of our exchanges: "We analyze several emerging tensions: the ontological crisis of human authorship, which challenges traditional copyright frameworks; the risk of 'openwashing' where proprietary models appropriate the language of the open movement," and some ethical issues. "This paper argues that the binary definition of 'openness' is no longer sufficient. We conclude that ensuring equity in the AI era requires a transition from open content creation to the stewardship of 'white box' technologies and transparent digital public goods." Now there's a lot of uncharted territory in that final statement. This paper just begins to touch on it, and (in my view) concludes without really explaining what we might mean by all that.

Tom Woodward links to three interesting graphing resources in one post. This first item, a tool for selecting the sort of graphic you want to use, is a number of chart type selections classified according to the number of variables you're looking at. Their poster is probably the best value of the three. If you prefer a more open-ended selection, there's this complete guide to graphs and charts. This page also links to "on-demand courses show you how to go beyond the basics of PowerPoint and Excel to create bespoke, custom charts" costing about $100 per. And how do you make the charts? You could use SciChart, a 'high-performance' Javascript chart and graph library. But the pricing is insane, starting at $116 per developer per month. I'm pretty sure ChatGPT will teach you about the types of charts (actually, I just made one for you while writing this post) and Claude Code will be able to write you a free version of SciChart. 

I like this analogy. "Some of you may remember the Apple ads that emphasized the computer as a 'bicycle for the mind.' GenAI is not like a bicycle for the mind. Instead, it's more like an automobile." Or, says Mark Guzdial, "As Paul Kirschner recently wrote, GenAI is not cognitive offloading. It's outsourcing. We don't think about how to do the tasks that we ask GenAI to do. As the recent Anthropic study showed, you don't learn about the libraries that your code uses when GenAI is generating the code for you (press release, full ArXiv paper)." Maybe. But it depends on how you use AI - there is a 'bicycle method' (to coin a phrase) when using AI, which is what (I think) I do - making sure I understand what's happening each step of the way. As Guzdial says, "Generative AI is a marshmallow test. We will have to figure out that we need to exercise our minds, even if GenAI could do it easier, faster, and in some cases, better." See also: To Trust or to Think: Cognitive Forcing Functions Can Reduce Overreliance on AI in AI-assisted Decision-making.