Stephen Downes

Not that I have time to even begin enjoying this 20 part series, but I pass it along in the hope that readers fine it useful. Via George Siemens.

What's interesting about this project is its use of spiking neural networks (SNN) that respond rapidly to dynamic environments. See the illustration. "We developed a system that uses SNNs to control a robotic manipulator in an air-hockey game. In this setup, the automated opponent uses SNNs to process data from an event-based camera, enabling it to track the puck's movements and respond to the actions of a human player." Obviously something like this would be an excellent trainer for human air hockey players.

This is a report (60 page PDF) from the Offline Internet Consortium, an organization that advocates for the provision of digital services in areas where internet connectivity is not available. "The persistent global digital divide, driven by an array of social, economic, and political factors, underscores the critical need for innovative approaches to ensure broad and meaningful internet connectivity." Sounds great, but offline internet also creates a locus of control over the content that doesn't exist in the wider internet; the report mentions 'individual control' but really, whoever has the keys to the local server controls the local 'internet'.

This is the first of 13 lessons in a series called 'Modern Day Oracles or Bullshit Machines'. I like the way it is presented; here's a lot of scrolling, but it's also interactive, embedding interactions alongside the argument. The thesis of the first lesson is that "LLMs are predictive text generators. It is remarkable how many complex tasks can be performed in this way. But don't let the impressive capabilities of LLMs lure you into thinking that they understand human experience or are capable of logical reasoning." So, not oracles.

The law will clamp down on "the creation and dissemination of AI-generated deepfakes by changing copyright law to ensure that everybody has the right to their own body, facial features and voice... It defines a deepfake as a very realistic digital representation of a person, including their appearance and voice." The U.S. has passed a similar law. I'm wondering what the effects will be on other extractive industries, like news and media. The evening news depends on being able to display unauthorized images of politicians, criminals, people involved in accidents, and innocent bystanders. Or will fair use apply, rendering most protections offered by the new bill moot? Via Slashdot.

This is the latest bit of pseudo-professionalism associated with people working with AI: "Context Engineering is the discipline of designing and building dynamic systems that provides the right information and tools, in the right format, at the right time, to give a LLM everything it needs to accomplish a task." If it were really engineering, it would take four years of really hard work. But in the current context, it requires thinking about the topic for an evening. Via Miguel Guhlin. See also this article in LangChain.