Stephen Downes

Benjamin Riley posted this article a couple months ago and referenced it in a discussion today on the nature of intelligence. The focus here is on higher order skills and the role they play in intelligence (indeed, what we describe as 'intellegence' often referes to these skills directly). The main argument here is that while large language models (LLM) may surpass humans in specific domains, they cannot transfer that learning into other domains. In fact, there is a domain in AI called transfer learning, which I don't address in my reply to Riley, but I could.

According to Creative Commons, "through engagement with a wide variety of stakeholders, we heard frustrations with the 'all or nothing' choices they seemed to face with copyright...  way of making requests about some uses, not enforceable through the licenses, but an indication of the creators' wishes." In particular, they want to be able to limit the use of their work to train AI. I commented in a meeting today that it was telling that this, of all possible preferences, is the one that surfaced as most significant. I would rank 'use of content to create weapons' or 'use of content to undermine social good' as more significant preferences. I also commented that, without access to open content, AI will be created and trained exclusively by commercial entities with licnesing agreements, which will mean there is no possibility of an open artificial intelligence.

The Bill & Melinda Gates Foundation's Open Access Policy is being updated for next year. It applies to all Gates funded manuscripts starting in January. It mandates preprints "as a preprint in a preprint server recognized by the foundation or preapproved preprint server." It requires a CC By 4.0 license on publications, though grantees retain copyright. The foundation will not pay article processing charges (APC) but doesn't rule out grantees paying the costs. And it asserts "underlying data supporting the Funded Manuscripts shall be made accessible immediately and as open as possible upon availability, subject to any applicable ethical, legal, or regulatory requirements or restrictions." The policy is broadly endorsed and (I am told) offered to governments as an example to emulate for their own open access policy. Image: Open Access Network.

Back when I was a program manager I called for a project to explore using AI to study how experts approach tasks in order to understand what skills they require. We called it automated competency development and recognition (ACDR). Here's a couple of papers from my colleagues based on the concept. Automated competency development has advanced in the years since. This product announcement takes the idea a step further, progressing from recognizing the competency and developing in humans to developing it in artificial intelligence, who then go on to replicate the expert behaviour. Sure, at this stage, maybe it's all smoke and mirrors. But there's nothing inherent in the concept that suggests to me that it would be impossible, especially in narrow domains like software engineering.

I don't think this is a great article, but the first section took me back to the days when search engines first became popular. One suggestion was for teachers to have 'Google jockeys' in their classes to look up and report on things that were discussed. Elizabeth Wells suggests in this article the idea of 'AI jockeys'. "Since these people are on their devices anyway, we might as well make them work!." The remaining five suggestions are variations on the themes of 'go live and go offline', taking us back to the days of writing tests on paper. "We used to do this all the time, so its not as onerous as you think." For people who otherwise don't write anything, requiring paper-based tests is indeed onerous.

This paper explores studies that "show that students of all ages, from elementary school to college, tend to absorb more when they're reading on paper rather than screens." I think 'absorbing' is an odd way to describe reading. Researchers have been looking at things like blood flow, brain waves and electrical activity, which they then (fancifully) interpret. Here's a link to the meta-analysis. But ultimately, "None of this work settles the debate over reading on screens versus paper. All of them ignore the promise of interactive features, such as glossaries and games, which can swing the advantage to electronic texts."