OLDaily

Welcome to Online Learning Daily, your best source for news and commentary about learning technology, new media, and related topics.

Feature Article
How Can Neural Nets Recreate Exact Quotes If They Don't Store Text?
Stephen Downes, Half an Hour, 2024/02/06

---

A simple intuitive explainer: How Can Neural Nets Recreate Exact Quotes If They Don't Store Text LLMs are not 'cut and paste' systems. That model does not work, not even as an analogy. Saying that an LLM 'contains a copy' of a given work is a misrepresentation of how they are actually constructed and how they work. (Corrections appreciated, if I got anything wrong)

[Link] [Local copy]

Freedom to exit
Gordon Brander, Subconscious, 2024/02/06

---

This post revisits an important issue that should concern anyone who uses technology: freedom to exit. "If you can take your data out, if you can move it from one piece of software to another, then you own your data. If you cannot, if the software can tell you no, then you don't own your data." Gordon Brander identifies three major approaches to address this: decentralized protocols, local-first data, and legal agreements. The idea of addressing this issue through contracting or licensing is absurd. There is, of course, an obvious fourth avenue that he does not raise: government legislation. Similar to the provisions under the European GDPR, it should be mandatory for any company hosting user data (including institutional data) to enable either access to all data through open protocols, or local-first data (with provision for update or deletion of aggregated data).

Web: [Direct Link] [This Post]

‘The situation has become appalling’: fake scientific papers push research credibility to crisis point
Robin McKie, The Guardian, 2024/02/06

---

This article traces the cause of the problem to 'China' but I think we can offer a more nuanced explanation in the form of misplaced incentives and academic publishers, and specifically, requiring publication in order to qualification for promotion (similarly, most institutions won't sponsor conference participation unless the person is presenting a paper, leading to similar issues in that sphere). The publishers blame 'paper mills' but the problem is that "corruption is creeping into the system" at all levels. I'm not sure whether it's possible to change the incentives, and if not, then it will be necessary for employers to pay for their own research assessment rather than offloading it onto unpaid researchers and editors working for commercial media.

Web: [Direct Link] [This Post]

A critical review of GenAI policies in higher education assessment: a call to reconsider the “originality” of students’ work
Jiahui Luo, Assessment & Evaluation in Higher Education, 2024/02/06

---

This (14 page PDF) is useful. "Drawing on Bacchi's 'What's the problem represented to be' (WPR) framework, we analysed the GenAI policies of 20 world-leading universities to explore what are considered problems in this AI-mediated assessment landscape and how these problems are represented in policies." This is similar to the approach I took here, though I did not know about WPR. Also, I sought diverse examples, not elite universities, so I could find things the elite don't consider to be problems. Anyhow, what we find here is that "the major problem represented is that students may not submit their own original work for assessment with the presence of GenAI." The table illustrated displays the full results. Via Paul Prinsloo.

Web: [Direct Link] [This Post]

We publish six to eight or so short posts every weekday linking to the best, most interesting and most important pieces of content in the field. Read more about what we cover. We also list papers and articles by Stephen Downes and his presentations from around the world.

There are many ways to read OLDaily; pick whatever works best for you:

• Read in your web browser on this web page, updated weekdays
• Receive a daily or weekly email newsletter by subscribing here
• Subscribe to the RSS feed using your favourite feed reader
• Follow @OLDaily@mastodon.social on Mastodon or elsewhere in the fediverse
• Follow @OLDaily on Twitter
• Integrate the open API using the JSON feed