Stephen Downes

This article starts from the perspective of a personal AI that learns about its user over time. We'll call that the 'personal context' for lack of a better term, along with the author, Duncan Davidson. How do we create and manage a personal context? Davidson lists some things that matter: provenance, for example, to distinguish between what the user wrote and what they read. Scopes are important, as they distinguish between work, home and family. Identities and types matter, to allow for a consistent ontology across sessions. Time matters: my views may change over time, and what I said recently on a topic should take precedence (all else being equal). The system should be available across devices and across locales, and interoperability should not require uniformity: "Different people and companies should be able to build different context systems for different ecosystems and trust boundaries. What they need is agreement on identity, permissions, provenance, and exchange, not one universal application."

The author describes Nine Timeless Human Dignity Needs as reason, security, human rights, accountability, transparency, justice, opportunity, innovation, and inclusiveness, and opposes them 'human nature', as characterized by emotionality, amorality, and egoism, "traits that, if left unchecked, can undermine trust, cooperation, and institutional legitimacy." As he explains in this paper (14 page PDF), " the findings indicating that we do not have an innate morality, and the compelling case for our actions being ultimately motivated by our own perceived self-interest." I'm generally sceptical of arguments based on accounts of human nature because it begs the question, what is 'natural' for a human? If we fear a tiger, is this 'natural' or 'learned'? Is there a principled distinction to be drawn between 'emotion' and a 'moral sense'? Neuroscience doesn't answer these questions, because the distinction itself is blurred and confused.

I like this discussion, and it raises a central point about meaning, text and authorship. Alex Danco contrasts two authors, Roland Barthes, who in The death of the author argued "that we should no longer assign a text's meaning as coming from the author. Text's meaning comes from text, and its relation to other text," and Michel Foucault, who argued in What is an author? that "there is nonetheless an emergent quality called 'authorship', which we create and employ socially as a way to categorize the text, assign meta-significance to it, or even to regulate it or police it." So, writes Danco, "LLMs, therefore, should be a total and complete triumph for Barthes," but that feels wrong, as we still need to "classify, compress, and critique the text; whether a person wrote it or not." In my view, the text plays a small role in meaning; far more important is how the reader comprehends the text (so, yes, the same text can have different meanings for different readers). The role of authorship, in my view, intersects with how the reader comprehends the text. Knowing the author allows the reader to recognize familiar ideas and concepts from characteristic turns of phrase. AI writing doesn't offer the same affordance, and so feels hollow and less meaningful to the reader.

"The economic models and market narratives EdTech has relied on for 30 years have collapsed," writes Ian McCullough. "By sharpening definitions and drawing clearer boundaries, we can focus our resource investments and set the stage for new entrepreneurial activity." The result is a framework that divides the industry into nine segments, defined by sector (consumer, business, government/schools) across one dimension, and by service (instruction, infrastructure, managed services) across another. The divisions make sense if you're concerned about things like the speed of decision-making or the threat of new competition. But the real segmentation is defined by, first, who pays, and second, why. In my opinion, at least.

This paper (17 page PDF) is a large survey of AI users in distance education. 44,035 out of the 76,485 students enrolled in distance education courses at IU International University of Applied Sciences (IU) used Syntea, an AI supported chatbot. Beyond the size, though, the study is disappointing, breaking down the sample by easy demographics (age, gender, subject, degree, mode) rather than, say, looking at what they were doing. Does it really matter than more women than men, or more young people than old, used AI? No it does not. If you're going to publish a study, do something useful, more than just reading the logs. Via Checkpoint.

As reported here, "The University of Michigan's College of Literature, Science and the Arts is trying out something new this coming term: a dual grading system in which professors give students ordinary letter grades based on the quality of their work, but what shows up on the student's transcript is either a 'pass' or 'no credit'." The argument is that this will reduce stress for first-year students, but "according to The Hill, the university will maintain final grades internally for athletic eligibility, compliance reporting, and students' eligibility for financial aid, among other purposes, (so) there is a question of just how liberated from grading some students will feel."