This is pretty impressive. "Today, in the journal Nature, scientists at the University of California, San Francisco, present a new type of brain-computer implant (BCI), powered by neural networks, that might enable individuals with paralysis or stroke to communicate at the speed of natural speech—an average of 150 words per minute." Instead of 'reading thoughts' and trying to infer linguistic representations, it instead "it translates brain signals into movements of the vocal tract, including the jaw, larynx, lips, and tongue." These movements are then translated into speech. The approach requires electrodes implanted directly in or on the brain; "external electrodes simply cannot provide precise enough data from small brain regions, experts agree."
Horizon Report 2019 Higher Education Edition
The 2019 Horizon Report 2019 Higher Education Edition, publiched now by EDUCAUSE, is now available. You can read it if you want. (44 page PDF) It will tell you that mobile learning and learning analytics are coming soon. In three years you might be experience mixed reality and artificial intelligence.
Geoff Cain responds to David Wiley's post from yesterday. "We need to bring OER out of the teaching & learning communities that use them, not the corporations," writes Cain. "David’s post is overly-concerned with business interests. I think we need to get away from commercial publishing because of costs and that it is an out-dated and inauthentic model for teaching and learning." Moreover, " By focusing on justifications for commercial models, we do a great injustice to all of the great work that is being done in Open Pedagogy." Quite so.
I like the thinking here even if I'm not keen on the overall result. The thinking begins with a set of astute observations about what students actually learn in a classroom. It proceeds by identifying several layers of curricula - the recommended curriculum derives from experts in the field, the written curriculum is found in the documents produced by the state, the taught curriculum is the one that teachers actually deliver, etc. What the students learn is ultimately different from any of these - as Dylan Wiliam says, “children do not learn what we teach.” John Barrett concludes by proposing, in diagram form, a 'learning alignment model' (based o John Biggs's. constructive alignment). It has the usual failings of educational theory - it's a taxonomy, in the form of a pyramid, that is ultimately unsatisfying.
I can't resist posting this, even if only because "the countries that had the highest average scores were Canada, Ireland and Australia." The article is referring to "the Times Higher Education (THE) university impact rankings, released for the first time on 3 April" where "THE has tried to measure universities on what matters, rather than on what can be counted." And what does matter? "Rankings are aligned to 11 of the UN’s 17 sustainable development goals. These goals – no poverty, zero hunger, good health and wellbeing, quality education etc – matter to us as a nation." Now while I like this ranking system, it does underscore a point about rankings made previously in this newsletter - ranking systems are advocacy instruments, where publications define their own version of 'good', and then attempt to convince institutions to align with them in order to improve their ranking. This remains true, whether or not I like the measurement or the outcome.
Continuing the discussion of AI-generated music begun yesterday, we introduce MuseNet, " a deep neural network that can generate 4-minute musical compositions with 10 different instruments, and can combine styles from country to Mozart to the Beatles." Significantly, " MuseNet was not explicitly programmed with our understanding of music, but instead discovered patterns of harmony, rhythm, and style by learning to predict the next token in hundreds of thousands of MIDI files." Via Helge Sherlund and Tony Hirst.
I like it when an author takes something that is seen as obvious by most observers and questions it from a meta-perspective. Such is the case here with learning assessment, which the authors note are provided by a market - "there are data producers, there are data consumers (countries, policymakers, international agencies and researchers), and there are goods and services exchanged for money (prices) to produce the assessment data." They then move to the meta-level: " While the specifics of a market will obviously vary, there are two central questions: does it allocate resources efficiently and equitably?" The answer? Not so much.
Martin Weller offers a response to my suggestion that connectivist MOOCs ought to be used to scale learning. "I’m uncomfortable with this over-reaching of connectivism," he writes. "Open universities across the world have been operating large scale, open, equitable learning for decades." Have a look (it's well-argued). Then you may want to consider my response. I write, "the cost of educational labour is what makes it so expensive... (but) The connectivist approach is to do whatever can be done to help students perform this labour themselves."
This newsletter is sent only at the request of subscribers. If you would like to unsubscribe, Click here.
Know a friend who might enjoy this newsletter? Feel free to forward OLDaily to your colleagues. If you received this issue from a friend and would like a free subscription of your own, you can join our mailing list. Click here to subscribe.
Copyright 2019 Stephen Downes Contact: firstname.lastname@example.orgThis work is licensed under a Creative Commons License.