How many quotes are you using in your papers that the insertion of quotes is 'tedious'?

Just reading this heading makes me not want to continue. What problem is so bad that this is the solution?

This isn't important to me. My reference library (in Zotero) just generates the citation, which I manually paste into my note (in Obsidian). However, I'm not trying to integrate my annotations on PDF into Obsidian either, so that may be important here.

Because 1) the tools change all the time, and 2) we're interested in better workflow rather than trying to get people to use specific tools.

Stephen Downes commented on this post:

According to the authors, "there are skills that AI cannot master: strategy, creativity, empathy-based social skills, and dexterity." The article goes on to explain why, and to describe how humans will apply these skills in order to work with AI. My perspective is that we would be very short-sighted if we assume AI will not be able to master these skills. AIs already display more dexterity than humans in some cases. They are also demonstrating interesting forms of creativity. A lot of what appear to be human-only skills are the sorts of automatic non-cognitive abilities we demonstrate. But that's precisely where AI will excel, since these are based on pattern recognition and response. An AI won't feel things the way we do. But it's not al all clear that feeling a certain way is essential for any of these skills.

But, aren't these multifaceted, situated, and values-driven modes also constituted of data? Isn't everything represented by data? Even 'subjective' understanding of the world is articulated as data.

Is there some 'standard' definition of data that I'm not aware of in the context of this domain?

This seems like an over-reach. There's nothing about testing that inherently implies that students 'should' progress at the same rate.

Is this bad? What is the alternative to using data to inform decision-making? How is 'data' being defined here?

Recommended by Ben Williamson. Purpose: It may have some relevance for the project with Ben around chat bots and interviews, as well as implications for the introduction of portfolios for assessment.

Bret Victor shared this post when making the point that some of the 'best minds' in Silicon Valley may not be the best people to be working on the problems we care about.

Together with the post, titled, "I saw the best minds of my generation...", he shared a screenshot of a quote: "In the beginning, EA [Effective Altruism] was mostly about fighting global poverty. Now it's becoming more and more about funding computer science research to forestall an artificial intelligence-provoked apocalypse."

Bret Victor shared this post to make the point that we shouldn't be worrying about sentient AI right now; that the melting ice caps are way more of a threat than AGI. He linked to this article, saying that corporations act like a non-human, intelligent entity, that has real impacts in the world today, that may be way more consequential than AI.

Ben Williamson shared this post on Twitter, saying that it's a good idea to remove the words 'artificial intelligence' and 'AI' from policy statements, etc. as a way of talking about specific details of a technology. We can see loads of examples of companies using 'AI' to obfuscate what they are really going.

In other words, don't aim for a life devoid of problems. Instead, look at problems as opportunities for meaningful engagement with your life.

See also Burkeman's conversation with Sam Harris on the Making Sense podcast.

Harris, S. (2021). Deep Time (No. 269). Retrieved February 16, 2022, from https://www.samharris.org/podcasts/making-sense-episodes/269-deep-time

This is a report of student satisfaction so it's irrelevant if you care about whether or not the product improves learning.

An important consideration when attempting to emulate real-world clinical contexts.

A + B is always greater than A.

Another variant on research design that proves nothing is the A versus A + B design. It arises frequently in studies of gizmos like iPhone apps, or PDAs or simulations, where one group gets the standard instruction and the other group also gets to listen to heart sounds or view ECGs on their phones on the way to class. Just as we need not prove that something is bigger than nothing, we also do not need to prove that something + something else is greater than something alone.

From Norman, G. (2014). Data dredging, salami-slicing, and other successful strategies to ensure rejection: Twelve tips on how to not get your paper published. Advances in Health Sciences Education, 19(1), 1–5. https://doi.org/10.1007/s10459-014-9494-8

We should not be gathering information because it's interesting.

Evidence that this is the case?

Two minutes to resolve a connection issue seems like not enough time. Also, knowing this will cause more anxiety.

Simply knowing that this is the case is enough cause for additional anxiety.

There isn't any evidence of this either (i.e. of widespread cheating). You have to question the motives of companies selling a product that relies on this being widely believed.

Again, we do this all the time with F2F assessments i.e. there are exam invigilators in the room to watch for cheating. Why have there been no complaints that "faculty members don't trust them"?

Compare this, in principle, to what an exam invigilator does with F2F assessments; 1) they ensure that students aren't communicating with anyone else, 2) ensure that students aren't reviewing banned material in the exam, 3) monitor where students are looking during the exam. No-one has ever seriously complained about that. If the assessment took place in a venue that wasn't a students' room (i.e. the process removed the concerns that someone is looking at their bedrooms), would everything be OK?

Students have to be involved in the development and policies around how the technology and data are being used. And above all, the process must be transparent.

You can change the nature of your assessment tasks so that you don't need the software. See Killam, L. (2020, April 6). Exam Design: Promoting Integrity Through Trust and Flexibility for a great example.

I certainly agree with this statement but I wonder why students have never resisted F2F proctoring during assessments. We've always had invigilators looking over the shoulders of students writing exams. Why is it different when the proctoring system is remote? And note that this isn't about the potentially biases AI that informs the systems; this article is (so far) only pushing back against the presence of the system and not it's accuracy.

The article is a bit confusing wrt to what it's actually about. The question around wearing a biobutton to track possible Covid-related factors is one thing, but remote proctoring for assessment is another. I get that the article is about 'surveillance' more broadly but the focus of attention is scattered.

This isn't the problem. The problem is when the button is normalised and the programme expanded. Even if no-one has any plans for this right now, it will absolutely be the case that whoever is paying for the system will want to see it used for other purposes.

Again, the framing of the situation is key. As is situating the process within a transparent policy that 1) students contributed to in the initial development, and 2) has a built-in sunset clause that causes the policy and process to stop unless it is re-evaluated and renewed by all stakeholders.

How the situation is framed makes all the difference. Many would look at this and think it seems quite reasonable, given what's going on with Covid atm.

Much like Apple's AirTags.

I'm more concerned about 'feature creep'; once the infrastructure is in place (at significant cost), no-one is simply going to dismantle it. Instead they're going to keep adding features to the system that sees it become increasingly integrated into school and social life.

I agree with all of these concerns. But I also wonder if these same people use Facebook? If they willingly post photos of themselves and friends in their bedrooms? And if they worry about social media companies building up profiles of them and their relationships that are never deleted?

Formal meetings should have administrators present who are responsible for capturing the content of discussions and outcomes of decisions. And nowadays, why not just record every meeting? And then use automated speech-to-text transcription to get a searchable set of notes.

Good to reflect on. Even if you never use the note again, you should probably make the effort to create it in the moment when it first seemed interesting. Because you never know what the future will bring, you never know if the note will be useful or not.

This means that you need to strike a balance between going down the rabbit hole of trying to capture everything you read (just in case it's useful), and capturing nothing (because maybe it will never be used again).

I'm not 100% sure I agree. You can still only keep a limited number of items in your working memory. Writing notes - for me - is about having access to a collection of ideas that I've had over time. It's more about a repository of information that's important to me.

Every intellectual endeavour starts with a note. - Sönke Ahrens (2017).

I used to think that taking notes was an important but relatively small part of my work. But now I think of note-taking as my actual work.

See also: Bailenson, J. N. (2021). Nonverbal overload: A theoretical argument for the causes of Zoom fatigue. Technology, Mind, and Behavior, 2(1). https://doi.org/10.1037/tmb0000030

What does the 'digital first' version of these publications look like?

I wonder if it would be useful to distinguish between research and scholarship, where formal research is but one type of scholarly practice?

If we look at a journal as a channel for promoting scholarship then there's no reason that we can't include essays as a category of writing.

The aim of the book.

I'm intrigued. I don't see an easy route from where we are to a society giving up on its professionals.

See here for more information.

And possibly law.

Such a natural fit.

Because the pictures convey some of the 'feeling' of what the participants shared.

This is probably more a function of how society treats comic books i.e. they're for children. However, graphic novels and comics (illustrated stories) are no less of a communication medium than academic papers, and are certainly more appealing to read.

This is important; in order to this kind of presentation and dissemination to be done well, it needs to be funded.

A graphic novel is probably more likely to be read by funders than another academic article.

This isn't only a creative way of sharing findings for a general audience but is also aligned to the needs of a very specific audience who would benefit the most from reading the study findings.

We are hard-wired to pay attention to stories that emphasise the human condition in all it's variations. Story is a format/communication channel that we may have evolved to privilege over other forms of information sharing.

Does the fact that the data themselves were already in the form of stories, lend itself to sharing findings as a graphic novel? Could an RCT also be published in a similar format?

Stop replying with things like, "Thanks for this".

The ultimate purpose of any email message is to be archived or deleted.

Email is a chaotic task list organised by other people. When you think of it like that, it makes sense to get those tasks out of your email inbox and into another system that you control.

Similar ideas discussed here: Ellis, B., & Rowe, M. (n.d.). Guided choice-based learning (No. 4). Retrieved November 29, 2021, from https://inbeta.uwc.ac.za/2018/02/09/4-guided-choice-based-learning/

I'm not sure that I agree. Do we want our students to be able to change the map, or to choose different pathways through it?

Similar ideas here: Stephen Downes (2015). Design Elements in a Personal Learning Environment. Invited talk, Guadalajara, Mexico. https://www.slideshare.net/Downes/design-elements-in-a-personal-learning-environment-52303224

What makes an 'online course' different to an 'online learning community'?

Everyone's PLE should have the capacity to integrate with formal (institutional?) learning environments. But, will every formal learning environment be able to link to informal ones?

We all think in different ways, so a good system for learning should enable every learner to structure their learning environment uniquely.

The best way to learn about something is to do it.

How do we find the connections we don't know we're looking for?

It's interesting to think that we've gone through maybe a decade of startups (e.g. Slack) telling us that we need to get away from the chaos of email, and now we're being pitched the idea that email is a quiet, more intimate space.

Scannable subject lines in emails: "your subject line should tell readers something specific and valuable. Summarize 2–3 of the biggest items in your update and separate them with a comma or em dash so ideas stand out at a glance."

As much as I like the idea of putting your content out for free, I get that people need to make money if this is the business model. Schools and universities are probably under less pressure to do make a profit but still need to cover basic costs.

Not only new media; this would probably also work for schools, or any learning community.

I agree with everything in this post; I want to listen to more academic work in audio formats but find the process quite unsatisfying.

Either I find my attention drifting, or I’m switching between apps to try and capture the essence of something I’d like to come back to later.

I’m hoping that things like Momento help get us closer to the ability to capture information from audio sources, but this would need to be built into ebook readers or the operating system itself, in order to be more broadly useful.

It’d also need to be more reliable with respect to the quality of the machine learning transcription. At the moment it’s just useable, and requires a bit of interpretation.

Karnofsky suggests that the cost/benefit ratio of how we typically think of reading may not be as simple as we intuitively expect i.e. we think that 'more time' = 'more understanding'.

If you're simply reading to inform yourself about a topic, it may be worth reading a couple of book reviews, and listening to an interview or two, rather than invest the significant amount of time necessary to really engage with the book.

A few hours of skimming and reviews/interviews may get you to 25% understanding and retention, which in many cases may be more than enough for your needs of being basically informed on the topic. Compared to the 50 - 100 hours necessary for a deep, analytical engagement with the text, that would only get you to 50% understanding and retention.

That being said, if your goal is to develop expertise, both Karnofsky and Adler ('How to read a book') suggest that you need a deep engagement with multiple texts.

Good advice on creating secure passwords.

I have no problem with publishers making a profit, or with peer reviewers doing their work for free. The problem I have is when there is such an enormous gap between those two positions.

If publishers make billions in profit (and they do), while at the same time reviewers are doing a billion dollars worth of work for free, that seems like a broken system.

I think there are parallels with how users contribute value to social media companies. In both cases, users/reviewers are getting some value in return, but most of the value that's captured goes to the publisher/tech company.

I'd like to see a system where more of the value accrues to the reviewers. This could be in the form of direct payment, although this is probably less preferable because of the challenges of trying to convert the value of different kinds of peer review into a dollar amount.

Another problem with simply paying reviewers is that it retains the status quo; we keep the same system with all of it's faults and redistribute profits. This is an OK option as it at least sees some of the value that normally accrues to publishers moving to reviewers.

I also don’t believe that open access - in it’s current form - is a good option either. There are still enormous costs associated with publishing; the only difference is that those costs are now covered by institutions instead of the reader. The publisher still makes a heart-stopping profit.

A more elegant solution, although more challenging, would be for academics to step away from publishers altogether and start their own journals, on their own terms.

I like the idea of designing the types of conference experiences that I'd like to have, rather than what participants have come to expect.

I like the idea of Discord as a backchannel but it suffers from the problem that it's a relatively niche app, and no-one is going to install and learn how to use it just for a conference.

I think that Discord would work well for a learning community though.

Just because a dataset is publicly available doesn't mean that you can use it to build commercial AI software.

For me, presentations represent a well-organised set of concepts that I'm familiar with, and that are relatively well-connected to other ideas.

I have a sense of what this might mean but I have a hard time conceptualising how I might implement it.

This is something I still haven't got around to in Obsidian. The only use case I can think of that might help with my workflow is when I'm doing a weekly or monthly review. But it only takes me a couple of seconds to open the notes I need for that anyway.

Obsidian has since released a mobile app, with a paid-for sync service. I was using Dropbox to sync my notes between devices but had paid for the Obsidian sync service to test it. It works well but I hardly do any work on my phone, so it's not clear to me what the use case is. I probably won't renew my subscription and will go back to syncing with Dropbox.

I tried Todoist, which aims to bring the resources into the to do list. It started as a to do list that you can organise into projects, and then added notes and attachments to items. I decided to go all in for a year, paid for the premium service, and put everything into Todoist. But it didn't work for me because my to do list was still separated from the rest of the work I was trying to do.

I use transclusion in Obsidian for my weekly review. The weekly review note (also defined with a template), has a section that gives me an overview of my week.

I only ever diaries for keeping track of appointments, but the concept of a daily note (that I define with a template) has been enormously helpful for me to structure my day.

My project 'overview' note includes headings like: Admin (meeting dates, etc.), people involved deadlines, related notes. Each of these other parts is a separate page, linked to with wikilinks.

I have Project pages in Obsidian, with sub-projects that are separate pages, linked to the main project page with wikilinks. I do it that way so that I don't need to worry about folders. But now I'm wondering if a folder/project would be a way to compartmentalise things.

An overview of how Ton Zijlstra uses Obsidian as part of his workflow

Use of tools has ‘gone back’ to what he used in the 90s, which is to say, he’s moved away from things like Evernote (silo'd) towards using mostly text with Obsidian (note-taking) and Tinderbox (on the Mac, for writing).

Didn’t try to bring all of his Evernote notes into Obsidian, because it was a massive dump of content - an unorganised mess - and would have just created a lot of noise. Instead, he mined his blog posts for the most useful information. Still has the Evernote notes as markdown files, so still available when necessary. So he started using Obsidian with a blank slate.

Uses Obsidian as a viewer of his text/markdown files. Plain text files make it easy to maintain your notes across different platforms and systems, while allowing creativity and innovation to happen as a layer on top of the files.

Tools change. Every tool is temporary, so avoid having tools that drift too far away from the core feature set. If you're going to install plugins that provide additional, more powerful functionality, that functionality isn't going to be present when you look at your notes in plaintext. Avoid plugins that are going to make it difficult to view the content of your notes in other tools.

You need to be able to move on to other tools with low/no friction.

Uses folders to organise information, mostly related to projects, with an ‘overview’ page that links to other parts of the project.

Finding a new tool provides a certain level of fascination. But, we need to ask if the tool is genuinely useful with respect to the work we do, or is it a shiny new hammer to hit nails with?

He has replaced several tools with Obsidian (notes, to do, outliner). He says that Obsidian has replaced about 2/3s of the steps in his system. This is my problem at the moment…I’m happy with the number of steps in my system, but not that each step needs it’s own tool.

Don't write about the tool. Write about how you use the tool to do real work. You see this in the YouTube channels of people who spend hours talking about the tools they but they don’t show themselves doing real work with the tool.

Still not happy with the relatively high friction of turning notes notes into output. Works in Obsidian a lot to generate notes but still struggles to push out artifacts. I also want to figure out how to move my notes and reflections into output, with less friction.

There’s a tension between writing for yourself and writing for an audience. Do you want to publish your daily notes? Would this lead to self-censorship? Would you be more self-conscious?

Use blogs as part of an obligation to explain. If you need to search for information to do something, maybe you should blog that process to try and make it easier for others to do things. I like the idea that we have an ‘obligation to explain’ because we have all benefited from others who have explained things for us.

Tools will come and go but your process should be more stable over time.

I agree with this in principle. However, the process of using the core Obsidian plugins to create Daily notes with templates is so simple that you could easily swap out the plugin with a manual plain text alternative.

I've been thinking about this under the frame of, "The more time I spend with a tool, the easier it should be to leave it". What I mean by this is, the more I invest in something (time, cognitive energy), the more reliable it needs to be. Since nothing is completely reliable, I need to choose a tool that allows me to move to another tool without any friction. I don't want the tool I'm using today to be bought out and shuttered by another company, so I have to work in formats that make it easy to adapt.

I had no idea of the number of monitoring services that were available and actively implemented in schools. The tone of the article suggests that these services are successful at reducing the risk of harm to children but provides no evidence other than anecdotes from those responsible for implementing the service. Hardly convincing. There's a halfhearted attempt to show the other side of the story, from critics, although the airtime given to the 'successes' makes it clear which side the writer comes down on .

What happens when other keywords get added to the algorithms search criteria?

If this is such an amazing service, why aren't we rolling it out to everyone, including adults? How would those parents respond if they knew that private companies were monitoring them, in the interests of their own safety?

Again, the language suggests that this kind of surveillance is actually making kids safer. Which I doubt.

So, you're going to be sneaky about it because you know it's wrong.

This is a bad idea.

Have these tools really prevented anything? Or is prevention the aim of the tools? This is written as if these tools are actually achieving this, which is hard to believe.

Improved your written communication.

Absolutely. If someone else, in a similar context to you, has been through the hard process of evaluating information to already make a decision that you're facing, why wouldn't you try and learn from them?

So it's really important to identify which decisions are important, and which really don't matter. Spending time on decisions that don't matter is a waste of time.

This is a good step. However, even in higher education where mentorship is reasonably common, few mentors will actually invite a mentee to watch them work. How may mentors will say, "I'm about to spend an hour working through emails. Would you like to come and watch?" And yet, managing email as part of workflow is a significant part of the administrative role of a professor in higher education.

And yet we have an expectation that our leaders, health care professionals, teachers, etc. are all able to make good choices.

International summer school run from the University of Bologna in July 2020, with the theme of AI in medicine and medical education.

Students attended lectures and interactive sessions, and then worked in groups on a project to develop an AI tool and literature review that identified a real-world need.

Potentially useful format for generating new ideas and projects, and providing a baseline foundation in basic AI concepts.

The use of a pre- and post-test survey design baffles me though. Who cares about student satisfaction re. self-assessed competency and quality of lectures, etc.? If you really want to know how useful the programme was, you get an external panel to evaluate the quality of the students' projects.

Some questions they could consider in the evaluation might include:

1. Does the project identify a real-world clinical or medical education problem that has a high impact on patient outcomes or physician efficacy, for example?
2. Is the project likely to make some significant progress in moving that problem forward?
3. Is the solution elegant i.e. what variables are being included in the model? How technically simple is the model?
4. Are there any blind spots in the students' perspectives, especially re. patient data, bias, etc.?

Compelling idea; Instead of working for years to make the mountain of knowledge bigger (and therefore harder for others to climb) you could spend years making it easier for others to get to the top.

There is no equivalent of this in academia; where someone with less experience gets to shadow someone in terms of their daily work. A simple example might be to watch how a successful professor manages email.

What scale should this list be at? "Climate change" is an important problem that people should work on but it's so big that it's practically useless. Or maybe the list is so granular that the connection to the larger problem isn't clear. What's a good scale to think about problems?

Choose good problems and develop expertise in the skills needed to explore those problems.

Focus on skills, not careers.See Wiblin, R. (2021). Holden Karnofsky on building aptitudes and kicking ass.

Holden prefers to focus on ‘aptitudes’ that you can build in all sorts of different roles and cause areas, which can later be applied more directly. Even if the current role or path doesn’t work out, or your career goes in wacky directions you’d never anticipated (like so many successful careers do), or you change your whole worldview — you’ll still have access to this aptitude.

It's really valuable to work at the intersection of 2-3 domains; you don't need to be the best at any one of them but you may be the best where they intersect.

This is such a compelling idea; explain difficult concepts well in an attempt to help others progress faster in their learning.

See Victor, B. (2014, December 22). The Humane Representation of Thought, for an insightful discussion of how we can think differently about information presentation.

The choice of the table itself must convey additional information; what is it about your data that is highlighted through the presentation method?

It's not only the information presented in the table, but the table itself, that must serve a purpose. You can often present the same information in a different format, so when should you choose a table over something else?

Interesting to note that this is exactly how things emerged from my thesis; I obviously had some ideas about theory early on (and I wrote about the theory early in the thesis), but I really 'arrived' at my theoretical position almost at the chronological end of my PhD process. And then I had to go back and rework substantive parts of the thesis.

But then is 'theory' really being used i.e. if you can separate out the theory into a standalone section/chapter, is it really part of the work you're doing?

Understanding theory well can give you deep insights into the work you're doing.

You have to do something with theory; it's a couple of paragraphs that you add to satisfy a committee or journal checklist.

See Varpio, L., Paradis, E., Uijtdehaage, S., & Young, M. (2020). The Distinctions Between Theory, Theoretical Framework, and Conceptual Framework. Academic Medicine: Journal of the Association of American Medical Colleges, 95(7), 989–994.

This is why I have a problem with university research committees that have now added the 'theoretical framework' requirement to proposals, regardless of whether it is indicated or not.

Unless you schedule this time to review your notes and work with them, they'll just scroll off your list.

I have no doubt that this would be valuable but it assumes that you're sitting at your desk and can make these summaries. But if I'm reading away from the computer (i.e. a lean-back experience) it's really inconvenient and impractical to do this.

I use a similar approach with some of the writing tasks I give to my students:

1. Start by writing a short reflection on what you think about topic.
2. Read a couple of mainstream news sources and update your initial reflection.
3. Read a selected paper (or more) on the topic. Update your writing again.
4. Send your draft to two peers for feedback on what's missing and how to improve what's there.
5. Update your piece again.
6. Submit your final version for grading.

If I'm not familiar with the topic I tend to read for information i.e. I'm just looking for the big ideas. I tend not to add my own annotations to these texts and only highlight. My purpose is to extract information. If I know the topic relatively well, then I'll engage with the text through annotation and more critical reading.

'Reading' doesn't start and end with the first and last pages.

We'll always be inclined to take the path of least resistance.

Set aside your most productive time of the day ('productive' in terms of when your cognitive energy is highest) and move an important project forward. In the moment it may not feel like a big thing but over time those small movements get you closer to achieving something of value.

And this is a value-based judgement, not an economic one.

You can never feel on top of everything because there are always more things.

Which is why it really makes sense to think carefully about our commitments. What is the opportunity cost of taking on this new thing? What does it prevent me from doing?

I'm often taken aback at how poorly we (or maybe just me) understand time over the longer term.

You don't necessarily have to be the first to describe an idea; you just have to describe it in a way that lots of people can understand it. There's some value to publishing 'explanations' of difficult ideas.

It's largely because of these questions that I decided to shift my research slightly, from educational technology in general, to AI and machine learning in higher and professional education. I think that AI and ML is an important and pressing problem, and while it's not neglected in technology circles, it is not even on the radar for most health professionals and educators.

Algorithms don't tackle problems using the same logic that we do. This is a good thing if we're interested in exploring the solution space for difficult problems, but possibly a bad thing if we want it to drive us to the airport.

Leadership may be lonely.

Sometimes you don't know what you think until you try to explain it.

Talking to a trusted friend can reflect your thinking and ideas back to you, in a different, nuanced way that you may struggle to do by yourself. It's another way to think about how you think.

The fact that they're still around suggests that there's something in them that has stood the test of time. Compare this to most books that, even if they're popular today, are forgotten by tomorrow. What are the lessons you can learn from books that have been around for a hundred years?

Cal Newport talks about the idea of craft. Of mastering a difficult skill through sustained commitment to challenging work.

The answers to the most important questions that confront us are not to be found in distraction.

See Postman, N. (2006). Amusing ourselves to death: Public Discourse in the Age of Show Business (20th anniversary ed.). Penguin Books.

We use media to distract ourselves from more useful and interesting pursuits.

Also see Huxley, A. (1946). Brave new world. Harper & brothers, which confirms that this is not a new phenomenon.

These other activities make you feel like you're busy when you're just running on the spot.

Cal Newport's concept of Deep work.

Context-switching comes at a significant cognitive cost.

Is this 'moral courage'? I would think of this as 'conviction', which is different. Also, given the scandal in his personal life, maybe holding him us an an exemplar of morality isn't ideal.

This needs a commitment to a vision.

Not just in warfare; these characteristics are necessary for leading in modern society.

"Managers manage within the context of a prevailing paradigm. Leaders take us from one paradigm to another." - Joel Bankeur

Leaders have a vision.

Leaders are thinkers.

After all, this is what basic education trains us to be.

This seems especially harsh. I'm not sure that this is a reasonable default position.

But we've also said that 'excellence' (actually, 'success') isn't the same thing as leadership.

I've often said that universities are big machines that keep going, more from momentum than from anything that any of us does. But this adds a bit of nuance. Yes, there's momentum but even momentum runs out eventually. You need people to keep the momentum going. These people, while in positions of authority, tend not to be leaders.

All systems have frameworks that you need to understand if you're going to navigate through them. When you don't understand how the system works, you're less likely to make progress.

Being successful doesn't necessarily prepare you to lead, but does that also mean that good leaders need not be successful? Is there any kind of relationship between these ideas, or are they completely separate?

The measures we use to determine 'success' are not the same measures we should use to determine 'leader'.

Exactly. These are not necessarily leaders; but they do seem an awful lot like conformists, or followers.

This doesn't seem reasonable. There's no way that all of those extra-curriculars had anything to do with what the student was applying for. It's just a signal that this is someone who puts everything else aside to build up their CV. Is that a good thing?

Distributed practice, Retrieval practice, Interleaved practice, and Elaboration.

ideally, you'd like the students to identify the differences between their own performance, and the standard or reference goal.

Again, an entirely reasonable conclusion to reach; a far cry from the "algorithms are racist" and "data kills people" tone of the article.

This is a far more nuanced and reasonable conclusion than almost the whole of the preceding text suggests.

We need different rules for AI models used in healthcare; if you want to use ML in healthcare, maybe you don't get to keep your algorithm proprietary.

Perverse incentives in academia.

This is a serious claim. It'd be good to link to something more suitable than an Ars article with the headline, "...may be killing thousands of people". Disappointing.

This isn't good writing. They're not "data scientists"; they're data scientists.

Isn't it more like "Algorithms can be trained on biased data that encodes racist behaviour among people"? Racism isn't coded into the algorithm. People are racist and the algorithms reflect that back to us. If the data were different then the algorithm would reflect that alternative to us.

One of the challenges of ML is knowing what models to use, when to use them, how to use them, and what to train them on.

See federated machine learning for a discussion on how we might avoid some of these challenges.

This doesn't seem to be a great analogy. A better one would be that we do have centralised processing, it's just mobile. When we do consult the internet or the library, we're still not going to a centralised source...those things are themselves decentralised. I'm not sure that this is helpful.

Short overview of federated machine learning.

This is exacerbated when dealing with patient information, which is regulated even more than other kinds of personal information.

It's more efficient and reduces the risk of data being shared.

This is especially problematic when sensitive data (e.g. patient health records) need to be sent to centralised datasets.

It's also very expensive.

How many times do you have to redo something that you didn't get right the first time?

Understanding takes time.

This reminds me of the "illustrated primer" in Stephenson, N. (1995). The diamond age. Although that's an AI-based system entirely, with no human interaction necessary.

So, not a 'teacher' who provides access to specialised information but someone who can guide and advise.

I'd like to see teachers working as coaches, reviewing learning dashboards and helping students interpret the data they're getting back from the learning system.

What's the teacher doing there?

Everyone says this. Now it just feels like 'the thing we need to say so as to not upset anyone'. It's almost like a polite in the right direction.

The challenge of professional programme accreditation and regulation is significant. If you're lucky enough to have a programme accreditor that allows flexibility, that's great. However, many/most professional regulators are conservative by nature and tend to push back against the kinds of teaching and learning suggested here.

Again, make undergraduate learning look more like the learning that professionals do.

Here's a reason: Many students around the world, maybe the majority, in fact, simply don't have constant, cheap, and stable internet connections.

I don't understand why it's so hard for teachers to recognise that this is what their own learning looks like. Why shouldn't we design learning spaces for students that reflect how we learn?

Is this anecdote? The experience of the author?

I'd like to agree with this but my experience has been that this has been the dominant response by far i.e. the delivery of content to students.

Also known as Goodhart's Law.

We should balance the use of both quantitative and qualitative data to inform decision-making re. research impact.

Recognise that citation count, impact factors and other quantitative metrics aren't necessarily aligned with broader contexts.

Odd phrasing language. What do you think informs judgements, if not data? There's this weird thing that happens when people conflate 'data' with 'numbers', which is just wrong.

Quantitative metrics are important but should be used alongside qualitative metrics. In the same way that mixed methods research designs see quantitative and qualitative data integrated so that each makes up for the weaknesses of the other.

It's not easy to marshal the resources necessary to make this argument; there's no single place that serves as a collection of the evidence you need.

Not just 'cultural' but also an ecosystem within which many stakeholders must function.