Module 3: Introduction

By Stephen Downes
Oct 25, 2021
Transcript of Module 3 - Ethical Issues in Learning Analytics: Introduction

Unedited transcript produced from audio by Google.

All right. It's not gonna let me do that. I can turn this off though. There we go.

So, welcome to ethics analytics in the duty of care, just checking life controls here. And I have nobody in the life meeting with me, it was gonna happen sooner or later, but that's okay. Not the first time, I've given a talk to an empty room and I'm sure it won't be the.

Last time I give a talk to an empty room. Let's just check now and make sure things are running. So I'll open up the the activity center. And there we go. And it looks good. Just check here, won't be over. Yep. Okay, so we're we're broadcasting live on YouTube and of course, I'm recording the audio.

So, and I'm the only participant in this zoom room chat session. But like I say that's fine. We can do this. We will do this. So the topic this week is ethical issues and learning analytics and I'm going to talk about that in a little bit. But I want to introduce first the assignment so that, you know, people have a good idea of what I'm looking at.

So I'm going to share this screen here. So what you should be seeing now is my my web browser and you should be seeing the activity sensor and there may be somebody watching live on YouTube. So if you are, I don't know if you are. I have no way of seeing whether you're watching.

I suppose I could. But but if so welcome and if you're watching this recording, welcome. So but what I want to do is go to the course outline and we'll jump into module three, ethical issues in learning analytics. And here's the task examples of ethical issues in analytics. So I invite you to take a look at the list of all ethical issues and I've created a list here and you might think perhaps reasonably that we're a missing and ethical issue.

Oh and we've got someone, we have a person. Hi mark. Oh my goodness. We're gonna need more to get his audio up and running and so on. Yes, I performed the unpardonable sin today of starting on time. I'm just teasing America. Oh there. Yes. Welcome. Hmm. I'm hardly hearing you.

I wonder why that is.

No, I doubt that it's you let's just check my microphone or my speaker settings. All that's why. Okay, try. Now you hurt me now. Oh yeah, that's much better. Okay. Yeah, it was going to we're in Iraq. I had bounced around to find this one. What was incorrect. The zoom links.

Oh really know how weird they were linked to the day. They were created not today. Oh for goodness sakes. So, I'm gonna guess readers, you know, like I was gradually, we did find it on one of the on, but not on the activities. Really through Friday. Oh, the yeah, I think reloading like I did update the activity center link, but if you have a cashed in your browser, you might need to reload that page.

Yeah, let's well, let's see, that's 29. Resume. Oh, I've done. Yeah, I've seen what I've done something really stupid. Yeah. In the activity room, the link simply links to the activity room. So yeah.

Will crank up the volume a little bit here too. So then I can hear him so. Well, let's fix that because it would be nice to have people coming into the right location. So, I should have a Invite contacts. Email, copy, invite link. And now I'll go to the activity center page.

Course activity video? No, of course activity slides and yeah, there it is.

And I was doing so well today, okay. And publish. All right I've put the correct link. Now, into the activity center, might be too little to late but at least it's there and now I'll just reload it. Sorry you have shared his email. No one. Well, maybe in theory, if she's subscribed but no, I don't.

That's what I said if she subscribed but then yeah. And yeah. Here we are. Well, okay. So what I was about to do is just to talk about the first assignment or the first task, there's actually going to be two tasks in this module. I've posted one of them.

The second one will be with the graph and I'm just trying to finish off the code for the graph. If I don't finish off, the code will use Matthias's original code, which works fine and it's a good exercise, but it's not 100% what I want, and I'm picky that way.

So let's come back then. So I'm going to share the screen so that people, including you can see it on YouTube or if you're in the in the live chat you can see analogs, just you. So oh somebody saying zoom is not appeared to be working. Maybe that was you.

It was me. Oh, okay, page.

Yeah, where

Yeah. Sheridan says yep, still looking for open. Okay, all right. So let me get the property, right. Can you drop the link into the zoom chat? Yeah, I'll do that. In fact, copy, you know. All right, so I have to break out of the share. So, stop share, chat.

There's the link. Now this is including something together. This is clinging something together, but it just goes to show that we go above and beyond to make sure that our second person is able to join the chat. What other course would do that aside from all of them?

And, Okay. Now, hopefully she. All right, so let me try out. There we go. We've got Sherita. I've just admitted her. Okay, and I guess she's just getting set up on microphone and

And video. So I had this thought this morning, I'll say it now, so I don't interrupt your shirt loan. That if we had an analytics of care set up, we would have reached out the Bernie to say I know. And if we had an ongoing learning community, yeah. Then we have relationships and we reach out to Bernie and say they're doing okay.

And is there anything we can do to help? Yeah. Yeah, exactly. Yeah. It's, it's one of the risks of being so decentralized by suppose. Yeah. Now selling another person landing or I don't know what now being was. So as just me making noise. Okay? All right. Now for I think it's the fourth time I'm gonna share my screen and we'll go to the first task, and I've made it a lot more straightforward than last time the same task, but it's more straightforward.

So we'll go to the course. Outline will go to the module module three. And here's the task. You can also just click on it and see it in its own page. Its own glorious page. But the idea here now is that we're looking at and categorizing ethical issues in analytics.

And this is actually an exercise other projects of done, and I'll actually even look at one or two in what I have to say a little bit later. But here's our list. Notice this time the page exists before the task. So, these are all the different ethical issues that have come up.

That I could find organized and classified by me. Now, I don't know if I've captured all of them, if you think I've missed one, click on submit your example here. Give it a title categorize. It, I'll have to fix that. Drop down and give it a short description. I'm looking here for, you know, types of ethical issues.

So if we look at, for example, surveillance, which is one that everyone brings up, right? So I've written a longer description that I would expect people to, but it's a type where I, if we look at lack of appeal, this will be a slightly shorter description again. I'm just providing an overview of it.

So, and then that's the first part, right? Have a look at this list. See if I missed any, if I have missed any added then the second part is to pick one of. These could be one that you've added could be one that already exists. Pick it and then read about it.

And then see if you can think of or see if you can find a paper or a web website. Whatever that talks about, this issue in particular and click on the link to suggest it. And here this is your submitting, a link. So we need the link title, whatever that might happen to be the link, you are all.

And then you provide a short outline of the of whatever's, on that page that you are all. So, if it's an article, just quote, right? A quick summary, the article, if it's a product, it might be a product that raises the issue, right about. That product, that's the task.

And what we're doing basically is we're feeling out this table with issues and examples of these issues so that we have references to show that. Yes people really have raised such and such as an issue, the second part and I don't know if I have the link open at all.

I'm just looking to see if it's too many of my tabs. I don't appear to have it is the one where we will, I know where I can find it. It's in my newsletter. Oops. Go home. Oh, well, daily, here we go. So the second task, which I haven't posted yet, we'll use this page.

The link for which doesn't appear to be working. That's kind of weird. Yeah, I'm having a day today, aren't I? That's post. Wow. Oh, we got a third person in the waiting room. Jim's gopher new record. Yeah. Okay. So, I've put the link in properly in like, what post.

I'm having a day I tell you but I do know, I can find it.

Okay, copy

This is what happens when I think I'm overworked. Is that I make silly mistakes. Here we go. So this is from Matthias Melcher. So, Jim, I'm just describing the tasks right now. I'm assuming you're in the in the chat room. Let's just see if you're there. Oh yes. There you are.

Hi Jim. And there. I had it on my list to get here at 10 today, but my supervisor ran, the meeting 15 minutes over. Well, yeah, that's what happens. This is, I just delighted to actually be in here now, welcome. And yeah, this is your your first session in the course that you've joined.

Although of course all the videos are available including the first 10 minutes of this one. So we're watching the videos and actually was just before our meeting with the supervisor. I was watching your discussion from Friday on high speeds so I could get through the. That's what people who know me in person, wish they had is a high speed button so that it doesn't take as long to listen to me.

Okay, so where I was was talking about the second task. The first task is very similar to last week where I ask people to add an issue. If I haven't covered all of the issues or examples of the issues. Now the second task, we're going to use something like Matthias, Melchers projects in these, put the demo of it up here and I'll just open it up.

So now you can, how do you make it bigger or smaller? I forgot. Let's get help. Move connect to them from doesn't say how to make it bigger or smaller. There's got to be a way was, oh my gosh. Control my asthmatus. Well, I know there's a way I just don't know what it is and that's really weird.

And of course, you know, well,

Let's try those. I'm assuming your browser window is, it's, I can zoom the browser window, but the problem is this canvas is way larger than the window I have for it. Well, anyhow, hopefully it'll render better in yours and I don't know there's I'm sure there's a way to make it bigger or smaller, Dragon like hunting move it.

Oh well anyhow. On the left hand side, we have, he called the motivations for learning analytics, as opposed to applications or useless. That's pretty good too. These are all the things that we look at last week, over here on the right, are the ethical issues and that's the list of issues that I just showed you.

That I've compiled for this week and as you can see there categorized into groups like predictive analytics, prescriptive analytics etc. And similarly the issues are categorized. But and we'll be linking each of these to, you know, links or references as examples. The question is what issues are raised by different applications?

Let's take dashboards. For example, does a learning analytics dashboard of your class? For example, raise issues of privacy. Oops. So click on all. And then drag, I always forget how to do that. Click on alt and then drag and it'll draw a little line. Similarly does dashboard raise the issue of content.

Manipulation does plagiarism detection? I click on it. Now, I'm clicking on alt and holding out down. Does that raise an issue of privacy? Does it raise an issue of fear and things, etc? Now, the way Matias is done here, he's got all of the links or all of the applications and all the issues.

In one grand big chart, I'm going to try to break it down so that it's a bit more manageable. So yet you're not looking at everything mapping to everything but you're looking at subsets. And then I'll just have you page through different subsets for as long as you want.

But that's basically the idea is to whoops I always press shift instead of halt, any shift? No, but that's the idea. Just draw these links. So it's pretty simple task. Then The other thing that I'd like to fix, but this is the way it works. Now, right click on the diagram and then click on export and what that'll do?

You can you can either open it with Firefox or save it, if you open it, it looks like this. It's an XML file. So there are all the issues which is really interesting to us are all the links down at the bottom and this will include the links that you've added.

So this is of interest to me because what I'd like to be able to do is add these links to the graph that we're building in the online course that we're all looking at. So if you were to send this to me, if I can't figure out a way to just have this save directly into the course without you having to export it and download it, I'll do that other.

You know what I'd like? Right is even a submit button or something but even right click and then send to course, but if I can't do that in the time, then I'll give you a place where you can just click the click a web form in the course, and upload it.

And then I can take it from there and integrated into the course graph. So the idea here is that you're helping build these relations between applications of analytics and the issues of analytics. And, you know, this would be really interesting if we had a hundred or a thousand people doing this.

And then I can take all of these graphs that people did and amalgamate them together and give the links different weights for how popular they were. But we don't have a thousand people but such as life, it'll still be interesting to see what all of you come up with and compare that to some of the things that we're looking at.

Some of the discussions that we're looking at with regard to ethics and analytics. So that's the exercise. That's this. Second task and I'll turn it back over to you. First of all, do they make sense and second any questions? So this time instead of submitting an example, we're going to actually, we're going to be drawing those connections for a second test.

So submitting the example is task one for this module, so I've set it up a bit easier so that it's a bit easier. So, if we, if you go to, if you go to the module screen and module three, and then, click on the page that says all ethical issues.

You can submit a new ethical issue here by clicking on that or you can click on any one of these ethical issues and click on the link of the bottom to add a link, which is an example of or resources URL, a piece of software or yeah, it's like that, that does that.

Yeah. Okay, this, but it works better this time if that is to say, it works this time. So, yeah, it's around until the weekend, so that's, that's fine. Yeah, not a problem. So that's task. 1 is to add these, you know, a link, maybe more and then task 2, which I haven't put on the module page yet because I'm just trying to write some software to make it work.

But task to is to is to draw those links and I don't know why it's not. Oh, I'm in the wrong module. That's why there we go. So when task, two shows up, it'll show up right underneath here. Okay, so it's not posted yet. No, it's not posted yet.

Okay? Because you have first second. And so those are and then there's another one and that and you have a link to where we can. Yeah. Draw those connections. That's it. Exactly. Yeah. I'll provide the link with that task, and it'll show up in the newsletter. So either on our, by RSS, or by email, whichever you're using or website, whichever you're using, tell us that sounds

Yep, Select the area. I'm hoping it'll be fun too about.

Of course, I put some links in the day and and added multiple logos, but realizing that I was sort of redefining them. Yeah. Right. So, I don't know if I should ask you to delete all that and let it go back and add proper extension of categories. Let it ride.

Um, our are these categories of ethical issues? Yes. Okay. I'll probably just leave it for now. You see part of what I'd like to do in the discussion at the end and we didn't really do that last time because we got involved in other things. And that's fine is to talk about the sorts of the sort of thinking that we need to undertake when we're talking about ethical issues.

Categories of ethical issues, how to how to name them, how to describe them etc. So, having examples of what you put in, even if you now repudiate, those examples, they're still examples of how someone thought at some particular point in time. So if not, you would could have been anyone, right?

So we still have something that we can talk about and you know, as I think about this and you know this thinking is sort of forming for me over time as we go through this course. You know, we're sort of emulating a lot of the stuff that AI people actually need to think about and go through as we go through these exercises, you know?

So we've had a labeling exercise, we've had a classification exercise. Now, we have a linking exercise and in one of the modules later on, in the course, it's going to be structured along the lines of the decisions we make when we design AI and analytic systems. And we'll see at that time, a whole bunch of these decisions that we've already gone through in, in our work, through the course.

And I think it'll give us a bit better of an insight. It's already giving me a bit better than insight to the sorts of decisions that need to be made. And, and how they impact, you know, our understanding of concepts, as well as the ethics of these concepts. That's, let's see me how it's shaping out right now.

Anyways. And that's already an unexpected. But interesting thing to happen through offering this course. So our errors and misunderstandings provide a record of learning kind of like, yeah. So our creating students don't erase your answer. Don't need line through it. And but you're new understanding there, and you'll have a record of what you've learned.

Exactly. And so that's why I'm not gonna remove your mistakes when you submit them because we're gonna go put them on video reason. Another issue, is it possible to go back and add new categories onto a posterity made? I don't see how to I can do it. But no what's what?

You've got access to is a submit but not an edit function. No. Yeah, because the article is good and dropping something like this, you'll hear. Yeah, highest discrimination. Yeah. But they weren't used exactly. You have said and I probably could have invented a couple. Yeah. But nowhere. Okay. So, but feel free to add more stuff, you know.

Yeah, I think that's that's the way to go here. Just isn't a side the way I'm designing this and that, you know, like you're probably sitting there thinking about you. Why didn't you put an edit capacity into that and and decide from the horrible, technical problems that it creates.

Although, you know, I did have an edit capacity before and I have a whole permission system in grasshopper to make that possible. But where I want to go with this approach, is that individual people taking the course, have their own instance of grasshopper. And so have their own instance, of a way of creating these editing, these drawing, the links of whatever by themselves.

And then it's the result of what you've done in your environment that gets harvested or aggregated by the course, and then brought together with everything else, can't do that and word press or blogger. They just don't have the capacity, right? All they have the capacity to do, is you write some posts, you know, I was thinking about that.

Even you know, I wanted people to suggest for example, additional applications or additional issues. And what do I have? You do write a post. Well, you could but then how does the aggregator recognize it that post is intended to be an example of an application and put it properly into the database and have maybe if I had AI, it could do that.

But I don't have anything like any I that could possibly do that and and you know in that introduces any waste the possibility of ever into reading and categorizing your post. So better if you do it better, if you do it on your own system, but no such system gives you that capacity.

There is no system that you can use where that's possible. At least none that I found. Yeah, there's some concept mapping tools and things like that and maybe there are ways to explore that. But you know, the the whole overall work with these examples create these graphs, add these links, send it to a central course.

Nothing like that exists, maybe this time. Next year, I'm still working on grasshopper the PLE edition as well as grasshopper the course edition. And when I tried when I offered the course in 2018, I tried it with the two and it failed miserably, but I'll try it again. I still have two and a half years before I retire.

So there's plenty of time to write a large and complex software application. So here's a ignorant question can so can we go up a level and think of using the internet and now that APIs are so prevalent and with the group of people who have been trained and making those connections.

Mm-hmm, possibly just build this in Europe not using gratifying in using APIs policies, Is that theoretically possible? Again a thorough it's not only theoretically possible. I imagine it'll actually be done at some point grasshopper uses APIs to communicate between instances. So your personal learning environment would use an API to connect with my course so that sort of exists and and you could connect with each other's instances with APIs as well.

So that's what creates that kind of community and that's how we would share. For example, things like information that I've bought the link to the to the live discussion but our own unique platform. Yeah, yeah, yeah all of this should be open. So, in theory, you could use any platform, you know, as long as that abided more or less by the API.

And what's interesting to me is the graph on your version would be different from the graph on my version different from the graph on Sharita's version. Everybody has their own representation of the discipline or domain. And and I think that's interesting. I don't know what the application for that is but I think that's interesting.

Jim. You had a comment requestion earlier. Yeah, I'm looking at the all ethical issues page and it's says submit an example, when I go to that, submit an example, it's very familiar, the title, but the categories still has your descriptive diagnostic predictive rather than the categories that I'm seeing on the ethical issues page, quite right.

And that's a coating error that I'll be correcting. As soon as we're done this session, I wasn't sure if I was misunderstanding something or is skipping something but but it's okay. It's something you're still working on. That's fine. Yeah, I watched. So I was watching the I said was writing a blog post last night and commenting on how I had been watching you struggle through on Tuesday, but I'd already looked at the the page and knew that you had succeeded.

So it was kind of a weird time work. Yeah. It's kind of odd. That way isn't it? Well, I mentioned it in my blog post. Not that I need a badge or anything but the logic of asynchronous learning. I love it. Yeah. Okay. Thanks. That was my interesting. So we're I've got no with urine indulgence.

All right. Well, do I want to do that? No, well, I don't know, I prepared some slides to introduce this module. Are you guys interested in that? Or do you want to keep discussing? I'm seeing two Sherita. What are your thoughts?

Shem. Very indecisive which I'm trying to make little things work. So I'm only have half be attention. I hear you have to. I'm gonna have to leave very shortly. So I'm not, you know, the person you should be asking, probably marker Jim. All right. So if you check the recording of this later, it'll be at about the, the 40 42 minute mark, and you can pick up on that.

All right, let me just it's not along thing, but it would be nice to to have it. And if I can do it now, instead of doing it separately, there we go and set up. So,

Sorry it takes a little bit of time because Microsoft always wants to default to showing a slideshow on me entire screen and I never want, not even one. I presenting, I never want to slide presentation on an entire screen. All right, so let's come back here now. Let's see how we go.

There it is. Voila. So now you should be seeing my beautiful slides not yet, but it takes a bit. There we go. Yeah, because yeah. Needs to to load some large and beautiful images. I just love the colors on this image. Okay, so yeah, this is to introduce the, the third module of the course and it's hard to believe that we're already into module three.

But where we're at? Now, we found a ton of applications of AI and analytics and learning far more far more than just the, the typical applications. That people think of, like, predicting how students will do, or recommending content to learners a wide much, wider range of applications, and that, and I hope that, that part you're convinced of now that there are many, many applications of AI and learning many benefits that results.

So it's not really an option. Just to say, no, we just won't use this. Technology is just too much benefit but that's where the ethical issues come from. Of course, taking all of these applications into account is important because it's precisely in these wider accounts and analytics that the relatively narrow statements about ethical principles are seen to be lacking people talk about bias and and they talk about surveillance to name a couple or privacy.

Perhaps and yeah, these are concerns but they are by no means. The only concerns nor is addressing any of these concerns. A simple matter when you're looking at dozens and dozens of different uses of analytics and artificial intelligence. And what's really interesting about this technology is that it's possible to use it correctly.

It's possible to use it precisely, as designed is still reach a conclusion or an outcome that would violate or moral sense. Maybe not everybody's moral sense, but are all personal moral sense and it's possible to use analytics and AI correctly. Still do significant cultural and social harm, the oven as well.

Look at Facebook, right? We we don't need to look any further than Facebook. We've got the the Facebook papers being released this week. The, the great revelations that happened last week and it's just the latest in a long series scandals about Facebook, but what exactly are the ethical issues here and how do we address them?

You know, I read a lot of times is writing, that seems to suggest that we can address the issues with Facebook simply by blocking offensive accounts about for one thing. That's kind of a wackamall strategy, right? They keep popping up. He blocking them but they never go away. They're like spammers but also the it might be that the moral enough of a problems are simply problems about offensive content.

Lot of people have suggested that there are fundamental issues with Facebooks algorithm or even yesterday, I read an article about Twitter's algorithm and revelations. I did favors right. Wing politicians, right? Wing issues, others have suggested that, maybe it's Facebook's incentives that are wrong. The incentive to link to engage drags as deeper and deeper, as they say into the rabbit hole deeper and deeper into more and more radical content.

And again, that's not something that just plagues Facebook, this sort of accusation was leveled YouTube, not too long ago and that's where at first surface, or the problem might be just who Facebook serves and what they want. The purpose of Facebook, is to make money. It's a public company, they have a fiduciary duty to focus on making money, but making money as an objective.

When running AI analytics, may buy itself, because, cultural and social harm. So I think all of these considerations together, speak against simple and superficial understanding of the ethical issues in AI. There is a widespread demand that something be done right. You know, it's it's a rising market. I'm doing this course and I started my own personal investigation because two years ago, the president of NRC made a promise to the government of Canada.

That energy would look at the ethics of AI and there wasn't a whole lot done about it because it's kind of hard to do that. So I took it on myself to address the issue so I'm hoping for anatomy from the president. No, I'm just kidding. And there are other things being done today at NRC.

Now around ethics in AI in a number of papers have been written, but at the same time, we have NRC and elsewhere in our field. Generally are working in an environment of misinformation about the development complexity and riskiness of AI of all the different applications. All the different issues we do need proper debate about its development.

But again, you know, this document from the world economic forum that I'm referring to also says something along the lines up and I'm paraphrasing here. It's a fallacy to believe that these issues cannot be addressed through regulation. Well, I don't know how they know that, because I don't know that.

And Well I'm not going to say I'm smarter than them because I'm probably not. But I mean, if I don't know that probably, it's not proven to be true. Don't want to give myself too much credit there. You Nesco talks about managing the discussion and, you know, points to the two sides of it, the dangers of, or the consequences of misuse can be devastating, they say, and I think that's quite true.

Particularly in countries. Where if the wrong information reaches the wrong people, it can have very serious personal consequences. On the other hand hate division and life are good for business. This is where the the legal imperative to make money. Works against our social cultural needs. Also globally, inequity global inequality is mirrored online and now it's just thinking this morning about all the stuff, but I've read on ethics analytics and so on how much of it's coming out of private, elite universities and associated academics and and the countries that it's coming out of including our own here in Canada.

But, you know, Britain United States, Canada, Australia. And are these the societies and cultures we want determining, what counts as ethical NII. I don't think we have a good history here. I mean, there are many things that these societies and cultures have done, that is good. But there's no question that they've played a major role in producing the global inequality that is now mirrored in the online world and mirrored in analytics.

On the other hand, the potential benefits are enormous AI and analytics. Could end the need for work. That would be pretty significant but says UNESCO we need to agree on international. AI regulation. Again is regulation going to be what solves the problem. I think we need to dig deeper hence, the purpose of this course.

I I think we need to dig deeper, not just into the issues, but the nature of the sorts of issues and and the environment around these issues that comes up. And I only just found this diagram today and I haven't had a chance to study it in detail, but it really, it brought out a concept to me that really helped a lot.

I it's actually looking at the quality of a clinical environment, but it seems to characterize the discussion of ethics and morality a bit more generally. And when I looked at the diagram, I realized that it maps for point to. This course, the different things that we have to consider in ethics are not just ethical rules.

Ethical principles. What's right? And wrong. But rather a complex set of interrelated, factors moral climate, moral community, more identity, distress, sensitivity, agency, and integrity. And without a whole lot of work, as you can see, with the light orange text, I was able to map each of those to one of the modules in this course.

It wasn't a big force to do that now, how well that matching stands up to closer scrutiny, I don't know. But nonetheless, I think it's still gets at the idea that our understanding of ethics isn't simply an understanding of what's right, what's wrong? That there's a lot more nuance than that.

And that's why, you know, I encounter when I encounter statements that there is a consensus on ethics and analytics, you know, I really wonder about that and there's a study for example, of from fuel and other suggesting that we have reached this consensus and the links to it. There are on a slide and I spend a lot of time studying this document in particular.

These researchers look at a set of ethical principles and guidelines, will be looking doing much the same exercise in the next module. And they mapped those to the ethical issues that arise. And when you look at them, you look at this chart that they present. And it looks like, oh yeah, everybody has reached this consensus on ethics and analytics.

We all have this common sense of what the ethics are. Again, you look at who produced all of those documents on AI guidelines and ethics and coming from basically the same demographic. So that should be one flag, but more than the point, when you look at any of these ethical issues in any sort of detail, the consensus falls apart.

Yes, it's true. Three, four, five, eight, ten. Twenty different codes of ethics. Say something like privacy is important but when you ask them, what do they mean by privacy? They're working with three, four, five, 18, 20 40 different definitions of privacy. Where the importance of privacy plays out in different places and even more when you get out, you know, when you go outside, the narrow domain of studies addressing ethics and artificial intelligence, and look at wider discussions of ethics and privacy, generally, that consensus really falls apart, a couple of examples here on this slide, one from public library of science.

The headline for that article is sharing is caring but is privacy theft and I think that's pretty good question. You know, how do we do open science with privacy? What are the ethical decisions there? It's not simply project privacy end of discussion. Another example, the document on the left is a representation of the Panama papers, the Panama papers.

You know, 11.5 million documents etc. Describing how people are avoiding taxes by secretly stashing money in anonymous bank accounts in Panama. Now is that the value we want preserved and AI analytics and and, and learning. Generally, it's not clear to me that it is. Those are just two examples out of many that I can draw into the discussion here on the issue of privacy and privacy is just one of the issues about which there will be many different points of view of perspectives, different definitions, different statements of what's important and what's not and crucially different statements about what's ethical and what's not.

So this is the stage where we're at. This is the last slide. That's why I'm not moving forward. This is the stage where we're at now, in this course, the stage where we look at, what all of the ethical issues are that people have talked about and try to find.

It's not just you know where they've used some words and common and we can't avoid doing that because we have to work in words but to try to find all of these examples of where these issues are discussed to see the different definitions of different perspectives, the different understanding of the issue.

And when we understand the ethical issues of analytics, I think that will be in a position to say more about whether there is a global consensus on what ethics and analytics should be. So like to say, what's the last slide, I really should have a nice closing slide. But that's what.

At least I am up to in this module and we still have a little time for comments. Questions reactions.

Nothing allow each other because I first so I'm stepping back but and he thoughts. Jim, you mentioned a global consensus and how access is privileged you and Iron Canada? I think, Mark, you're in the US. How did we bring in? What were the some mechanisms we can use to bring in marginalized voices?

That's a good way. I'm not sure. I know how to do that, but I think it's and especially when most of this course is it is an English. Yeah, you know. How do we how do we bring those other voices in? Yeah. And let's be clear and we should be clear.

They're probably not going to be in this course at least in this offering of the course, because there are how many people in it and we all seem to share the same demographic and we can't avoid being what we are. But I think it's very clear and I think you make the point that these other voices should be heard and that, you know, any discussion of the need for the ethics, what the ethical issues are and and what the resolution of them would be requires this global consensus and that's why I have that section on the duty of care sitting there near these near in the second half of the course.

Because overall lots of philosophy that brings in many of those aspects. And you know, again the others there there are ongoing discussions of AI and ethics that, you know, I'm watching and listening to and sometimes offering comments to where this need is just simply, not considered. In fact, contribution from people outside the field of technology simply aren't considered as needed.

You know, it's a technical issue. Will solve it internally. I will present you with the solution That seems to be the attitude, and, you know, there's no way, right? Always title thing. So, don't think either of you think so either. Because based on what I've heard from you in the past gym, based on what I've heard from you in five minutes, AI translation.

I don't think it's there yet but that might help bring some diverse voices together. Yeah, it does help and helps quite a bit. Not helps me quite a bit like right? Yeah, I said good morning to a friend in Chinese not too long ago and got his got good morning and his name in Chinese.

All correct and he was quite surprised at this, you know. He was like, how did you do that? Because, you know, he doesn't even present his Chinese name when he's communicating with us in English, but it was possible, but it's not easy yet, you know, I have a, you know, publish on Twitter.

Publish a wordpress etc, capacity, and I'm building a grasshopper, right? I'd like to have a publishing French publish in Spanish or publishing all languages or even better. Just my browser, just converts whatever I'm reading into the language of my choice but we're not there. Yet getting there, but we're really close.

We're really close. I was reading a nutshell, right? And I actually am wondering is, if the original post was written English and then used chrome dressing with that. Because and there were no somatical errors that I saw. Oh, wasn't the native English speaker. There you see. I don't know.

Yes, you know. I don't know the person because somebody has stumbled onto. Yeah. Working it and open access, indie web and stuff. Yeah, so maybe an English native English speaker that posted up in an alternating right there. That was a lot of the second translation is to get hell compared to other experiences.

The interesting to hear someone like yeah. Tell us whether the Dutch was any good. Yeah, right. Yeah, right. Yes, you have no way of but, you know, even the translation is just the first step. I've had quite a few correspondences with people in Arabic over the years and it's not simply that they speak a different language, but they're the whole character of how they express themselves is different.

You know, there's the obvious, there's the religious overlay on everything but then also the way they relate to other people, the way they relate to each other is it's just a fundamentally different way of expressing one's self and the translation system translates. It literally but I don't think the literal translation is really getting at what they're actually saying to me.

You know, and we get that in French even as well. Whether there's a lot of idiomatic expressions, but Google is getting better at translating, but let's still some of them creep through. Yeah. I'm taking a course in agricultural communication right now and that's what it focuses on. Yeah. You know, there are these culturals ways of speaking him that we're not even aware of, and that it may, you know, at some point we're gonna come to where it's just a little too complex.

Yeah, get the full moon. Yeah, you know, even with human a layer of human conversation happening to the physics, it's still, we would take a, you know, we take a United nation's lifetime expert type person. Actually communicate most of the meeting and then there's those things that just can't do.

Yeah. So yeah and that's where we can't put aside relationships because having a relationship with someone well will prevent me from making a certain amount of Rome assumptions based on just hearing certain words. Yeah, yeah, that's where trust is assuming that intention. And yeah, all these parts of a relationship and screwed over those little three cars that is that a bias we can code into AI was that's a good question.

Yeah, to keep that question in mind for I think unit six or module six or seven. When we go back and talk about how we're building our AI systems in the decisions we make, can we build trust and relationships into it? You know, I would argue that different topographies will produce different results.

Along those lines, I'm much. You know, I tend to prefer something that might be called a community of communities model where we're not trying to connect everyone to everyone all at the same time. But even the community of communities model you need to form communities from different cultures and people tend to associate with their own culture.

So Well that's our time and I'm sensitive not just to your needs but also to the needs of the people watching on YouTube either live now or later on, but through the rest of the week I'll be adding more discussion of these issues. I'll be aggregating and reading your blog posts for sure.

I'm putting them in the newsletter. Yeah, count on. It was not just me that freezes and I'll be adding that other tasks and trying to make that make that work. And I'll also be catching up on some of the stuff that I didn't cover from last week because I had horrible technical problems that ate two full days, out of my life.

I'm starving a blast going this. I don't care. That ate two days over in my life. But I have fallen behind in some of the stuff that I wanted to distribute. So I'll be getting all of that in the newsletter through this week. And then, of course, we'll get back to together on Friday and talk about it.

Yeah, I have a strategy finding session, I can't get out of them Friday. I put it in my calendar just in case it gets a drop. By the way, I like it. You have a calendar links to it. That's really helpful. Good. I'm bad it was yeah, mostly it's like by the weekend I have time to catch up and dabbling.

Oh, that's fine. That's why these sessions are recorded, is sounds good. So that you can benefit from that. You know, this whole move is an experiment in, how do we make the whole mook accessible and useful and open to people. So and and so I'm trying to do that.

There's more stuff, I'd like to do. I wish we had a live show. Cast audio broadcast but we don't have that but maybe in the future. All right. Then I'm gonna wave goodbye and tell you. I'll see y'all next time when I see you again. You're never minutes even after you start recording, huh?

Yeah.

Stopped live stream. Okay, we're no longer recording. There you go. So this is brought up an issue. This course, is brought up an issue that I one of my hobby.

Force:yes