Artificial Intelligence in Education focusing on the Skills3.0 project

Editor's Notes

1. (Image from UNESCO mobile learning week 2019). Good afternoon ladies and gentlemen, colleagues. I am very grateful to be able to talk at this wonderful Elearning Fusion event. Thank you Bartek for introducing the topic of AI from such a broad perspective. In my talk I will focus on AI in education, and move from the general overview to a more concrete example of a project I am currently working on.
2. In the last hour, while I was in Poland, I used the following tools
3. At the end I will show you the learning solution I am working on, the so called skills 3.0 project.
4. The first online learning teaching machine was invented by Skinner (Harvard university) in 1954. At the same time machine learning took off, mostly shaped by computer scientists. It took a couple of decades before the term eLearning emerged, embracing all types of online education. In 2005 Roger Mougalas from O’Reilly Media coined the term Big Data for the first time, only a year after they created the term Web 2.0. It refers to a large set of data that is almost impossible to manage and process using traditional business intelligence tools. CCK Connectivism and Connected Knowledge course was a course organised by Stephen Downes and George Siemens, both Canadians. Dave Cormier was one of the participants of the course, and he named it a Massive Open Online course or MOOC. The first Learning Analytics & Knowledge conference was organized in 2011 in Canada. This development was crucial for the next steps in big data combined with education. Some researchers assess that the October 2012 ImageNet victory anchored the start of a "deep learning revolution" that has transformed the AI industry.  "Why Deep Learning Is Suddenly Changing Your Life". Fortune. 2016. Retrieved 13 April 2018.
5. Squirrel AI, the machine that regularly outperforms human teachers and redefines education by Wei Zhou Squirrel AI is an AI to respond to the need for teachers in China. Based on knowledge diagnosis, looking for educational gaps. A bit like an intake at the beginning of a master education for adults. Human versus machine competition for scoring education, and tailored learning content offerings. (collaborates with Stanford Uni). Also recognized by Unesco. (sidenote: it is clearly oriented at 'measurable, class and curriculum related content testing).
6. An algorithm is a set of rules to solve a problem. It can be any set of rules that is mathematically viable and programmable. It can also be used to manipulate data, so that a specific result is achieved.
7. AI is the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages. Based on algorithms. Machine learning focuses on the development of computer programs that can access data and use it learn for themselves. The process of learning begins with observations or data, such as examples, direct experience, or instruction, in order to look for patterns in data and make better decisions in the future based on the examples that we provide. The primary aim is to allow the computers learn automatically without human intervention or assistance and adjust actions accordingly. (e.g. Google draw) Deep learning: based on pattern recognition in big data
8. Algorithms enter our homes, work, schools, institutes, habits… but in most cases they are invisible.
9. AI risks to replicate the norm (filter bubbles prove it).Explained in part by the similar profiles of the creators of these algorithms. Easy to proof: in order to use online learning, English is used most frequently, simplified English with English jargon added. Written texts are the most frequently used form in content delivery and certainly in assessments. This leaves out non-written language based societies, and learn-by-mimicing behaviour. Do we ethically support the use of English, written assessments as proof of learning? A product is always a mirror of its creator. See community bounding: https://medium.com/@joaomilho/fixing-the-filter-bubble-e360a2c9bfdc
10. AI risks to replicate the norm (filter bubbles prove it).Explained in part by the similar profiles of the creators of these algorithms. Easy to proof: in order to use online learning, English is used most frequently, simplified English with English jargon added. Written texts are the most frequently used form in content delivery and certainly in assessments. This leaves out non-written language based societies, and learn-by-mimicing behaviour. Do we ethically support the use of English, written assessments as proof of learning? A product is always a mirror of its creator. See community bounding: https://medium.com/@joaomilho/fixing-the-filter-bubble-e360a2c9bfdc
11. Cognii builds upon the classic learning platform, but enhances it with AI for tutoring and assessments Cognii : Virtual learning assistant (chatbot integrated)
12. magpie offers: intelligent learning recommendations
13. Goodmorning ladies and gentlemen, colleagues, The last couple of days, we were able to listen to the challenges that we are all facing. And one of those challenges is How to build a future-proof wind workforce. This is not a simple task, as we know that wind energy is a rapidly evolving field. I will share the skills3.0 project, which might provide a solution to some of the challenges to build a future-proof workforce. But first let me start with a personal story. In my life I have switched careers RADICALLY four times. … and I am taking into account the one day I worked as a bullfighter to earn money… no, I am talking about Four longlasting career changes. The fourth time I did this, I really thought about it. Why did it work? This is what I do, I look for emerging industries, as these are still developing. Then I looked for skills that were needed in these emerging fields, and then I started learning feroucisouly, with all my heart…. And in the end I got a new career… So how do we build a future-proof workforce?
14. Value in purely providing courses, but to be competetitive we need to offer something more specific.
15. First of all we need to understand the emerging industry needs. We must find a way to predict where the industry is moving towards. How do we do this first step? By screening industry reports, road maps, white papers etcetera for emerging skills and competencies. We screen these reports by using an Artificial Intelligence or AI engine, that allows us to see which concepts are emerging, and which evolutions can be expected. If we know the evolutions of the wind industry, we can predict which profiles or which type of workforce we will be needing. This is where step 2 of the cycle comes in. As soon as we understand the upcoming competencies and skills that will be needed, we need to pinpoint skills gaps that exist between the industry’s needs, and the competencies and skills of the workforce that is currently available. Which brings me to step 3 in the process: analysing the profiles that are available and ranking them in terms of usefulness. How do we do that? By using an AI engine to screen resumes from the workforce (both blue and white collars). We look at their current employability, after that we triangulate which of these profiles would be the best fit after training. This is more than just looking at resumes, this is looking at skills taking into account available courses, and then predicting who is the best fit. Let me recap the first three steps: Step one: we have situated the industry needs Step two: we have pinpointed the skills gaps that need to be addressed Step three: we found those profiles that will be able to fit future skills needs, given they will get specific training … which leads me to step four: building a personalised training trajectory that addresses the skills gap. InnoEnergy has partnerships with content producers, universities and corporations that are experts in the field, but if the field is rapidly changing, we must be able to build some specific content on the go. That is why we are testing an AI engine that produces interactive learning modules from reports. I will show you links to a recent example of how we turned the WindEurope report into 5 learning modules, with only 15 minutes of wo/manpower invested in it. This automation of the learning cycle allows InnoEnergy to roll out blended courses that have experts coaching the learners as guides on the side. With these four steps InnoEnergy believes we can build an upskilled and reskilled future-proof workforce.
16. (Inge add link to wildfire learning) These are some of the examples coming out of our Artificial Intelligence engines. The employability check mixes profiles of workers to see which one’s come out on top after having followed specific training. The modules listed under the learning experiences, are the modules that were created from the WindEurope report that was launched recently. Now this approach is only one option we could use to build a future-proof workforce. But this afternoon we will sit together, bring all of our intelligence to the table to see what all of us can come up with as a roadmap for building a future-proof workforce. Thank you for your attention.
17. Multiple choice to measure student progress Teaching machine, designed by B. F. Skinner CC BY 3.0 File:Skinner teaching machine 01.jpg Created: 31 March 2008
18. Making reading and writing more accessible, including for those with a learning challenge