ad-free
Channel / Source:
TEDx Talks
Title: Artificial intelligence & the future of education systems | Bernhard Schindlholzer | TEDxFHKufstein
Published: 2016-08-04
Source: https://www.youtube.com/watch?v=ZdHhs-I9FVo
if a mineralization will have a profound impact on our economy and if we want future economic growth we have to rethink our approach to education ever merrily say sin is a term coined by the U. S. visionary and philosopher Buckminster fuller who used it who you use the word to describe the tendency of technology to do more and more with less and less until eventually you
can do everything if nothing and what sounds like a visionary and philosophical statement it's actually something that we have all experienced if we look back in time and look at the first computers who were huge machines field rooms with limited storage and processing capacity and if we ban look into our pockets find the smart phone which dramatically more power more computing or processing power and storage
that is now able to integrate all these other functions see a concrete example of ephemera the station I also think we are reaching a new tipping point a new tipping point in the development of feuding power it will allow us to I'm look a completely new set of use cases and applications these use cases artist simulation of the brain through neural networks deep learning and ultimately
artificial intell now just to give you an example how big these changes have been in the last actually twenty four to forty eight months %HESITATION here's a chart it basically gives you some statistics how long it Terry takes train up the machine learning algorithm in this case Alex net which is used for image classification purposes the blue bar it's one of the traditional but high end
processors and it takes this processors forty three days in the green bars see the already launched and soon to be launched products from a company called nvidia who is able to approach process thirty sign in a completely new way and to deliver a twenty fold increase in training performance which means we are able to decrease training time from forty three days two days only that means
we can train computer us much much more much much more scope and give them a better understanding off what's happening in the world they're already have very concrete use cases where we see these changes in action just a few months back %HESITATION the computer Alfa go bet the Korean beat the Korean go World Championship but he said dole in four out of five games Verhoeven industry
insiders were surprised how quickly this happens we're also seeing the rise off intelligent assistance where you can ask your iPhone are you specifically Syria on your iPhone not just simple questions anymore but also more sophisticated questions to show you photos from your last vacation in uta in August Google has also announced a new intelligent assistance what did you cannot ask what a tool book is any
good and it will give you the answer without having you to search for information by yourself the ultimate goal or one of still timid goals of artificial intelligence the self driving car and why the self driving cars for many of us here in Europe still sounds like something out of a science fiction movie if you visit the Silicon Valley they go to Mountain View and then
sit down on a road that the Googleplex by rote wait for thirty minutes and you see this cars driving around you will realize Utah has already arrived William Gibson set it frees already here it's just not evenly distribute and I'm asking myself what happens then Dick knowledge she's evenly distribute it what happens to our economy when self driving cars to make it to Europe and intelligent
assistance become even more intelligent how many travel agents to still neat when your assistant can book a flight for you how many customer support agents do we needs when we can use machine learning to extract the ride on Sir from thousands of pages of help center Clinton's and how many employees do we still need in the back office of a bank or an insurance company that
are making decisions soon can also be potentially automated I think it's a very very the cult question but I think we have to answer it I also believe we have to be realistic and for me being realistic means we have to admit the demand for certain chops will Chris Pacific early the demand for routine knowledge work will be decreasing we will need less travel agents and
we will meet less employees working in a box backoffice because that will be automation there will be artificial intelligence that will perform decision but there is also good news there will be an increasing demands in another area future economic growth will come from non routine trade if knowledge work she's work it aims to define solutions solve problems in the world this is already done today by
scientists researchers physicists programmers and engineers the good thing is there is an unlimited number of problems in the world so they'll always be work or somebody solve real so if we accept premise I'm asking myself how can we be per and again we need to be realistic that nobody will hire you anymore what can who knows everything right and it will know even more it will
give you even more precise answer maybe not in two years maybe five years maybe ten years maybe it's not even going to be Google itself rights because maybe another startup will come around and we'll be able to give you deride on service at the right time so if knowledge itself becomes less important what we need to focus on in order to get hired is the ability
to apply the knowledge you will get hired because of what you're able to pretty straightforward to universities occasion to some degree follow this principle already but looking at universities all over the world still the most common view we have lectures sometimes great and sometimes enrich qualities standing in front of a class of students trying to transfer their knowledge today our minds following what I said before
that's moved a fitted cation is not really the best use of anybody's time why not take the best lecturers in the world figure out a way how we can make it more interactive learning experience they can take over a part of pure knowledge transfer and den we think about away and to set up and a new structure for universities where students get a chance apply their
knowledge and they are actually universities who are really driving %HESITATION pushing the envelope this area for example Stanford University that it is teaching a course that actually goes back to the sixties and seventies where industry partners students problems do you dance have to work on these problems come up with solutions what you see here on the screen is a typical classroom session where you see the
two professors actually three of them sitting there where can lift a student in the Hylian gaze manner over many many months to really understand the problems and he signed solution the outcome of discourse it's been quite impressive it's not just number of patents companies derive from the work of the students in collaboration with the university or the number of products companies develop or students develop in
the chorus it also acts a breeding ground for start ups students themselves ID eight in this course and the skills that they learn its help them afterwards launching their own companies to it's not just happening in the United States we have transferred to system also to the university of Saint Gallen where we teach a course on the sign thinking where students spent more in around ten
months between ten to fifteen hours a week in a highly engaged environments under supervision by professors internal and other external advisors to work on these problems come up with solutions and I believe that's the characteristics we seen these courses will be shaping future of education it will be more about problem based learning about immersion and about simulation how what do these three things mean problem based
learning means you dense are challenged to apply their knowledge to real world problems and they need to rephrase and rethink and re frame the problems to identify new solutions and maybe even realize that they have been trying to solve the wrong problem all along that's something that we typically don't to because most exams still try and force you to come up with the one ride ons
are into only one right hands shifting away from the smoke where transfer knowledge then we salmon whether you have successfully acquired does not will be a thing of the past it's also specific to this kind of teaching and education environment is immersed in it's a real time decision making over longer periods of time as I mentioned before these courses run over ten months so if a
student makes a decision turns out to have not been the optimal decision they actually have to deal with the consequence turn around the project figure out the new power forward but the last point that is important the aspect of simulation that at the end of today all this happens in a safe environment first you dance are free to experiment but again try out things they can
feed and they have a chance to try again I think it failed again and if they even at the end of the course ve continuously be failing course aims not to only charge on the outcomes but also on their style and their ability to apply knowledge and to try and solve and I believe that's our education system needs to go beyond pure knowledge transfer we need
to find smarter ways a silly take that and we need to change our universities into a real training grounds where students are have a chance to train under the supervision of professors solve real world problems in an immersive environment while experiencing assimilation of real life because only if we go to distant only if we go there then we don't have to be afraid the rise of
