About free online education on the example of the Stanford AI Class
Not long ago, Stanford University organized free online courses on artificial intelligence. It would seem, here it is, happiness - go booth! Competently conducted marketing and tried to make everything as friendly as possible: the videos themselves were on YouTube, the official discussion group was on Reddit, the announcements and news of the course on Twitter promised to even integrate an account with Facebook ... And now a record large audience reached - 160 thousand listeners ( proof ). This is a victory, giving higher education to the masses!
Only here something strange happens with the attendance of this course. Due to the fact that video lectures are launched from YouTube to the embedded object, you can go to the YouTube page of the KnowIt company, which deals with technical support of the course, and look at the popularity of videos (number of views). For those who are lazy to go to YouTube:
And then the number of visits slowly, but steadily falls - from 5 to 2 thousand views. The minimum number of views of the examination videos at the time of writing the post - 878 (reception of answers is already closed). So, out of 160 thousand enrollees, the exam has been tried to pass a little more than 0.5%.
PS Video views through the embedded object are fixed by Google correctly, I checked it on one of my video with a near-zero number of views (which makes sure that it was my view that was recorded, and not someone else accidentally decided to watch it too).
PPS In this case, it remains incomprehensible to me why the organizers themselves said on Sunday about 80 thousand active students ( proof ). There are three options: either the teachers have distorted the facts, or the statistics of Google’s views — not to hell, or I’m getting pretty big somewhere. In the latter case, please indicate the error.
PPS Another option: the majority of students read a textbook, do not watch videos, and answer questions directly without even viewing a video question.
')
TODO: There should be a multi-layered conclusion about the prospects of higher education on the Internet, but I don’t have it yet. The point is not in the percentage of registered and active users, but in the total number of world-famous scientists who are really interested in the free course. In the absence of basic training, everything will not succeed: the course website has links to the necessary Khan Academy videos, which provide all the material necessary for understanding AI Class, starting from the school level.
Update Fixed a typo: not 989, but 878 views. Accordingly, this is 0.54% of the number of those who signed up.
Update 2 According to quite numerous testimonies ( example ), YouTube’s counting system is working, if not at all, then at least unclearly: in some cases the viewing is fixed, in others it is not, in the third it is recorded with a great delay. However, the above calculations are consistent with the number of subscribers of the official Reddit discussion group (<4 thousand).
Update 3 Habrayuzer radiant prompted that Norvig published official statistics on students. Who handed over the homework, according to official data - 46 thousand. It’s still premature to draw conclusions about the fitness of the current YouTube video counting algorithm for operational analysis (and generally about the correspondence of the number of views to reality), but the situation is puzzling.
Only here something strange happens with the attendance of this course. Due to the fact that video lectures are launched from YouTube to the embedded object, you can go to the YouTube page of the KnowIt company, which deals with technical support of the course, and look at the popularity of videos (number of views). For those who are lazy to go to YouTube:
- "Welcome" looked 50 thousand people
- Introduction and "What is an intelligent agent" - 13 thousand each
- Terminology - 8 thousand
And then the number of visits slowly, but steadily falls - from 5 to 2 thousand views. The minimum number of views of the examination videos at the time of writing the post - 878 (reception of answers is already closed). So, out of 160 thousand enrollees, the exam has been tried to pass a little more than 0.5%.
PS Video views through the embedded object are fixed by Google correctly, I checked it on one of my video with a near-zero number of views (which makes sure that it was my view that was recorded, and not someone else accidentally decided to watch it too).
PPS In this case, it remains incomprehensible to me why the organizers themselves said on Sunday about 80 thousand active students ( proof ). There are three options: either the teachers have distorted the facts, or the statistics of Google’s views — not to hell, or I’m getting pretty big somewhere. In the latter case, please indicate the error.
PPS Another option: the majority of students read a textbook, do not watch videos, and answer questions directly without even viewing a video question.
')
TODO: There should be a multi-layered conclusion about the prospects of higher education on the Internet, but I don’t have it yet. The point is not in the percentage of registered and active users, but in the total number of world-famous scientists who are really interested in the free course. In the absence of basic training, everything will not succeed: the course website has links to the necessary Khan Academy videos, which provide all the material necessary for understanding AI Class, starting from the school level.
Update Fixed a typo: not 989, but 878 views. Accordingly, this is 0.54% of the number of those who signed up.
Update 2 According to quite numerous testimonies ( example ), YouTube’s counting system is working, if not at all, then at least unclearly: in some cases the viewing is fixed, in others it is not, in the third it is recorded with a great delay. However, the above calculations are consistent with the number of subscribers of the official Reddit discussion group (<4 thousand).
Update 3 Habrayuzer radiant prompted that Norvig published official statistics on students. Who handed over the homework, according to official data - 46 thousand. It’s still premature to draw conclusions about the fitness of the current YouTube video counting algorithm for operational analysis (and generally about the correspondence of the number of views to reality), but the situation is puzzling.
Source: https://habr.com/ru/post/130672/
All Articles