August 11, 2022

I, Science

The science magazine of Imperial College

By Vaishnavi Mohan
25th February 2022

For anyone who is obsessed with TV shows and movies, we are familiar with Netflix, one of the biggest streaming services in the world. Netflix is fundamentally changing the way the movie and film industry functions in many countries.

I am sure that there will have been countless times when you or I have browsed through Netflix to find the next movie to watch or tv show to binge. According to Business Insider, this decision is made in around 60 seconds! That is an incredibly short amount of time for one picture (along with its title) to convince the viewer to invest time in a programme…

Netflix relies heavily on using the most advanced artificial intelligence (AI) and applied data science to fine-tune the image which each viewer sees on the thumbnail. The thumbnails which I may view for a specific movie might be drastically different from what you see. This is because Netflix is aware that if we spend a long time sifting through titles which we would never watch, it would demotivate us from using the platform at all!

Indeed, the AI that Netflix uses is so sophisticated that it doesn’t only track basic analytical data, such as the number of movies watched or which movies are your favourite, but also which aspects of each movie you enjoyed, in order to give you a much more tailored selection to choose from. 

Here  are some examples of the data that Netflix collects:

  • Ratings
  • Searches
  • The date the movie/tv show was watched
  • The device on which a programme was watched
  • When was the program paused?
  • Do portions of a program get re-watched?
  • Do the credits get skipped?

The answers to questions such as these help create a profile for each user, and each profile is unique because of the number of variants involved. Thus, the algorithm needs to be equally sophisticated enough to provide accurate recommendations for each user. 

So, next time you switch on Netflix and sift through the reams of thumbnails, think carefully about the unseen AI busily whirring away and collecting your data. Do the tailored recommendations make it a better viewing experience for you? Is Netflix justified in collecting this amount of data to make our viewing experience less of a hassle or is this subtle manipulation an exploitation of our attention? 


Vaishnavi Mohan is a sub-editor for I,Science, and is currently completing her Masters in Science Communication here at Imperial College London following her undergraduate in the History and Philosophy of Science at UCL. She is interested in the history of medicine as well as how science and culture intersect.