The Prediction Project

The Past and Present of the Future

LogoIcon3_edited_edited_edited.png
  • Facebook
  • Twitter
  • YouTube
  • Instagram
  • Spotify
  • HOME

  • ABOUT

  • MATERIALS

    • Prediction Essentials
    • Omens & Oracles
    • Rise of Theory
    • Modern Prediction
    • More to Explore
  • COURSES

  • TALKS

  • WRITINGS

    • Uncertainty about Uncertainty
    • Data Driven Dilemmas
    • Uncertain Risks
  • PRESS

  • FORUM

  • Introduction to Uncertainty

  •  

    Use tab to navigate through the menu items.
    To see this working, head to your live site.
    • Categories
    • All Posts
    • My Posts
    Sravya Kuchibhotla
    Dash  ·  
    Mar 29, 2021
      ·  Edited: Mar 31, 2021

    Google Flu Trends

    in Health

    Ben Schneiderman


    The conversation about Google Flu Trends was extremely interesting to me. I never knew that Google tried to predict flu hotspots using search strings and the sale of commodities back in 2009. It appeared very revolutionary in the beginning because being able to predict future flu hotspots is a great way to reallocate hospital supplies before an area is hit with the disease (as we saw with the allocation of ventilators and other medical supplies during the SARS-COV-2 pandemic). After initial success, the predictions of Google Flu Trends actually resulted in worse allocation because the predictive model was not as accurate as expected. By 2013, Google shelved the program because it became a sort of embarrassment for the company.


    I think it’s very easy to maintain that the future is in Artificial Intelligence or that robots will be able to predict and control our lives in the future. How can we not? So many Sci-Fi Dystopias paint this picture. However, I believe that we are truly a long way off. Whether it was predicting Flu hotspots in 2009 or COVID-19 in 2020, the trends have been very similar. Having the foresight to predict hotspots was virtually impossible; having the hindsight once an area is affected to delineate the contributing factors not so much. I don’t think that has changed much between 2009, 2020, or (I would argue) 2040.


    1 comment
    Julie Alarifi
    Dash  ·  
    Apr 06, 2021

    I found this point extremely interesting as well! It also adds to a previous thought I had about the implications that predictive systems can have on our future decisions that also directly impact the future and create a sort of paradox. I feel as though this is somewhat an example of that because based on the predictions that google flu trends made that might have affected where they decided to allocate resources which might have inadvertently been the reason their predictive system failed although we can never truly be sure. But I completely agree that we are a long way off from being able to predict hotspots and also being able to predict how choices we make might further change those results.

    1 comments

    content brought to you by

    Font-Harvard-Logo.png

    follow us

    watch us

    listen to us

    imageedit_8_5179044945.png