DONALD TRUMP HAS BEEN IMPEEACHED!
Look at you getting all excited (or sad)! Of course it’s fake news. Did I do a good job at jolting your emotions in that one line even just for a few seconds? That’s the world we are increasingly going to live in – a world where news articles and reports are going to come from many different sources, some of them unverified with malicious intent. How then do we know what’s true and what isn’t? Very clever people in this world are developing A.I. systems that aimed at helping us sift through the noise and hopefully get to the truth. That is what I read about this week.
First, let me unashamedly promote a post I published here on LinkedIn discussing one fuel for fake news: Deep Fakes. If you are going to read one article in this post, it better be this one. Not because it is my article (well, partially) but because Deep Fakes are capable of ruining your life in a heartbeat! What is worse is they are becoming increasingly easy to create. I aptly titled it, “Deep Fakes. Be afraid. Be Very Afraid.” Let me not spoil it. The article explains what they are and gives you examples that will hopefully get you thinking about how it’s going to become harder and harder to know what the truth is: https://www.linkedin.com/pulse/deep-fakes-afraid-very-houston-muzamhindo/
The world trusts video content without much questioning but with Deep Fakes, it is time for us to reconsider the pedestal we have put videos on. As they are right now, it is quite easy to tell that something is off with the video but as improvements are made to the algorithms, it’s going to be easy to fool the human eye. Facebook has been facing this uphill battle of detecting fake news in photos and videos. Enter A.I. They are training “a machine-learning model to detect potentially bogus photos or videos.”: https://www.zdnet.com/article/facebooks-fact-checkers-train-ai-to-detect-deep-fake-videos/
Fake news is not only in the form of photos and videos developed by some Generative Addition Network or GAN. I explain what this algorithm is on my blog with an example where I used it to generate handwritten digits (can you use it to generate a fake but real-looking signature?). Fake news can be in the form of a blog post that goes viral on Facebook as we saw during the US presidential election in 2016 and the news from 2014 about Ebola in Texas. Some people are thus working to detect fake news that is in the form of text. They are deploying various A.I. algorithms that look at different metrics such as the score of the website, its reputation etc: https://www.marutitech.com/artificial-intelligence-fake-news/
Detecting fake news is extremely difficult. It takes sophisticated algorithms, some of them requiring cutting-edge research in fields such as Natural Language Processing. I can write a news article using terms that are understood locally but not universally. State of the art algorithms might stumble on such texts if they are not trained using local slang for example. Our only hope is the clever people working on combating this very real threat of fake news can quickly develop ways to do so before we get excited (or sad) at news articles with titles like “Donald Trump has been impeached…” before it actually happens.
It’s a brave new world!