Role of Sentiment Analysis and AI in Social Media
Sentiment analysis plays an incredibly important role in social media monitoring. It is better known as opinion mining and that’s exactly what it does. This analysis yields deeper insights into popular opinion regarding specific topics. Most companies utilize it to assess the feedback they get from customers but it’s not just corporations that gauge public opinion. In fact, even government administrations use sentiment analysis to measure extreme reactions to controversial policy announcements and respond accordingly. The issue is that this application is not a perfect science. Not at all. Just think about it. The human language is more layered than we give it credit for. The meanings of certain sentences can change on a dime given the context behind them. That being said, human beings possess enough self-awareness and intuition to interpret the tone of a written piece. Machines on the other hand? Well, that’s a different story altogether.
For better or worse, artificial intelligence is a prominent part of our lives. AI has made remarkable breakthroughs in linguistics, psychology, mathematics and neuroscience that would have otherwise been impossible. No one can deny that. And I’m not someone who is fearful of it but I definitely have my doubts. These doubts range from practical concerns to more sinister problems. For starters, machines will not always understand the tone of a sentence. The grammatical nuances tend to fly over their heads. What you and I perceive as sarcasm might not register as such with machines. Contextual understanding is the key here. Think back to the various times you have lodged a customer complaint online. I know I have used the word “brilliant” more than once sarcastically but a machine will take that at face value. The widespread growth of social media makes this dilemma worse. A platform like Twitter limits one’s expression to 140 characters. Hundreds of abbreviations and slang have been born from the necessity to get more across by saying less. If you think you can barely keep up with the new digital shorthand, how will the machines comprehend it?
This brings me to the more disturbing possibility. Sentiment analysis is clearly a powerful tool that helps quantify and study subjective information. Do we really want AI to be heavily involved in this area? Teaching machines to emotionally evaluate data can be dangerous. Moral reasoning is what separates us from them. This tool will only further the hypothetical threat that gradual progress in AI will inevitably lead to some sort of global catastrophe. After all, the reason humans are at the top of the food chain is that our brain is more advanced than that of any living animal. If AI were to match or even surpass our distinct capabilities, it would be difficult to control. Driverless cars will soon become the mainstream norm in a couple of years. That’s going to be interesting to watch. We are basically putting our lives directly into the hands of these devices. Yes, you could argue that we already do but at least our current automobiles are still subservient to our physical handling. This is just my roundabout way of saying that we need to consider the things we teach AI. The machines have become more capable than we could have ever imagined. And that isn’t going to stop anytime soon.
The latest debate raging on the internet is about ‘artificial intelligence’ vs ‘natural stupidity’! Stay tuned.