I want to believe - Fake News and Big Data
Stories about fake news continue to make the headlines. Facebook is currently struggling to avoid a barrage of criticism and questions over whether it is actually a media owner, and therefore should be regulated as a media company. As I read these stories I was struck by the relationship between the controversies over fake news in the consumer world and debates around big data in the business world.
In the age of big data, we look to algorithms and artificial intelligence (AI) to supplement human abilities. According to Stephen Hawking, amongst others, AI may actually supplant humans altogether.
With this incredible processing power, we should live in a time where we can understand exactly what is happening in the world around us. Yet we also live in an age where someone can make up a data set, let’s say “pigs can fly”, put this into digital space, and find others repeating and seemingly validating this ‘data’ simply because it is out there.
Circulating inaccurate data.
At school (a long time ago), I spent weeks repeating a fake story about a fictional band (with the implausible name Wickedly Victorian) to trick class-mates. After made up albums and supposed live gigs some of the class started to tell us how much they liked WV and how brilliant the new LP was. Whilst a silly prank, we’ve seen in recent months potentially far more serious consequences when inaccurate data is spread, most recently on the eve of the French Presidential election when real emails were leaked mixed with false documents.
The input problem
What does fake news have to do with big data? I think we’re in danger of creating AI-generated fake news in our organisations. Poor quality data can only yield poor quality outputs: garbage in, garbage out. We are all familiar with the mantra, but increasingly, organisations are looking to technology to solve a problem at the output point without sufficiently addressing the input quality.
This problem is felt particularly in sales and service intelligence. Companies are investing increasing amounts in CRM systems to try to improve productivity – a number estimated at $37bn globally this year alone and predicted to rise to $51bn by 2020.
Reports sell software, we are told, yet often the systems that produce the reports rely on human agents with little motivation to enter timely, accurate and quality data. How do we ensure that we can capture the best of human observation and insights into our enterprise workflows? Because without high quality inputs our AI may simply be giving us more convincing fake news.
Next story: Wickedly Victorian reform after 40 years!