You can enable subtitles (captions) in the video player
Could be someone spreading misinformation, they're spreading across Twitter and the rest of the web. And that means that we need to escalate it as a crisis.
Might involve product recall. It might be getting the CEO out of bed to make an apology. It might be getting a chief marketing officer out of bed to make a decision at whatever time of day - three o'clock in the morning on a Sunday.
It seems an impossible but vital task. How does a business find out what people are saying about it online and where? Crisp Thinking believes it has found a way. The company from Leeds in the UK is paid by multinationals to scan social media for harmful comments so they can stop crises before they get out of control.
The artificial intelligence - it does the heavy lifting. Then we apply human intelligence on top of it, which means that we don't create false alarms.
We're gonna run through a crisis simulation of a real world event that happened for Unilever and one of their marketing campaigns some Hindu nationals took offence to. We've detected an anomaly ingested into the platform. I'm looking to see if it's anything we need to be concerned about. There's a hashtag included in this saying, boycott Hindustan Unilever. Well, here we are tracking our topic reach to see how many people have seen that particular hashtag.
But the challenge is huge. Every second, the world produces almost 9,000 tweets, views 80,000 videos on YouTube, and uploads almost 1,000 photos to Instagram, while 3.2 billion Facebook users interact.
Platforms like Facebook employ similar techniques to what we do. But they're going after things that are essentially breaking their terms of service across their entire platform, whereas we're focused on what would be harmful to any particular brand.
So the initial post by the brand was when the crisis started. And they noticed that the crisis was happening. They deleted their tweet and replaced it with something else. The actions of the brand actually amplified the crisis in this particular instance.
The crisis started on Twitter. It's now moved to Facebook. Can see here, within the space of 15 minutes, top two news articles, Twitter influences, and various other data sources have picked this up. And you can see it's reached ten to 15 million people within a very short space of time.
Social media can mean that something can be spread about you whether it's untrue or false within minutes. And you can go across the globe to millions of people. So to a brand, that can have significant damage.
But the social platforms themselves are struggling to detect explicit violent content, false accounts, and fake news. With no checks before users post, Crisp cannot prevent a viral infection. The best it can do is provide a diagnosis in time to try to reduce its spread and the harm it causes.
Someone criticising a brand isn't something we're interested in detecting. That's not what we deem as harmful content. Spotting needles in a haystack is a really good way of understanding that. And it's a hard thing to go and do - describe ourselves as not the actual firefighters solving the problem. But the fire alarm letting people know that there is a problem.