Artificial intelligence (AI) is currently one of the hottest topics in the communications industry, but it’s also been met with a fair degree of uncertainty. The very mention of AI sparks varying sentiments of fear, excitement, scepticism and curiosity in PR professionals.
"It’s quite easy to get carried away with ourselves thinking it’s going to be this revolutionary thing that blurs the lines between human and machine," Orla Graham, senior insights manager at Cision, told the PR360 audience. "Or it’s only a matter of time before they take over the world and take all our jobs. But in reality we’re using them more like our butlers. They’re enhancing our lives, by doing a lot of those monotonous tasks we don’t want to do ourselves."
"When it comes to PR and communications what we’re seeing in terms of AI is how it can help you get information. We’ve come a long way from using scissors and Pritt Stick to put print articles on a piece of paper before sending it off to a client. Machine learning is allowing us to do things better, faster and cheaper," said Graham.
How can machines help?
"Machine learning helps bring in more relevant and accurate content. The technology is getting better at assigning topics and trends, so you can see what is resonating with people and picking up traction, so you can adjust your strategies accordingly.
"At Cision, we’re partnering with ad tech companies to get access to lots of information from a huge variety of different websites so we can understand how people are engaging with editorial content. How long are people spending on that page? Are they actually reading it or moving on? What are their behaviours and activities like? And who are they?" said Graham.
Sculpting your strategy
"We had a client who is a major tech company and they wanted to understand about tech trade show coverage and how they could reach out to more women – an audience who had typically been quite underserved by their coverage.
"They wanted to learn which media types are resonating more with female readers, which other competitive events they should be benchmarking themselves against and which topics they should be focusing on.
"We were able to provide insight into these areas using automated technological innovation because we are able to tell you where your audience is, what’s of interest to them and how you can start to sculpt your strategy to actually speak to them," said Graham.
Natural language processing
Natural language processing is most simply defined as the ability of a computer program to understand and analyse human language as it is spoken. The capabilities of natural language processing cover a broad scope of use cases, but the ability to identify audience sentiment automatically is key for communicators when it comes to planning and executing impactful earned media programs.
"We’re seeing a lot of innovations and developments around some of those typically human types of metrics that are often seen as being too nuanced to automate in any kind of way," said Graham.
Sifting through coverage to identify stories and company mentions that had the desired impact on target audiences would take a lot of time and resources. Traditional sentiment analysis would not simply involve locating keywords, it would require a person to read through an article at length to interpret its tone.
"Today’s automated systems are equipped to interpret context and attribute true meaning, growing smarter over time. With a natural language processing engine built into your monitoring solution, you can quickly and easily determine sentiment from a simple string of text across any relevant channel," said Graham.
Big data & smart data
"When it comes to this ‘art vs science’ of communications and data measurement, you do have this dichotomy of needing to get the data in a quick and efficient way but needing to make sense of it. Humans are still the only things capable of doing that and then coming up with a strategy. The data’s all well and good and it’s really helpful when it’s accurate, relevant and it comes to you quickly. But it’s not going to figure out your strategy or make sense of the data – humans are," said Graham.
Cision have a number of clients who are now using a combination of big data and smart data. Big data describes massive amounts of data, both unstructured and structured, that is collected by organisations on a daily basis. This big data can then be filtered and turned into smart data before being analysed for insights.
"For instance, an automotive client wanted to understand what was going on with automotive trade shows and whether they were really worth their while. So they commissioned us to do some research into the Mobile World Congress – not typically an automotive trade show.
"So we looked at these bigger conversations to see what topics were resonating in regards to which brands. Then we dived into more detail into the human coding side of things to be able to help them understand which conversations were ‘really’ resonating. How the brand was coming across, were people thinking it was a good idea that they were appearing at these types of shows. And they were able to make a strategy about which shows to attend on the back of that.
"Technology might be making us better, faster and stronger in a lot of ways, but it can’t replace us. You need both, you can’t use technology on its own because you’re never going to get that nuance of human emotion and connection.
"AI is looking after the tedious, repetitive tasks leaving humans free to make sense of the data," said Graham. "But we also need human intervention because what we’re trying to do as communicators is engage with humans – it’s at the core of everything that we do."