Automation is coming. What used to be far off science fiction is fast becoming science fact. The future is AI, so how do you make it work for you?
AI and CMS fit together seamlessly in a way that might surprise. In the past, it appeared that “creative” gigs such as content creation, design, and other tasks of this nature were under no threat from AI. We can all imagine a self-driving car and perhaps even A/B testing done by a bot. But what about authoring full-length articles indistinguishable from human work written by a computer program? This concept might seem far off but it’s already being implemented by companies across the world.
In a previous Gartner study, it was reported that 20% of business content would be authored by machines by 2018. Well, here we are, and programs like Quill are already molding the CMS of the future. Quill creator, Narrative Science, already rents out the program to financial institutions to create 10-15 page financial reports. It is a compelling proposition, considering that it takes mere minutes for the program to create what a human writer might take weeks to produce. According to MIT, Quill is producing millions of words per day; words that no longer need a human touch to be produced.
It’s startling to think of the changes this will make for lower level content production in the short to medium term. We aren’t yet at a point where AI could be expected to create long form, ultra-creative work – we still need humans for that – but when you read that the Washington Post published 850 entirely robot generated articles in the last year, you can see where the future of the business is going.
The Washington Post used a program called Heliograph to produce 500 articles around the US election that generated more than 500,000 click throughs. That isn’t a lot when you weigh it up against the rest of their output, but when you consider these articles were in an area that they weren’t going to dedicate staff resources to, it’s clear how the numbers stack up. That’s 500,000 clicks generated by one source that didn’t need a weekly wage.
At the moment, AI helps add value to human writers by taking a lot of the dirty work off their plate. Take financial reporting, for example. Why would you pay someone to dig through realms of numbers when AI can do it quicker, easier, and error-free?
“In the case of automated financial news coverage by AP, the error rate in the copy decreased even as the volume of the output increased more than tenfold.” - Associated Press strategy manager and co-AI lead Francesco Marconi
From a CMS perspective, why wouldn’t you want to bring AI in to take over some of your content creation? It will eliminate mistakes, dramatically increase production, and it’s a solid solution for a large scale need.
AI can be helpful in a number of CMS use cases.
- Suggest the content editor to use a different style based on sentiment analysis, for example.
- Filter or suggest to remove words/phrases which may not be suitable for the individual or company to publish.
- Enable the CMS to dig deeper into data for more advanced customer insights.
- Help to publish the approval workflow. In theory, an article could be automatically approved by AI without a human reading it over.
But when will we reach a tipping point at which companies compete with each other on cost benefits entirely facilitated by AI advantage? This is one of the strongest advantage of AI, along with time-to-results. In a media world where costs are increasing, ad revenues are decreasing, and realtime information is king, how long before AI becomes a standard tool in the content creation process? A program in development called Stats Monkey is already producing passable sports copy based entirely on data point entry.
NLP and NLG: How this Actually Works
NLP stands for Neuro-Linguistic Programming. It’s basically about learning the language of the mind. In AI, NLP uses computational techniques to analyze and combine natural language and speech.
There are three types of NLP:
- Inquiry: uses text analytic tools.
- Conversational: engages the person in conversation to clarify information and refine inquiries.
- Reasoning: this goes beyond the basic understanding of AI and deals with more abstract concepts such as beliefs and emotions.
AI can generate bucketloads of content 24/7, no caffeine required. It monitors how that content is consumed, who consumes it, where it’s consumed, and how readers respond. In addition, it can also learn from this to become even better at creating content that feels personalized and targeted to the needs of the reader, i.e. it can become more human.
Natural language generation (NLG) is a subset of the field of NLP and focuses on generating content that mimics the language a human would create – except it’s produced entirely by a machine. It typically operates using a repository of content, or knowledge base, and translates the underlying data into natural language that you and I can make sense of. NLG has existed for a long time but has made tremendous leaps in recent years. Turning financial data from a set of databases into plain English (or German, Farsi, Spanish etc.) is a quintessential use case example.
Analyzing Consumer Behavior
The area of consumer behavior analysis is a huge one for AI, because – quite simply – robots do it better. AI can monitor behavior in realtime at a speed and standard that’s beyond the ability of humans.
A/B testing can be automated and the results of this testing can be implemented straightaway to improve UX based on the AI analysis. In terms of content, personas and advanced segmentation can be utilized to create content that presents varying text based on each visitor profile. This is one of the most profitable and valuable usages of AI at present.
AI even has a role in the future of social media. The role of social media is only going to increase when it comes to content promotion. Spending was up 60% for social media advertising in 2017 and companies are already using Natural Language Processing to increase the feeling of real interaction from bots and AI. As advertising spend increases on social media, companies will look for more bang for their social media manager buck.
Social media management can be quite complex but if your needs on Twitter, Facebook, and Instagram are basic (reply and tweet), AI can and will replace human input sooner rather than later. The biggest issue in removing this cost base is the ability of bots to properly process language and appropriate responses. The field of NLP is based on greater understanding of language from written inputs. And as it develops, there’s an obvious advantage to customer service – which is also handled on social media – by improving customer drop off rates. It’s estimated that by 2020, 85% of all customer interactions will be handled without a human agent being required. But the further NLP progresses, the more we could see AI used in actual social media interaction.
But Don’t Fire Anyone Yet!
AI will change the way CMS is managed from the ground up. It will change how it’s costed, it will change how it’s created, and how it’s eventually marketed. The ultimate destination isn’t to replace your copywriters and marketing staff. Instead, view AI as a tool that will evolve the role and change the daily tasks of your content creators.
NLP still needs that subtle touch of human influence when content is being delivered in any kind of creative aspect. There’s no point in generating 100s of pages of content that no-one will actually take notice of. Sure, there is a lot of content that’s suitable for AI takeover, but it’s best to leave the creative content to your skilled human workers. The job of your content writers is to connect with the reader, and that authentic human touch is something that AI is a long way from mastering.
What AI can do for your CMS, is make life a whole lot easier by streamlining processes that were previously tedious and allowing your content creators, marketers, and designers to work on what they do best.