Writing is largely considered a craft, a manual task that marries both the cerebral need to think about each and every word that sits on the page (or screen) whilst also manually placing the letters onto the paper with a hand tool, whether pen or keyboard. Portrait galleries around the world throng with images of scribes struggling to make their art near to perfect. Writing, though, is far from immune to the move towards greater automation.
Artificial intelligence (AI) writing tools are becoming more prevalent. With CEOs demanding greater automation, digitised business processes and seeking market growth, it is perhaps inevitable that just as the quill made way for the pen and then the keyboard, AI-powered content creation will become essential to write the sheer volume of material a modern enterprise requires.
Demand for the written word is increasing. “Sectors such as IT, professional and financial services need long-form content that is optimised,” says Adam Benzecrit, co-founder of Inflo AI, an automated writing assistance technology. “If you’re a scale-up with a small marketing team that is stretched, then you are struggling to deliver enough content to all the channels. They can outsource to an agency, but those agencies don’t have domain expertise.”
This rising tide of information is not only sales and marketing material - although a glance at the email inbox or social media feed suggests otherwise. Internal communications are on the rise too. In the UK, the NHS Digital First strategy states that £413 million in savings can be made from making appointment communications digital. Northern Lincolnshire and Goole NHS Foundation Trust were able to deliver an extra 900 appointments to patients due to a move to digital communications in 2021. CIO for the Lincolnshire NHS Trust Shauna McMahon says that not attended appointments have a major impact on the efficiency of a hospital.
Enter AI
Digital writing tools are far from new; spell checkers began the trend. Robert Dale, a noted academic on Natural Language Processing (NLP) at Macquarie University in Australia, wrote: “the category has been given a substantial boost from recent advances in deep learning.” Dale says the development of the OpenAI GPT-3 language model, which uses deep learning to create almost human text, has led to: “plausible and convincing texts on any topic”. He adds, this “opens the door to a new set of capabilities, where responsibility for the content of a text is shared between human and machine.”
“Our automation technology aims to help people through the writing process, and that will save a lot of time,” Benzecrit, co-founder of Inflo AI says. As the London headquartered company name suggests, the platform used AI and the OpenAI GPT-3 language model. “Users don’t have to start from scratch. The user is just polishing something that is 70% complete. The writing and content process is a burden; there are so many factors to be considered ranging from search engine optimisation to the needs of the reader,” he says. Inflo AI learns the tone of voice used by the marketing team, scans for plagiarism and SEO strength and recommends sites and content the user should use for inspiration.
“Artificial Intelligence (AI) is going to fundamentally change how people live their lives and communicators are going to have to navigate major societal change while we work out what it means for our profession,” Kirk Millis-Ward, Director of Communications and Engagement, Isle of Wight NHS Trust told Askem, a provider of web feedback services in the NHS Comms report. “With COVID-19 resulting in a huge uptake in audiences consuming digital content, it makes sense that AI will become more and more commonplace in our comms,” adds King’s College Hospital NHS Trust Head of Communications Rebecca Kingdon-Kruszewski in the same report.
NLP academic Dale has doubts about the strengths of AI: “The outputs are often weak…especially so for longer texts, but occasionally you will be genuinely impressed, and even the weaker results may provide a source of inspiration if you have writer’s block.”
“Authenticity takes a level of creativity that is subjective,” adds Dr Stylianos Kampakis, a data science advisor and author at the London Business School.
“Anything and everything that includes AI technology needs to have disclosure about how it was trained,” says Chris Leong, AI advisor to Change Gap and ForHumanity, the nonprofit organisation looking to address the ethics, bias, privacy, trust, and cybersecurity in artificial intelligence and autonomous systems. “With disclosure, people can be informed as to whether they want to be involved.” Understandably, customers will have concerns about AI shaping the material they read from the organisations they do business with, and Leong warns CIOs and CMOs of unintended consequences from automation.
More content less tools
Marketing technology blogger Scott Brinker annually publishes an online graphic of the number of software applications CMOs can utilise. Despite having looked at it several times over the years, it still takes some believing; there is a mass of vendors out there for interactive content, video marketing, mobile apps, native content, display advertising, and the list goes on. Discussions with CMOs and CIOs reveal that many are looking to rationalise their technology estates to keep budgets in check. “There are keyword research tools, editors, content management systems (CMS) and optimisation tools,” Benzecrit says of creating a tool that allows comms teams to collapse tasks into a single platform.
Of course, the elephant in the room is that it is one thing to create more content and have tools that make the creation process easier. But customers, especially enterprise technology customers, demand authenticity, accuracy and they are time poor, so they want content that delivers insight and value.
The Edelman-LinkedIn B2B Thought Leadership Impact Study for 2021 says: “This year’s study reveals a tidal wave of content marketing that is making it harder for companies to connect with B2B customers and prospects.” Adding: “A pandemic-induced glut of low-quality content is diluting the perceived value of thought leadership among B2B decision-makers.” If there is a glut of low-quality material out there, then AI faces the threat of learning from poor quality data, another case of rubbish in, rubbish out, something that CIOs and CTOs could do with less of. On this point, Kampakis advises organisations to make sure that the content models they train their AI on are “very strong”.
However, the Edelman-LinkedIn report finds that 54% of business leaders spend more than one hour per week reading and reviewing thought-leadership content, a similar statistic to 2019, suggesting that there’s still plenty of value in the practice.
Technology will always allow more material to be produced, but the real challenge for AI in comms creation is not only the content; CMOs and business technology leaders thirst for the right audience and to secure the right audience, content must speak to a community. To be a member of a community requires authenticity. Will the AI content creation story lead to an increase in intelligence or yet more artificiality?
