How to create content in a world with bots

In 2014, I had a lively conversation with a colleague. He argued that in just a few years, robots would replace copywriters in content marketing. I rolled my eyes so hard that my brain hurt. The idea of ​​machines making sentences as elegant and nuanced as the human writer could do was both heretical and impossible, I thought.

It turns out to be right, but development is not the creative dystopia of science fiction that I imagined … in most cases. Today, companies are actually using machine learning to generate natural language and marketers are using AI-based tools such as Narrative Science and Automated Insights to produce practical, well-rounded content. built and very powerful.

Marketers use AI-based tools to produce practical, high-performance content, says @clare_mcd. Click to tweet

Everything looks good, yes? Not so fast. Some robotic writing applications pose serious problems – not because of the poor quality of the language they generate, but because of the credibility of their humanity. I'll come back, but before raising the red flag of caution, let's talk a little about all the applications available for robotic writing.

Today, in content marketing, there are mainly three forms of machine-made content:

Modular Content: Content strongly based on a model, based on a formula; it is often data-driven, and the breadth and scope of these content projects make them great candidates for machine-created content.
Creative Support: Tools that Give Writers Superpowers; Rather than writing from scratch, these AI-based tools reinforce the creative and analytical powers of human authors.
E-mail: AI-based interaction bots using automatic natural language processing maintain one-to-one conversations with prospects and customers – via social media, social media, and social media. web applications or even email.

Robotic content becomes widespread

The most interesting possibilities for using machine-based modular content are predictable, repetitive, model-based, and, in most cases, heavily data-dependent types of projects.

Years ago, when I left my traditional job to become a freelancer, one of my first gigs consisted of transcribing quarterly earnings calls and preparing summaries at the same time. intention of the press. Writers like me felt immense pressure to complete the tasks quickly, as each task was billed at a modest and uniform price. If you do not move at breakneck speed, your hourly rate may drop close to the minimum wage. (I lasted about two weeks.)

Today, these tasks are supplemented by machine learning technologies. Associated Press, for example, uses Automated Insights' writing robots for business income appeals. AP released 20% of journalists' time, reduced errors and improved turnaround times.

The editors for corporate profits have freed up time for journalists and reduced errors, says @AssociatedPress. Click to tweet

When the program was launched, Philana Patterson, assistant editor in charge of business at the AP responsible for implementing and overseeing the system, explained, "One of the things that we really wanted the reporters to be able to do was to get out of not having to focus on the initial numbers. That's the goal, to write smarter texts and more interesting stories. "

(Even if it is possible that some machines replace writers, believe me, when I say that this type of work does not lead to a satisfying writing career.)

Artificial intelligence tools also help companies convert data into stories. Dominion Dealer Solutions allows auto dealers to create ready-to-use profiles of cars on their lots. In the past, a dealer could cut and paste information from disparate sources to create a sales sheet – an extremely inconvenient process for dealerships with thousands of new and used cars in inventory. With the help of Narrative Science, dealers can access a sales sheet in real time, which helps unlock clear English descriptions from a wealth of vehicle data.

This application is not just about the sales record, of course; it is also a question of referencing. The use of robotics has allowed Dominion Dealer Solutions to increase the number of page views by 20% (including a 50% increase in the number of pages viewed on used vehicles), as well as to increase revenue from inventory.

RELATED CONTENT HANDPICKED: Content Creation Robots Are Here [Examples]

Applications Support Human Creativity

These cases show that AI replaces the human, but there are many examples of AI completing the work of human writers. To understand the future of robot-assisted content, it is useful to look at the manufacturing sector, where more and more companies are using "cobots", robots designed to interact with humans and help them in a better way. shared space.

Cobots often perform repetitive tasks in partnership with a human being (eg, packaging, assembly and quality control, quality control tasks) and eliminate or minimize the more tedious and dangerous tasks on assembly lines.

Writing tools with artificial intelligence are very similar to the relationship between humans and cobots. The most common content cobots are Google's Smart Reply and Smart Compose features in Gmail. As you type, Google predicts the words that will probably follow. To this day, predictive content cobots such as Google's do not keep their promises. Google Smart Compose does not save a lot of time: the keystrokes needed to complete the sentence are clumsy so that in most cases it is simply faster to type the same words.

Give him time. As Google gains access to more of your data (thrill), its deeply apprehended artificial intelligence will predict more accurately what you want to say. Ray Kurzweil is leading the effort and his team is working on a way to develop Smart Reply. He explains, "You could have similar technology to help you compose documents or emails by giving you suggestions to complete your sentence."

Apart from these types of predictive cobots with natural language processing, there are dozens of cobots with creative support on the market. Among my favorites, Frase, a word processor powered by artificial intelligence with an integrated search assistant.

Dozens of creative support cobots, like Frase, help make storytellers more effective, says @clare_mcd. #AI Click to tweet

While I'm writing an article, Frase is looking for media and third-party research, and summarizes the relevant sources. The tool even creates footnotes / endnotes based on the information cited. Frase is not intended to replace human storytellers, but rather to make them more effective and efficient.

RELATED HANDPICKED CONTENT: How to Secure Your Work as a Content Creator

Peeps can spoil the party

All of this sounds relatively benign, does not it? This is largely the case, but a small and growing number of AI-based interaction tools blurs the line between machine-created content and something more disturbing. .

Marketers are aware of chatbots – AI-based interaction tools that can answer customers' simple questions and replace the role of humans for more mundane conversations. Most chatbots disclose their machine identity in these interactions, even though a clever observer can usually tell when they interact with a machine. But all this will change quickly.

Lisa-Maria Neudert is a Ph.D. candidate at the Oxford Internet Institute and a researcher at the Compagational Propaganda Project. Among other things, she studies how businesses and political groups use chatbots to interact with their audiences. Lisa-Maria predicts that as machines become more adept at imitating the quirks of human language, while pursuing broader conversations, people will be much less able to identify robots. And this change will facilitate the exploitation of chatbots by bad actors.

People will be less able to identify bots because machines become apt at imitating the quirks of human language. @lmneudert Click to Tweet

She explains, "(Propaganda robots) will present themselves as human users taking part in the online conversation in comment sections, group discussions and message boards. , one-on-one, private conversations … and do it on a large scale.

In the MIT Technology Review, Lisa-Maria writes that it will not be long before "chat robots can search for susceptible users and approach them via private chat channels. They will navigate eloquently through conversations and analyze a user's data to deliver personalized propaganda. "

To be clear, content marketing has powerful and positive applications for robots. Conversation-bot maker Spectrm, for example, has created a public health chatbot in order to give advice on the next day's contraceptive pill. Spectrm co-founder Max Koziolek explains: "This is one of the cases where a person may prefer to talk to a bot rather than a human because they are a little embarrassed. They talk to the bot about what they should do after having unprotected sex and it naturally includes 75% of the queries, even if they write in a language that is not very clear. "

Still, good examples aside, marketers will soon develop messaging chatbots that mimic human language so closely that they will be tempted to use them more secretly.

Marketers are not required to disclose when the content is created by a machine, but they should do so when the conversations are done by a machine. California law, Botaggedon, states that companies must disclose information when users interact with a bot (for example, "Hello, I'm a bot"). The law is primarily addressed to social media chatters, thus leaving businesses a way to apply them to other cases of use.

Marketers Should Absolutely Disclose the Moment of Machine Conversation, @clare_mcd #AI

The content is still about the message, not the method

Despite California legislation, marketers are in a gray space as technology evolves faster than the regulation of this technology. Just look at how governments are trying to catch up with the privacy breaches of companies like Facebook.

People have known for years that Facebook, Google and thousands of other smaller players monetize their personal data. But governments and consumers have been slow to understand the implications of this data exchange.

The same goes for conversations powered by robots. Governments are still wondering how to limit the two armies of influence on elections and political activism. They have not yet begun to struggle with the implications of company-sponsored bot conversations. (Let's be realistic: it's not even in their radar.)

Given that regulation is unlikely to happen in time to curb inappropriate behavior in machine-driven conversations, consumers' choices may be. A study conducted by the Pew Research Center in October 2018 found that only half of American consumers felt that companies using robots to promote their products were acceptable (it was not really a resounding endorsement). And among those who understand the robots better, the acceptance rate is much lower.

While marketers are striving to find new and different ways to reach their audience (hello unattractive locked screen advertising, unsolicited text messages and web browser notifications), consumers are finding new ways to limit access. In the end, it's always the message, not the method. Publish great content, provide useful information and your audience will open the door to let you in, whether that content is created by a machine or a human.

RELATED CONTENT HANDPICKED: Data Protection Act: Ignorance is not an excuse

Note: All the tools included in our blog posts are suggested by the authors, not by the CMI editorial team. No post can provide all the relevant tools in the space. Do not hesitate to include additional tools in the comments (from your company or those you used).

Post more content on robotic writing – answer a quiz, explore robotic writing platforms, and more. – in the Chief Content Officer digital magazine. And while you're here, be sure to subscribe.

Cover image of Joseph Kalinowski / Content Marketing Institute

Related posts

Leave a Comment