Five Practical Use Cases for AI Generated Content in 2018
Strategy | Personalization | Artificial Intelligence
Marc Vieleers

By Marc Vieleers | CEO 09.01.2018.

Five Practical Use Cases for AI Generated Content in 2018

Gartner predicts that 20% of all business content will be machine generated by the end of 2018. While AI has always been a sort of mysterious topic adorned with bells and whistles, the practical applications can be quite simple and are becoming more real every day. We wanted to share five simple and practical ways AI generated content can and should be utilized in 2018.  

(Technical note: For the sake of simplicity, this article considers ML as a subset of AI.) 

SDL recently shared some great information on how AI generated content and machine translation can save big companies a ton of money and time. They compared costs of human created content for a travel site such as Booking.com to the same content being generated by AI and ML. Creating a 100 word hotel “inventory” in 8 languages costs about $22 each and for 100,000 hotels, a total of 22 million dollars. With ML this same content can be created for $5 each and in three seconds time, resulting in an end cost of 5 million dollars. 

AI can easily generate content from data, rule sets, and models, and any platform requiring repetitive information input and output can use it to become more efficient. Here are five basic use-cases to support the SDL and Gartner sentiments: 

1. Hotel Inventories:
There is great potential for Natural Language Generation (NLG) when it’s paired with the hotel and hospitality industries. Every hotel has numerous specs when it comes to location, hotel facilities and amenities, and even extensive review data. AI can consume all of this and generate simple, highly-readable paragraphs and experiences, in essentially any language, and distribute this content via any platform. With the big hotel booking sites (booking.com, expedia.com, etc) supporting millions of hotels around the world with properties added and removed daily, machine generating this stock content saves millions of dollars and countless hours.  

2. Real Estate and Rentals:
Another example of where stock formats are used repeatedly and across multiple websites, apps, or other outlets. Using image recognition neural networks, content can even be generated from photo input. A highly applicable case when it comes to real estate sites and apps. Top ranking real estate websites such as zillow.com, trulia.com and funda.nl could streamline content posting by using AI that can generate home inventories from seller submitted photos.

3. Sports Reporting:
The Washington Post has already tested out AI generated content for its sports reporting with a program called “Heliograf”. Statistics are meticulously recorded in sports, making it an ideal input for AI to output a descriptive, journalistic-style synapse of the game or event. Broken down by innings, quarters or points scored, just a few minutes after the buzzer rings, a fully detailed story can be sent to any publication or news site.

4. Product Descriptions:
Why are the big online retail giants so instrumental in creating AI NLG technologies? Because they’re benefitting from it – big time. There are hundreds of millions of products being sold online at mega e-retailers like Amazon and Alibaba and they all need descriptions. The more detailed and helpful the description, the more likely a person is to buy because of it. Smaller, country specific retail platforms, such as bol.com in the Netherlands, could also benefit from AI generated product descriptions as their platforms scale.

5. Travel and Events:
As SDL pointed out, the travel industry can stake many claims in NLG and Machine Translation (MT). World famous theme parks such as Disneyland provide tons of stock information, directions, tours and itineraries to their web visitors. Every time there is a new dining option, show, play or event, there is an opportunity for AI to go to work. Especially considering the fact the Disneyland and affiliated websites are offered in 15 languages, they can also greatly benefit from using MT. 

We’ve shared five simple yet practical use cases for AI generated content and it certainly doesn’t end there. Profit and loss summaries, quarterly business reports, real time stock insights, and any significant source of data can be consumed by AI to generate human-readable content. Rather than thinking of AI generated content as a threat or a concern, think of it as an aid and a tool to help get people away from spending time on small repetitive tasks and instead working on things that AI can’t (yet) accomplish. 

In a future article, we’ll also dive into the even more exciting world of AI generated personalized content – where personalization and content generation meet in real-time.

Marc Vieleers

By Marc Vieleers | CEO 09.01.2018.