AI in 2024: Best Practices and What to Expect

1. AI Is Not Replacing Journalists

Writers and editors can do certain things that AI cannot do.

Joseph Teasdale, Head of Tech at Enders Analysis (London, U.K.) states, “Everyone got terrified that AI would replace journalists, but that hasn’t really been the case — journalists still have access to the real world to conduct investigations and discover breaking news; a human element helps for some of that.”

With that said, AI is clearly a powerful tool and companion to journalism.

“So much great work we do as humans is through language, so AI can be very powerful with its strength in language learning models,” Teasdale added.

What We’re Doing at RebelMouse: We are building our Entry Editor with AI tools that empower writers to produce more and attract more traffic to their content. With tools like our AI Headline Generator and additional editorial-assistance tools, writers can expedite the creation of optimized, high-traffic pages.

2. Be Experimental

You’ll extract the most value out of AI by running experiments and testing different types of prompts.

“Run experiments with AI, but do them in a safe way. Know the red lines you’re not going to cross with AI,” said Lucky Gunasekara, CEO at Miso Technologies, which specializes in AI and predictive insights. “You want to surf the wave but you don’t want to drown — bring as many people from your organization into AI as possible, from business side to legal, etc.”

The best types of tests are not one and done — they are continuous.

“Run a test, iterate, strategize, run a test again — have a cycle for experimentation,” said Louise Story, recent Chief News Strategist at the Wall Street Journal and a New York Times alum. “Be OK with failure — that’s part of experimentation.”

What We’re Doing at RebelMouse: We are constantly running AI tests, learning what leads to the best results, and replicating that in future tests. That’s how we learned that AI recirculation units for related content lead to a 2–3x better click-through rate, and AI-generated teaser headlines improved CTR up to 3x when A/B tested against original headlines.

3. Get Your Data in Order

The backbone of the most successful AI endeavors is data.

“If your data is in a lot of different pockets, it can’t really play in this game of AI, which is a data transformation,” said Story. “All of your media, videos, photos, everything should be in a big data lake, and if that doesn’t exist yet you should build it. You should have a good data scientist on staff with machine-learning knowledge. Build a pipeline so you can take action on your data in real time.”

What We’re Doing at RebelMouse: We embed the content archives of our publishers, converting them into vectors that natural language processing can efficiently map. After that is finished and the data has been properly processed and stored, we can deliver the best results with AI, such as powering search functionality, chatbot interactions, and recirculation units.

4. Build Your Knowledge of the AI Toolset

The more you use AI products, the better you’ll get at using them more efficiently.

“The best thing you can do is play with the models, whether it’s Bard or ChatGPT — build an intuition for the types of capabilities they possess,” said Paige Bailey, Lead Product Manager at Google DeepMind. “Use them to help address problems you face in your business every day.”

What We’re Doing at RebelMouse: We constantly experiment in the sandboxes of different AI models, spending hundreds of hours collectively on writing the best prompts that generate the best results.

5. Think Visually

AI has altered the multimedia world with new opportunities. Expand your horizons to consider visual implications.

“AI isn’t just about text — think about questions like ‘Show me how to improve my golf swing’ or ‘Show me how to make an omelet’ that provide videos or images that are relevant,” said Bailey.

Marketing representatives at the Google Initiative see AI as an opportunity to make coding and visual design more accessible to a wider audience.

What We’re Doing at RebelMouse: We’re developing new features in our Entry Editor that incorporate AI images and delve into the multimedia world. We’re also thinking about video and how we might implement new tools in that area.

6. Help Your Readers Engage

We’re not just in the business of building tools for editors and writers. We are also creating functionality with AI that readers can interact with directly.

This is something that Google has been thinking about, too.

“(With our partners) we’re looking at adding a box at the end of articles that allows the reader to ask any question they want — the ability to ask follow-up questions to what they just read,” Bailey said.

What We’re Doing at RebelMouse: We’ve developed chat-based experiences, such as Narcity Chat and the Reliabilityweb Chatbot, that allow readers to have natural language conversations with AI-powered content engines. We’ve also improved search with AI, delivering outstanding results based on contextual mapping.

7.  Inform Your Editorial Staff

Improving insights and analytics at an editor’s fingertips is one thing that AI can do really well.

“We have a project called Article DNA — what is an article, how is it written, what is the style, who is it about, and what is the emotion of the article,” said Bram De Ruyck, Head of Data Science & AI at Mediahuis Group. “We use these insights to grade the editorial department, how they did this week — a dedicated, informed dashboard.”

What We’re Doing at RebelMouse: We are currently working on a scoring system that will evaluate your content for maximum search engine optimization and provide recommendations that AI can quickly and simply fix for you with a few clicks.

8. Think About Speech

AI goes beyond text and images, as it can expand to the world of voice, too.

“We’re working on a generative chatbot that you can speak to,” said Olle Zachrison, Head of AI at Swedish Radio. “It sifts through two million articles and we’ll deliver answers based on trusted news sources.”

What We’re Doing at RebelMouse: We currently have a text version of this same concept, but voice is an area we’re actively researching and thinking about around how we can use it to improve our product and the user experience.

9. Consider a Virtual DJ

Custom projects like a virtual DJ are within reach with the power of AI.

“We’re creating playlists and an AI voice does an intro to the playlists, personalized based on the songs — like a virtual DJ,” said Zachrison.

What We’re Doing at RebelMouse: We work with our clients to create custom AI implementations that address their objectives. For example, with PowerToFly, we have an AI-powered engine that helps sift through hundreds of thousands of résumés to match up candidates with specific jobs. We want to use AI to achieve anything you can imagine, and we’re always open to working together to create unique solutions.

10. Reassess Frequently

It’s still early days in the world of AI, ripe for pioneering and innovation. That means constantly reassessing your approach to AI along the way.

“The work you do around AI should be approached as living and breathing, as things can change quickly as the technology continues to develop,” said De Ruyck.

What We’re Doing at RebelMouse: At RebelMouse, we keep an eye on industry trends and develop our AI knowledge by using growing AI technology on a daily basis. We are never content with our current suite of AI solutions, so we always work to enhance them and look for new opportunities.