AI Revolution in Journalism: What It Means for News

Watch the video:

Channel 1 aims to revolutionize journalism with AI-generated news anchors delivering personalized news content to anyone in their own language. Even the voices you hear will have live translations for you. And this is not a possible future but is set to happen this year. 

Are we doomed to consume AI-generated content and avatars from now on? Not being able to differentiate generated content from real content?

When it comes to journalism, I think journalists are there to stay, but things will change. There will be fewer of them, and they will leverage AI tools. Let’s dive into how AI impacts journalism as we speak and discover what we should be anticipating… or fearing… along with some personal advice for current or upcoming journalists.

I’m Louis-François Bouchard, and I want to note that I am not a journalist myself. My goal is to make AI more accessible here on YouTube and with my company, Towards AI, and here’s a new series where I intend to cover the impact of AI in today’s industries. In this first episode, I dove into the world of journalism to cover everything AI-related that is or will be applicable, from generated articles, the future of journalists, dealing with biases in AI, and keeping objectivity to the ethical dilemmas posed by deepfakes and misinformation.

Join me in this mission of making AI more accessible by subscribing and sharing the article or insights from it with your group chat!

Now, the questions many journalists asked me: how will AI change journalism? Is my job doomed?

Let’s answer that by first diving into a little throwback of how new technologies have affected the field to help us grasp what’s to come.

A quick history of the field and technologies impacting it…

Journalism has evolved significantly alongside technological advancements. Journalists used to rely on print newspapers and hand-operated presses. They then had access to steam-powered presses, allowing mass production of newspapers—the first mass communication of news to the population. We saw the telegraph arrive not long after, enabling instant long-distance news transmission. We could now reach far and wide.

Then came the radio and television, revolutionizing news delivery with audio-visual content and live broadcasts enabled by satellite technology. We could now hook people even more with interesting visuals and add more and more personality to the content we share. 

Finally, the digital revolution arrived, shifting from print to digital platforms and the rise of online news portals, and blogs, further enhanced by the advent of computers, word processors, and mobile devices, including the phone you have probably checked a minute ago. These technological strides have continuously transformed journalism into the dynamic and instantaneously accessible field we know today. AI is another step in that direction, changing how journalists create stories.

Current use cases of AI for the field

But what about how it works right now? The current workflow involves various digital tools and platforms. Research and investigation rely on digital archives, Googling, and databases, with social media offering real-time updates and public opinion insights. Something we can keep an eye on, but no human can analyze this vast amount of information in real time on their own. Data journalism utilizes big data and analytics to dig up stories, while content creation has shifted to digital mediums like Google Docs and CMS platforms, integrating multimedia elements such as videos, images, and audio for the pleasure of the users. Verification of misinformation and fact-checking are increasingly supported by digital tools.

As you already know and use, the distribution of news has now moved to social media platforms like Facebook and Instagram and aggregator apps, with reader engagement encouraged through interactive online platforms with comments you should probably never read and articles optimized with tons of analytics tools. All of those already leverage AI in some way. An old 2019 research points out that using AI in journalism can help automate an editor’s job by 9% and a reporter’s job by 15%. Yes, 2019 is quite old for the AI world. Imagine, ChatGPT didn’t even exist yet. That was from some early adopters! I will claim that the numbers are more in the hundreds of percent for an experienced user today, at least doubling their productivity.

As another fun fact related to jobs, American newspapers employed nearly half a million people in 1990, but by 2016, this number had nose-dropped to 183,000, around 60 percent less! It should be even smaller now 8 years later. Ok, maybe that wasn’t such a fun fact after all… This continual drop is partly thanks to tools that make current journalists more efficient, requiring fewer professionals. These tools are now called “AI tools,” a necessary evolution in how we approach journalism — similar to the shift from pen and paper to keyboards.

AI's impact on journalism is already evident. It assists in generating summary texts for reports, sports events, interviews, and debates, optimizing article titles and keywords for better online discoverability, and analyzing large datasets to identify trending topics and find the next Star Wars kid viral phenomenon before everyone else.

You can also directly use tools to improve your work, like Grammarly and Antidote, using AI for editing and correcting. Speaking of improving news reports, even though AIs are biased (nothing isn’t), it should be less biased than one human being and can help you find your own biases in your piece to correct or change them. We also now have access to AI transcription services that provide real-time transcription instead of transcribing everything yourself or hiring someone to go through the video, debate, yesterday’s sports game, or meeting and write it down. It’s also much better than the transcriptions you were used to seeing on your muted gym TV. Just be careful when dealing with heavy accents or slang, as AIs are quite good for general purposes but much worse when dealing with an edge case like a unique or unconventional accent.

Even though we see some failure cases, it can do some pretty incredible things, like generating entire articles or journalists themselves. The Sports Illustrated magazine recently had a controversy revealing the potential for AI where they had fake journalists writing articles. They used AI to generate articles but also to create journalist profiles. Of course, I don’t recommend doing this, but there’s a lot you can do using AI to benefit your work, which we’ll dive into in a few minutes.

Channel 1 is another example of using AI-generated content and, here, even generated avatar reporters. They create fake people, including their voice translated live in your language of choice. Of course, as of the time I am writing this, we have little information on the actual technology and just a demo video, but we should expect to see it start in 2024. I am sure it will be quite impressive, as in the video demo, though not perfect. But what’s some buggy pixels if you can get any news from anywhere in your own language?! Plus, these issues will be fixed faster than we expected. Progress in AI comes sooner than any technology we’ve seen, which makes the future unexpected use cases not that far away.

Potential future use cases of AI (some fun guesses and predictions)

Let’s see what AI will soon be capable of, drawing from examples like Channel 1. I believe this text, audio, and even video with lip-syncing technology will soon become available in most languages for anyone to use cheaply. It’s already available for the most popular languages and relatively cheap using tools like Sync Labs. This leap forward will enable an even larger global reach breaking down language barriers. Something even the internet could not achieve yet. This single concrete example shows me how AI is the next technological revolution following the Internet. And this example is just one of the many we’ll see happen soon.

Now imagine a TikTok-like algorithm for news, providing better personalized content that resonates with individual preferences. AI's ability to predict trends will be a game-changer—a scaled-up version of what we have now. Journalists will be able to know what to cover even before it becomes a trending topic. Ok, maybe not always, but journalists will be so efficient that they will have the content ready if any of the identified topics become trending. Journalists who leverage these new tools will be incredibly productive and have enhanced report quality. AI will be used to create content that mimics a journalist's writing style or voice, bringing a human-like persona. This is not just to replace them but to allow them to produce more. Of course, there are always two sides, whether you are the bad guy or the good guy. I have some advice coming next, but please skip them if you intend to be one of the bad guys!

With these advancements come risks. Improved recommendation algorithms might lead to an echo chamber effect, where individuals are only exposed to news that reinforces their existing beliefs. This could further deepen ideologies and conspiracies for the better or the worse. I’m personally quite excited to see more and more content proving that the earth is flat on my feed. The truth will prevail! I’m just kidding, by the way.

But it’s true that the control over these algorithms raises concerns about population control and bias. Instances like the recent fake videos of Justin Trudeau and Elon Musk used to promote a scammy investing app are just one of the first cases showing the potential for misinformation on a larger scale. 

The job landscape in journalism will inevitably evolve as it has in the past, thanks to the printing press and the Internet. It doesn’t mean the end of journalism. It just means traditional roles need to adapt to include this new technological step. We might see fewer journalists in the end, as we’ve seen since the 90s, but their roles and talent will be more critical than ever, focusing on ensuring the accuracy, quality, storytelling, and neutrality of AI-generated content. Editing generated content, making it better, or simply writing what they truly enjoy writing about and having the AI edit and improve their text, similar to an editor or simply a good friend reviewing your work to give feedback. It’s just that this new friend will remember everything, unlike me, as it has incredible knowledge and memory, plus direct and instant access to the Internet and any database you may have.

Speaking of knowledge, these roles could ultimately transform into more data-focused and managerial positions, overseeing AI's outputs and understanding where the data comes from. There's also likely to be an increase in fact-checking roles to address potential AI hallucinations and biases that I believe are there to stay with generative models. The field will increasingly welcome data and AI experts to guide and enhance journalistic practices.

The future of journalism with AI is about balancing human judgment with machine efficiency to tell stories that matter: not to replace journalists with AI. Tasks will simply evolve as they have evolved with every new technology affecting the field. I want to make it clear that I believe humans will stay and remain important, and here is some advice to stay in the loop…

Advice (pure guesses, don’t invest based on these!)

Now for the part you may have been waiting for: I have some advice for aspiring and current journalists on leveraging these new technologies. Remember, these are just my perspectives and not investment advice. I’m just posing guesses on what could happen and some advice assuming it does. This is to take with a few grains of snow! Yeah, that’s how we say it here in Montreal when it's -20 degrees.

Firstly, it's crucial to understand the basics behind Large Language Models (LLMs) like ChatGPT. Get to know the data used to train these models to understand what they expect to receive as prompts, why they hallucinate, what those hallucinations are, and how to prompt them best, considering all those points. By the way, AI models like ChatGPT are constantly hallucinating. They predict the next word based on previous words you sent them and on all they have seen in their training. When was the last time you tried to think of the next word to say one by one? We don’t do that. Just try doing this to convey something you want. Going step by step like this works for walking, but it’s just impossible to come up with a proper story while doing so. We think of an idea or story to share and, on the opposite, skip words with no problem. Being a native French speaker, I often lose my words in English, and I just get around with other words or simply skip them when I see the other person understood what I meant. LLMs don’t can’t do this. They don’t work like our brain does. They generate numbers, which we associate with words in a big list, and, fortunately for us, statistics and lots of examples make it so that those word-by-word hallucinations are true most of the time, which is super impressive for sure. Still, ChatGPT and other LLMs have no concept of what is really true or not, and I don’t even want to enter into the consciousness question here. They work differently, they think differently, and they create differently. You can simply see them as a new tool you can leverage. Large language models are a Google on steroids. Nothing here replaces human intelligence. We have much more than mere intelligence and facts knowledge, anyway, which is a real blessing for those who don't have a memory like me.

These AIs have skills we do not have, like the ability to come up with okay text at lightning speed. And don’t worry about the current lack of personality or the many errors they make. Those are superficial mistakes. They will keep improving and evolving, which is why we have to stay updated with these ongoing developments. For those interested, my videos and newsletter offer a free way to understand AI technologies and their impact on our lives every week. If you want to dive in a bit more and are not that familiar with the field, I really liked the course "AI For Everyone" by Andrew Ng on Coursera. It is an excellent resource for gaining a good base understanding.

Now, let’s talk about leveraging the tools that are already available—starting of course, with our game-changer favorite app: ChatGPT. If used correctly, it can significantly aid in editing articles, brainstorming, expanding or detailing specific ideas, suggesting a few additions to your article, selecting just the best ones, simplifying complex topics for your next article, and even generating some cool visuals. You don’t need much other than ChatGPT or another similar language model like Gemini. You just need to learn to use it properly, and you’ll see how powerful it can be. There’s no magical general tool to help you, but ChatGPT is pretty damn close! I believe it's still essential to approach them with what we can qualify as informed skepticism. The problem is that what they generate looks real whether it is or not. Always verify the information and understand that, despite sounding convincing, the AI has no idea if it is true or not. A quick tip would be to force it to look on the internet to confirm its sources so you can easily double-check, something ChatGPT Plus users can do. Another more advanced tip is using something called RAG to augment your language model with a database of knowledge that you control. I made a whole video about this technique if you are curious.

Oh, and please don't dismiss ChatGPT after a single try; it's a tool that performs better with time and practice. Its efficacy and your proficiency with it will grow, provided you give it the chance to evolve and you adapt your methods to utilize it correctly. Even if I’m working in this field, I waited for 6 months before fully utilizing ChatGPT for coding and other tasks because each time I used it, it was unsatisfying. Trust me, give it a shot for 1 month using it daily, and it will change your life. If you don’t want to, no big deal; just make sure you keep improving and creating value, and you’ll be good. Some writers still work with a pen and paper. It’s not all about efficiency.

But life is not just about text. If there's a new software for automatic video generation like InVideo or Runway, a translation tool like DeepL or copy.ai, or audio production like the Adobe podcast AI toolset, try it in your next report. The quality of these tools is improving rapidly, and they can add a surprising edge to your content. They can bring something you didn’t even think about or that you were convinced you could not come up with by yourself. Likewise, you can use those new tools to come up with some very fun and unique experiences on your platform for the users to make the content more relatable or interesting, like having cool guest voices reading your articles automatically using something like Voice Actors by ElevenLabs—a platform where artists and individuals can permit and monetize the use of their voices, which can even be complemented with a cool avatar using something like D-ID, add AI-generated music to fit the style of your article or video with Suno or Soundraw, or simply to have your personalized chatbot either for your own archives or to put on your website.

But technology isn't the end-all. Developing creative skills, particularly in storytelling, is more important than ever. In an era where AI handles the bulk of data processing and allows many more people to get started and write more, the human element of crafting compelling narratives becomes more crucial than ever. Quality is more and more important as quantity and offer increase. Understanding what makes a story resonate with your audiences is key. 

The human focus should be on producing quality content and providing value. Once this is achieved, AI is there to help with the quantity. If your articles or products are widely read and appreciated, you’re on the right track, and nothing needs to be changed. This section is particularly crucial advice for our juniors in the field or smaller networks. Leveraging AI to augment your capabilities is essential, but never at the cost of losing storytelling skills and practice. The backbone of journalism is from individuals we can relate to, understand, question, doubt, love, or even hate. We enjoy unique events like personal anecdotes or even the errors from performers that we were one of the few to see happen live. We want to be part of it. This cannot come from a machine, regardless of the quality. What makes a piece of content good is partly our personal interaction with its creator, whether it's music, a movie, or, as is the case here, an article. Making these algorithms as good as expert writers and journalists will not happen in a day or a year, and it may never happen. It’s much easier to start writing than to become one of the best that people want to read. Who knows if we’ll ever be able to replicate this in a machine?

Another point to consider is that journalists don’t only write. They also conduct research, investigate, and do interviews. As I discussed with Karim Benessaieh, a renowned journalist here in Montreal, interviewing someone requires a human connection and skills to extract valuable information or even seduce them to get the interviewee out of their automatic responses. The interviewer needs to understand the feelings and leverage that information to the best of their ability. It’s a lot of psychology and skills I believe only humans can replicate. Of course, AIs can scan you and learn to do that, but you won’t feel such a deep connection believing the AI understands your situation as much as another human would. It cannot share your suffering or joy.

To go back to the advice, I also think journalists should gather some SEO skills or at least a basic understanding of what attracts clicks online or on their distribution platform to better manage and work with generated content and ideas.

Finally, transparency is crucial when using AI technologies in journalism. Being open about using AI tools maintains your credibility and helps shape a responsible future for AI in journalism. In short, try out the new tools; if you like the outcome and use them, say it!

Conclusion

As we’ve seen, AI is not just a futuristic and far-away concept. It’s a present reality transforming the field of journalism in many ways. From automated reporting to personalized content curation, AI is reshaping how news is created, distributed, and consumed: affecting everyone. And this is just the beginning. The potential for AI in journalism is vast, and its full impact will be just as immense, if not more, than the Internet. That’s why you need to stay curious and play with these new technologies. Imagine deciding to ignore the Internet in the 80s. Would you have kept your word 40 years later?

I know lots of these technologies are very different from what we are used to, so I also recommend staying informed about the latest developments and understanding the changes, which is quite easy if you subscribe to my channel or to my newsletter!

Of course, AI has risks and downsides, but it will ultimately change it for the better for the population, as we’ve seen with all technologies in the past. I’m quite optimistic about the journalism industry with the conviction that AI will be leveraged to help current journalists do better work and focus on what they like to do most: write engaging stories that report the truth in an objective, accurate, and fair manner to inform us, the public. In the end, just be transparent in your use of AI and have fun experimenting! AI doesn’t have to be feared. It just has to be leveraged properly.

Thank you for reading the first article of this series, in which I will cover how AI impacts different industries. What industry would you like me to cover next? Let me know below, and I will see you in the next one with more AI explained!