How journalists and newsroom managers in Africa should think about the rise of AI
As artificial intelligence (AI) disrupts journalism around the world, African newsrooms are beginning to feel its impact too.
With news organisations in Kenya increasingly integrating these tools into their daily work, the most common feeling among working journalists is fear: many think AI will increase job redundancies at the worst possible time. But a few reporters and editors are a bit more positive, and are already using these technologies to report more accurately on huge databases and to fight disinformation in a country that has been targeted by bad actors in the past. This piece provides an overview of how newsrooms think about AI in Kenya with the help of five working journalists and experts on the field.
Fearing fear itself?
Mercy Chelangat, a journalist at Nation newspaper, says that journalists shouldn’t be alarmed about AI. “I don't think it threatens journalists' jobs,” she says. “In the case of ChatGPT, for example, it gives you information based on what has been fed into it. That means a journalist must go the extra mile to confirm the authenticity of the information and whether it’s up to date and credible or not.”
Chelangat stresses that AI tools have no emotions, and they wouldn’t even know if what they create makes sense. AI should be used to enhance the work of journalists, not to do their job for them.
This recurring fear of AI is the result of a lack of understanding, says Phyian Karinge, the founder of YoungTechiez, a Kenyan start-up that makes technology familiar through media and digital content. AI is meant to make our work easier by automating any repetitive tasks that would take a lot of time to be done, she says.
What AI is for
Karinge explains that AI is very useful when sifting and analysing big amounts of data. “If your main work in the newsroom is repetitive in nature, then you should be concerned,” she says. “But I don't think it affects many journalists because there's an element of creativity in a journalist's work. AI can’t do that. AI is very useful when sifting through data. When using it, the room for error is very minimal.”
For Mwende Mukwanyaga, an open-source intelligence expert and data journalist at Odipodev, Kenyan journalists seem to be preoccupied with concerns about what AI is taking away and overlooking how fast AI is evolving and what it can do.
“We are scared. And when you're scared, it's either fight, flight or freeze. A lot of us are fighting. Others froze and are in a state of wait-and-see, hoping everything will be fine. And others are running, and AI is like that monster that catches you in your sleep. The more you run, the more it feels like a bigger animal,” she says.
Mukwanyaga says that most Kenyan journalists are yet to understand how to use AI on a day-to-day basis while others do use AI tools without noticing. An early adopter of AI herself, she says that the tools she uses vary depending on the story she is working on. She encourages journalists to check OSINT Techniques, a directory with information for conducting open-source internet research. “The tools you use for a health story might not be the same as the ones you use for a climate story,” she says.
A slower progress
Kenyan newsrooms have been making progress in the adoption of AI albeit at a slow pace compared to the West.
According to Peter Mwangangi , a senior business journalist at BBC Africa, AI is mostly being used by digital teams on their online platforms as compared to other traditional media such as radio, TV, and newspapers.
Mwangagi co-authored the paper The Adoption of Artificial Intelligence in Newsrooms in Kenya: a Multi-case Study together with Prof. Nancy Booker, Dr. Njoki Chege and Paul Kimweli. The paper focuses on the extent of AI adoption in newsrooms in Kenya, the factors driving it, and the challenges presented by this technology.
“AI has been adopted in varying degrees, from news gathering to packaging in a manner that appeals to a target audience,” he says. He notes that management buy-in of AI and the financial cost of technology have been the major factors contributing to whether a newsroom adopts AI or not.
“Developing AI models for newsrooms here in Kenya can be very expensive and we are at a time when media revenues are shrinking,” Peter says. “So, are you going to invest or allocate resources in something you are not sure about?”
Mwangagi’s words suggest Kenyan journalists might be concerned about matters that might not have a major impact on their work yet.
Mukwanyaga says that almost all the AI being adopted in local newsrooms is developed from the West. So journalists should be more concerned about the ownership of this technology. “If you google news about Kenya, sometimes you will get articles from international outlets with a very western gaze,” she says. “This is because AI data analytics is very Western.” She notes that Kenyans and Africans at large should channel more efforts into feeding the machines data that is Africa-centred. “We should start directing our resources at this issue so that, for instance, when you do a Google image search, your main results are African,” she says.
Mwangagi agrees and notes that some of the challenges of using AI in Kenyan newsrooms include language and the unpredictability of the technology. “As a journalist, are you willing to leave AI tools unsupervised? It could post unintended content without your knowledge,” he says.
Research indicates that AI has various biases including race, gender and class. Karinge explains that if the machine has only been shown full pizzas, it will not be able to recognise a slice of pizza as pizza. The same applies when using these tools for face recognition.
“Human beings naturally have a negativity bias where they want dark stories, and the same bias feeds into machine learning. So, if you are a journalist using AI tools to find stories for what’s trending, you have to watch out for that,” Karinge says.
A route to better journalism?
AI adoption has been marred with fear and scepticism in Kenya. Some journalists fear it will water down the quality of journalism in the country. Even though no news organisation has created news stories using generative AI tools, recent revelations that some newsrooms are doing it with fake bylines and quotes from non-existent individuals have been a major concern.
Amos Mabinda, a media researcher and journalist at Standard Group, says that the fact that AI has the potential to write complete stories with fake scenarios and present them as reality will increase the number of fake stories with fake quotes and will potentially damage the journalistic integrity at a product level as well as the personal integrity of the journalist.
“AI is also a good medium to spread propaganda, especially negative propaganda. If left uncontrolled, proper journalism would be chasing after propaganda on a larger scale,” he says.
Mabinda says that newsrooms also run the risk of some journalists using AI as their primary research, and not doing work on the ground. This would result in stories being watered down due to the shallow research as AI is still in its baby steps.
The 2022 Kenyan elections presented journalists with a real challenge, as misinformation and disinformation were not only circulating through written articles but through AI-generated deep-fake videos and other visual forms.
This was not the first Kenyan election where journalists were caught unaware of the scale in which technology was used to manipulate voters. In 2018, a Channel 4 investigation revealed that Cambridge Analytica played a significant role in Kenya's election in 2017 while working for the then President Uhuru Kenyatta’s campaign team. This company manipulated Kenyan voters with attack advertisements, deep fakes, and content painting opposition leader Raila Odinga as violent, corrupt and dangerous.
Mabinda says that Kenyan media need to be better prepared to fight against these technologies. “AI-powered deep fakes are powerful. So news organisations should invest more in AI to cover future elections because these deepfakes are going to spring up even more often,” he says.
As journalists embrace the use of AI, ethics and the capacity of the current laws to prevent potential harm are now on the agenda.
“Someone recently posted a video of a girl who was photoshopped into pornography. That is sexual harassment, right? But go to our sexual harassment act, what does it say on such a case? Does it recognise that act as sexual harassment? It doesn’t,” Mukwanyaga says.
The growth of AI cannot be ignored. Kenyan newsrooms will either adopt it or lose out. But more needs to be done in terms of training, and investment in AI tools that are more country or continent specific. Media managers also need to pay more attention to AI, learn to sieve out the noise, and focus their attention on how best to use AI for the efficiency and growth of their newsrooms.
Journalists also ought to call for transparency on how the tools they use are trained, and what cues or key things have been used for the machine to learn. Journalists should also familiarise themselves with the technology behind the AI tools they use as that will help them make an informed decision on which tools to use.