In the initial days of the ChatGPT excitement, I conducted an experiment with the Artificial Intelligence-based chatbot. I used it in a race against another writer, who was using the traditional research route, to produce an article on the Iranian morality police. Even with fact checks and touch-ups to the language, I won hands down.
So, what does this mean for the profession of journalism?
I posed this question to the purported villain of the piece itself. I asked ChatGPT, “How will ChatGPT impact journalism?” The AI politely suggested that it could be used to generate “simple news articles,” thus “freeing up journalists to focus on more in-depth and investigative reporting.” It even gave a caveat that it was not to be trusted as it can give “biased or inaccurate information.” It ended on a rather philosophical note: “Ultimately, the impact of language models on journalism will depend on how they are used and integrated into the news production process.” If it has more nefarious plans, it is definitely good at hiding them.
The AI is by and large correct in its assessment of its own impact on journalism. At least for the near future, AI will be a tool in the hands of a journalist rather than the journalist itself. If any journalist is worried that AI may take their job, the chances are higher that the job will be taken by a journalist who knows how to use AI.
There are, of course, areas within journalism that are ripe for disruption with AI-based systems. Any reporting based on structured data such as corporate earnings reports, sports score cards, and press releases may be out of human hands soon. In fact, it already is in some newsrooms. AI may also take over the writing of context-building articles and proof-reading. However, there is much to journalism that an AI application will find hard to replicate now, such as finding the location of a secret corporate board meeting in New York City based on who had hired the most limousines for the day. This took some non-artificial thinking on the part of the business journalist who pulled it off. There is nothing to match eyes and ears on the ground, at least for now.
Not new to the industry
While journalists may just be getting introduced to the possibilities of AI, the news industry itself has been dabbling in it for some time now. When you visit any major website, chances are that at least some of the articles you see have been specifically selected for you. These websites often track your reading habits and then apply machine learning to figure out which articles to show you so that you stay on at the site. For all our grouses with the social media companies, news media did learn a trick or two from them.
At The Hindu, we are trying out AI tools, particularly in Natural Language Processing, to create versions of our articles for various modes and durations of consumption. This involves training the machine to simplify an article without the crux of it being lost. In this context, it was self-affirming to see the vendor of an AI service confidently feeding in our editorial for the day and the machine failing to provide a coherent output. We now had virtual evidence for what we and millions of India’s civil service aspirants have always believed — there is no better way of stating the fact of the matter than The Hindu’s editorial.
george.pj@thehindu.co.in