‘We should not be so afraid of jobs going away as much as jobs changing, new roles and new tasks emerging.’
Writing templates for automated news systems is emerging as a new skillset, says Professor Nicholas Diakopolous. In a brief interview with DIGIDAY, he says these kind of new tasks will shape the future of journalism. Journalists being assisted by AI tools will be more evident than AI systems taking over from human journalists.
- Before autoscribe systems can churn out stories by the thousands, the underlying logic needs to be written as a template. This requires human news judgment and anticipating future story scenarios that can be derived from a data set.
- Journalists and AI systems are co-evolving to handle complementary tasks. Diakopolous sees more hybrid workflows ahead.
- As well as automated writing, AI systems can tailor layouts and news line-ups to better match existing content with user interest, he says.
Diakopolous heads the Computational Journalism Lab at Northwestern University in Evanston, Illinois. His 2019 book, Automating the News: How Algorithms Are Rewriting the Media, is often cited and a leading source in the field.
OUR TAKE
- So far, human judgment is needed both before and after AI outcomes in journalism.
— The ‘before’ is to set the parameters of the system and assess the inputs it will be using, for example, the integrity of the data set being fed into the system.
— The ‘after’ is to verify the results for clarity and accuracy. - Specialty functions will require specialty training before practices become widespread. Current capacity has been bootstrapped by individuals, sometimes with assistance from suppliers. This isn’t scalable. More organized instruction will be needed if it is to become a general practice in industry.
- Having more ‘humans in the loop’ is likely for the foreseeable future. Variables will be the sophistication of the systems and confidence in their outputs.
- The trend to watch over time is how much AI tools are assisting human outcomes v how much humans are assisting AI outcomes.