In a short space of time, generative AI – defined as artificially intelligent technology able to produce rather than analyse or regurgitate existing content – has gone from being roundly dismissed to a force to be reckoned with.
The content generative AI can produce spans imagery, video and text, and has been increasingly visible on social media platforms as users test and promote the technology’s limits and capabilities.
However, ethical questions remain around the technology and its use cases – should publishers really be using AI to generate content? We dig in to some of the issues surrounding generative AI and its use in publishing today.
Can AI be trusted?
While AI is not exactly new to publishers looking to improve personalisation, create intelligent paywalls or iterate quickly, this new generation of generative AI tools are novel and not without risk: the rise of artificial intelligence chatbots such as OpenAI’s ChatGPT and Google’s Bard has already given rise to ethical dilemmas for journalists attempting to compete with the tools’ rapid creation abilities. Can AI really be trusted to deliver accurate, unbiased information?
Open AI released GPT-4 in March 2023, bringing with it the ability to accept visual as well as textual inputs. Cofounder Sam Altman calls it their “most capable and aligned model yet”, claiming the tool is more creative, less likely to ‘hallucinate’ incorrect information, and less biassed than previous versions.
Proponents argue that the efficiency gains in production and related reductions in labour-intensive tasks such as interview transcriptions or data modelling may trump concerns over remaining accuracies or bias.
Others maintain that the prevailing need for human judgement and fact-checking diminishes these benefits to the point of uselessness. As ever, the truth likely lies somewhere in the middle of these two opposing viewpoints.
The business case for generative AI
Publishers cannot afford to be blind to the competitive media landscape driven by hyper-relevant social media platforms and the audience centricity users have come to expect. Today’s consumers expect tailored content delivered to them at lightning speed, but manual production at this rate would necessarily result in astronomical costs: not an avenue most publishers would relish taking given the limited budgets available.
AI therefore seems like an effective bridge between journalistic capabilities and audience wants, provided publishers have the necessary safeguards in place to independently factcheck and edit AI-generated content.
Some publishers have already found a middle ground, albeit a contentious one: US media, news and entertainment company Buzzfeed recently announced that it would be relying more heavily on ChatGPT to produce lists, quizzes and other entertainment content, resulting in a share price surge of 150%, although this subsequently retraced. Feedback from Buzzfeed staff centred on whether this would result in workforce reductions, although a spokesperson provided assurances that the tools would be used to allow employees to be more creative and efficient.
Embedding AI technology in publishing workflows
It’s clear that AI isn’t going away any time soon, and publishers will feel the pressure to get on board before they’re left behind.
With robust frameworks for the technology’s use and clear pathways for content creators to indicate their use of generative AI, there’s no reason why publishers should miss out on the benefits while simultaneously lessening the associated risks.
Almost every stage of the editorial process could benefit from AI input: subject research, brief writing, editing, layout and design, SEO, back link creation, testing, tracking, maintenance…
The list goes on. However, we’re still early in this adoption curve, with no one tool providing an adequate solution for all publishers’ needs.
Today, a suite of complementary AI-enabled tools embedded directly into the CMS feels like the dream ticket for publishers. And, though it’s not here yet, it also doesn’t feel too far away. It’s possible to see an AI-enabled WordPress, for example, as being the one-stop shop all editorial teams would love to use.