Explosive revelation: clickbait is dying, and you’ll never guess who’s responsible!
These titles are as annoying as they are essential. If they tend to annoy some people, who are careful not to click on them on occasion, they remain (sadly?) essential to generate significant traffic. But a new player in the bustling news world is looking to change that. So, is it good or is it bad?
ChatGPT to the rescue!
Artifact has just arrived on our phones, but the application is already flowing ink. Dreamed up by Instagram dads, it promises to propel print media into the 2020s, and there’s no shortage of innovative ideas. Recently, its developers launched a feature allowing users to report clickbait, those catchy headlines that sometimes (often?) exaggerate the facts of the article concerned, if not worse.
But it’s 2023, and it seems like no developer can release anything without sprinkling it with a bit of generative AI. Here, it is of course GPT-4 that comes into play. While until now, Artifact was content to report the clickbait to its readers, or even to give less visibility to the article concerned, the application now seeks to modify it.
Thus, as soon as a title is reported by a user, the OpenAI tool comes into action and generates a title more in line with the content of the article. GPT-4 is particularly good at identifying key points and summarizing short or long texts, so it is not surprising that it is used in this way.
To make the process smoother, Artifact developers are looking to automate clickbait detection without relying on manual reporting. Once this system is in place, the application should automatically detect and rewrite the titles, which is not without its problems.
From the clickbait to the title without any relation to the subject?
ChatGPT doesn’t have the best reputation in the editorial world. While many fear that it will bog down the work of many editors and journalists, the reality for the moment is quite different. Indeed, as far as we know, the tool has been widely used to create news farms with average quality content. He is also very good at creating clickbaits and fake news. So, can we really trust him to rewrite titles, without the agreement of their authors? Even, at best, with human moderation?
Especially since the titles chosen by news sites and newspapers remain editorial choices, whether they are optimized for SEO or not. Before reading these lines, your first connection with this article, and perhaps even with Clubic, was a title that the editorial team deliberately chose. This raises the question of the relevance of what Artifact has just put in place, and of its legitimacy to operate what are similar to editorial choices.
For Kevin Systrom, one of the app’s founders, that doesn’t make much of a difference. In an interview with TechCrunchhe declares : ” In fact, building the algorithm is an extremely editorial task. Because what you choose to train your algorithm – the objective function, the data you introduce, the data you include, the data you don’t include – is editorial judgment. »
Whether the competition will follow remains to be seen. It would not be surprising to see, for example, the developers of Google News seeking the help of Bard, the in-house generative AI. From there, will this new approach concerning the titles radically change the written press?
Source : TechCrunch
9