WSJ’s Doppler turns millions of social videos and podcasts into election stories
Cartoon · AI-generated
The Wall Street Journal is using its Doppler AI system to analyse more than 25 million TikTok videos, 500,000 podcast episodes and other media for political themes traditional polling may miss.
The Wall Street Journal has begun publishing stories informed by Doppler, a custom AI system it developed to broaden its midterm election coverage. According to Andrew Deck of Nieman Lab, the tool has already processed more than 25 million TikTok videos, about 500,000 podcast episodes and thousands of Instagram clips and cable news segments, giving the newsroom a way to track political discussion far beyond the usual polling sample. Aja Whitaker-Moore, the Journal’s deputy editor-in-chief, said the aim is not to forecast the outcome of the election, but to map the themes running through conversations about politics as they surface in culture, work, money, family and community.
The Journal’s approach reflects a wider shift in newsroom use of artificial intelligence, but with a more ambitious scope than the transcription and summarisation tools adopted elsewhere. Rather than simply helping reporters sift through material faster, Doppler is being used to conduct large-scale analysis and turn those findings into publishable journalism. Nieman Lab reported that the paper intends to use the tool to surface voter concerns, priorities and sentiments that more traditional surveys may miss, then fold those insights into its election coverage ahead of Election Day.
Industry summaries of the system suggest Doppler relies on embedding models that convert text and images into numerical vectors, making it possible to group content into thematic clusters and identify patterns across vast amounts of online material. The Journal has also signalled that the project may not remain confined to politics. According to one report, editors are considering whether the same analytical framework could be extended to other desks, which would make Doppler a test case for how far AI can be woven into reporting without replacing the judgement of journalists themselves.
That distinction remains central to how the Journal has presented the initiative. In a letter to staff dated 21 September, editor-in-chief Emma Tucker reportedly stressed that the system is intended to assist reporters rather than substitute for them, with journalists still responsible for deciding which questions to ask, checking material and making editorial calls. The broader coverage plan also includes on-the-ground reporting, suggesting the paper is trying to pair machine-driven pattern spotting with old-fashioned field reporting rather than treat the technology as a stand-alone news engine.
Inspired by headline at: [1]
Sources by paragraph: - Paragraph 1: [2], [6] - Paragraph 2: [2] - Paragraph 3: [3], [4] - Paragraph 4: [5]
Source: Noah Wire Services · Original