A new AI tool by by Google’s Jigsaw to help journalists detect manipulated images: Demonstrating the idea of “mixed teams”

According to the New York Times, a new tool to help journalists identify “doctored images”, has recently been announced by Jigsaw. Disinformation comes at the top of the list as a threat to “international security” which is relevant to the Google owned company’s main mission and hence motivated the creation of their new AI tool Assembler to aid journalists in their tasks.

Using machine learning, this tool combines “multiple image manipulation detectors” together in one AI tool. Each of these detectors on its own can identify one type of manipulation techniques – for instance whether something “was deleted from the background” of the image. Accordingly, combining different detectors in one tool gives a further comprehensive feedback regarding whether an image has been altered.

In accordance with the discussions that took place in the OI2 Seminar 2020: The Challenge of Artificial Intelligence in Audio-visual Media, the development process of Assembler and its purpose shed light on the importance of adopting the idea of “mixed teams”. This concept was highlighted by Cristina Pulido, Serra Hunter professor in the department of Journalism and Communication Science in Universidad Autónoma de Barcelona (UAB), and confirmed by Laura Cervi, the postdoctoral researcher in the same department, UAB during the round table discussions.

First, the idea can refer to mixed teams between communication and engineering researchers. Assembler demonstrates a collective effort among researchers and engineers to serve journalists and fact-checkers. Researchers from “the University of Maryland, University Federico II of Naples, and the University of California, Berkeley each contributed detection models”

Jared Cohen, CEO and founder of Jigsaw, wrote on Jigsaw’s blog:  “Assembler is an early stage experimental platform advancing new detection technology to help fact-checkers and journalists identify manipulated media. In addition, the platform creates a space where we can collaborate with other researchers who are developing detection technology.”

Second, mixed teams in terms of promoting the idea of AI tools/robots working side by side with journalists to support them in tedious time consuming tasks. Today, quality journalism is most difficult to attain and maintain with the challenges journalists face; manipulated content widely and rapidly spread together with time pressures, which lead journalists to choose immediacy of breaking the news at the expense of thorough verification. This is where AI comes in to “humanize” the profession, as emphasized by Pilar Bernat, Professor of New Technologies in Universidad de Nebrija and David Llorent, founder and CEO of Narrativa, a company dedicated to the automatic generation of content in the Seminar 2020 round tables.

Cohen emphasizes the importance of journalists and robots working together by mentioning that Assembler was built “to help provide journalists and fact-checkers with strong signals that, combined with their expertise, can help them judge if and where an image has been manipulated”.

Assembler is one more real life example confirming the importance of integrating AI into journalism progressively.