As part of the QUEST project we have developed an AI-powered tool that helps journalists report about science more effectively, while leaving them in control of the writing process. You can explore some of the tool’s features.
[Note: JECT.AI is the new name for the tool previously called INQUEST]
Journalists, more than ever, are writing about science and topics related to it.
Science journalism is no longer the preserve of the science journalist. Because of the COVID19 pandemic, journalists are publishing stories about how a virus impacts on not only our health and our lives, but also on our politics and our economies.
However, most journalists lack tools and training to support them to write about science. The few existing tools that break down academic papers into bite-sized chucks for non-scientific audiences, or detect misleading scientific claims to influence public policy, are unlikely to be enough.
Therefore, to provide more support, QUEST has developed a new interactive tool to enable journalists without specialist science knowledge and training to write about science more effectively.
The new tool is called JECT.AI.
JECT.AI was designed to support journalists to write about science when their time and other resources, such as direct access to scientific experts, is limited. To do this, it uses different forms of automation to support journalists to undertake activities associated with good science journalism. Three of these activities are:
- Using more diverse scientific information sources to develop stories. As well as reading scientific papers to research and write stories, science journalists often use more diverse sources including science alerts, scientists’ tweets, science newsletters, and general news sources;
- Tailoring content to reach specific science journalism audiences, and the different styles and channels to reach audiences that are excluded from, disengaged with or only moderately interested in science news;
- Explaining science to these audiences, using different strategies, such as reporting the context and the background shown to be more effective.
JECT.AI provides automated support for these and other activities, but it still leaves the journalist in control of the writing process.
What JECT.AI can do
JECT.AI can automatically discover and present information from diverse sources to journalists related to their stories when these stories are being written. The sources include 1000s of peer-reviewed papers from over 50 science journals and magazines, 1000s of views expressed by 100s of science experts and journalists, and millions of stories published in 100s of newspapers. JECT.AI discovers this information at the press of a button, and presents them to a journalist in a common format in one place, as demonstrated below in the current JECT.AI prototype.
JECT.AI can also present journalists with different science audience personas. Each of these personas is a fictionalised reader with characteristics typical of researched science audience segments, such as sciencephiles, those critically interested in science, passive supporters of science, and people disengaged from science . Some also reflect people from ethnic minorities and with lower incomes .
JECT.AI presents each persona name, background, interests and roles. The current personas are designed to be used by journalists in different EU countries. Below shows the current JECT.AI prototype displaying some of its audience personas.
To automate explanation support, JECT.AI uses theories of the organization of text using predefined types of rhetorical relations – relation types such as background, description, circumstance and cause. Because journalists writing about science from different sources are seeking to develop stories that are connected and coherent, these relation types provide a valid start point for designing explanation guidance. JECT.AI automates this guidance based on relations such as context, example, motivation and cause.
Furthermore, JECT.AI testing by broadcast journalists revealed the need to automate support for a fourth activity, to support the use of metaphors to communicate science effectively. Metaphors are heuristic tools for science communication that enable non-scientists to learn knowledge that may be too complex or abstract to understand in its original form. Well-known metaphors include the greenhouse effect and herd immunity.
Therefore, a first set of metaphors, each with a title, image, description, and guidelines when to use the metaphor was implemented in JECT.AI. Each metaphor also comes with simple guidelines to encourage journalists to think about its limits, as outdated metaphors can limit scientific inquiry and cause public misunderstanding, and wrong language can re-enforce stereotypical thinking and political meaning. Below shows the JECT.AI prototype displaying some of the science communication metaphors.
A full version of JECT.AI will be available at the end of the summer. QUEST will be offering the prototype to journalists, to use and provide feedback on. If you are interested, contact us at firstname.lastname@example.org.
Using JECT.AI now
You can explore some of JECT.AI’s features interactively now here. Click on one of the two predefined options, related to COVID-19 or the climate crisis – and explore what JECT.AI discovers. Use the options above the results to prioritise content from different sources, and/or its periods of publication. And use the icons on the bottom left to explore the audience personas and science metaphors.
 John C. Besley. 2018. Audiences for Science Communication in the United States. Environmental Communication 12, 8, 1005-1022, DOI: 10.1080/17524032.2018.1457067
 Emily Dawson. 2018. Reimagining publics and (non) participation: Exploring exclusion from science communication through the experiences of low-income, minority ethnic groups. Public Understanding of Science 27, 7, 772-786. DOI: /10.1177/0963662517750072