With the "Deus Ex Mach­ina? — Test­ing AI Tools"-series we want to show you dif­fer­ent tools, that aim to sim­pli­fy writ­ing, design and research by using Arti­fi­cial Intel­li­gence. More on the "Deus Ex Machina?"-series can be found here.

Overview

"AI-powered intel­li­gence for journ­al­ists" is what the AI tool JECT.AI is sup­posed to offer. JECT.AI prom­ises to help users broaden their hori­zons, adopt new per­spect­ives and thus cre­ate more diverse and high­er-qual­ity con­tent — espe­cially in the field of sci­ence journalism.

To this end, JECT.AI provides users with vari­ous applic­a­tion options. Its core ele­ment is an AI-based search engine that sug­gests art­icles, journ­al­ists and sci­ent­ists for the sci­ence top­ics users searched for, thus provid­ing a broad and detailed over­view of cur­rent research and report­ing on this top­ic. Five fil­ters allow users to tail­or the res­ults to their own needs. In addi­tion to gen­er­al con­tent, users can search spe­cific­ally for facts and fig­ures, cent­ral char­ac­ters, back­ground inform­a­tion and devel­op­ments. For example, if you enter the term "plan­et­ary bound­ar­ies", the search engine com­piles the fol­low­ing res­ults, graded accord­ing to related art­icles and sub­ject areas, sci­ent­ists and journ­al­ists, as well as pos­sible com­bin­a­tions of top­ics. These res­ults can be fur­ther tailored, for example, if a spe­cial focus is to be placed on facts and figures.

In addi­tion to the search engine, JECT.AI offers its users five oth­er tools to help com­mu­nic­ate sci­ence bet­ter. Under "Sci­ence Audi­ences", for example, the tool bundles exem­plary address­ees of sci­ence com­mu­nic­a­tion, which journ­al­ists can use as pro­to­types for their read­ers. In addi­tion, com­mon meta­phors that reappear in one of the three main sci­ence fields that JECT.AI focuses on, storytelling roles, explan­a­tions of how sci­ence works and indic­at­ors of good sci­ence journ­al­ism can be found among these tools. They are all stat­ic con­tent and have not been fur­ther mod­i­fied since the release of JECT.AI.

To use JECT.AI, all you need to do is register using your email address and a pass­word. The basic ver­sion is free of charge, but the com­pany offers addi­tion­al ser­vices and exten­sions on request, which are sub­ject to a fee.

The AI behind the application

JECT.AI emerged as a tool from the EU-fun­ded pro­ject: QUEST (QUal­ity and Effect­ive­ness in Sci­ence and Tech­no­logy com­mu­nic­a­tion) and is now run as a start-up by its developers Prof. Neil Maid­en and Dr. Kon­stanti­nos Zachos. The aim of QUEST — which phased out in sum­mer 2021 — was to improve the quant­ity and qual­ity of sci­ence journ­al­ism, par­tic­u­larly in the three focus areas of cli­mate change, vac­cin­a­tions and arti­fi­cial intelligence.

Accord­ing to the developers, JECT.AI was developed togeth­er with journ­al­ists and adap­ted again and again to fit their needs. Every day, the AI reads more than 1000 newly pub­lished art­icles from over 400 sources (papers, journ­als, news­pa­pers, etc.) in six European lan­guages (as of 2021) and exam­ines them for their sci­entif­ic con­tent. Diversity plays an import­ant role in the AI algorithm. It is trained to present the broad­est pos­sible spec­trum of res­ults. If, for example, 20 sci­ent­ists are out­put who are research­ing a spe­cif­ic area of a top­ic, JECT.AI ensures that at least 10 of them are FLINTA* (women, les­bi­ans, inter­sex, non-bin­ary, trans and agender per­sons). In oth­er words, belong­ing to groups that are under­rep­res­en­ted in main­stream soci­ety and academia.

How­ever, the com­pany does not provide any inform­a­tion on which data set the search engine accesses or wheth­er it has been optim­ised again since its pub­lic­a­tion in 2021. The AI on which the search algorithm is based and the data used to train it is also not dis­closed on the JECT.AI website.

The rhetorical potential of the tool

JECT.AI is aimed at journ­al­ists who are sup­posed to write about sci­ence but have no in-depth sci­entif­ic expert­ise. The tool is designed to help them cre­ate high-qual­ity con­tent and make sci­ence com­mu­nic­a­tion more access­ible. It helps to find inform­a­tion, know­ledge and con­tacts that facil­it­ate the pro­cess of cre­at­ing inform­at­ive art­icles on sci­ence topics.

Rhet­or­ic­ally, besides the search engine, which is rel­ev­ant for find­ing top­ics and inform­a­tion, the pre­pared fea­tures are par­tic­u­larly inter­est­ing. The per­so­nas sum­mar­ised under "Sci­ence Audi­ences", for example, which rep­res­ent the address­ees of sci­ence com­mu­nic­a­tion, their atti­tudes and needs, are inten­ded to help journ­al­ists write more spe­cific­ally for these groups of address­ees. These per­so­nas indic­ate, for example, which sci­entif­ic know­ledge, pre­ferred media or forms of inter­ac­tion apply to cer­tain tar­get groups. Journ­al­ists can take these per­so­nas as a mod­el and adapt their writ­ing to them so that their com­mu­nic­a­tion meas­ures are as effect­ive as pos­sible for the inten­ded tar­get group.

The pre­pared meta­phors and storytelling sug­ges­tions also help to build tar­geted lin­guist­ic images and nar­rat­ives that are sus­tain­able and con­vin­cingly con­vey the sci­entif­ic content.

The search engine can be used to fil­ter out com­mon topoi and nar­rat­ives on indi­vidu­al top­ics. This fea­ture can be instru­ment­al­ized to make one's own journ­al­ist­ic con­tri­bu­tions more inter­est­ing and impress­ive through new per­spect­ives and more diverse points of view. The pos­sib­il­ity to get in touch with oth­er journ­al­ists and sci­ent­ists offers users of JECT.AI the chance to col­lect and sub­sequently com­mu­nic­ate inform­a­tion of high­er qual­ity through exchange. This increases the qual­ity of the con­tent and the logic­al per­suas­ive­ness of the con­tri­bu­tions. Espe­cially in sub­ject areas that are less access­ible to address­ees and require a great deal of expert­ise (as is usu­ally the case in sci­ence), the author­ity of an expert — such as a sci­ent­ist — can cre­ate trust.

How­ever, the rhet­or­ic­al poten­tial of JECT.AI remains lim­ited out­side of the search engine func­tion. The built-in fea­tures offer per­so­nas, meta­phors and nar­rat­ives that cov­er the three sci­entif­ic fields of vac­cin­a­tion, cli­mate change and arti­fi­cial intel­li­gence at a bas­al level, but do not go deep­er. There is a lack of terms that are less com­mon and a lack of cur­rency. Since the tools release, these fea­tures have not been expan­ded, which means that changes in dis­course, such as new terms or nar­rat­ives, are not reflec­ted in it. Indi­vidu­ally adapt­ing and com­mu­nic­at­ing to a spe­cif­ic audi­ence at a spe­cif­ic time is a core ele­ment of any suc­cess­ful com­mu­nic­a­tion. Due to the lack of agil­ity and top­ic­al­ity of the tool, exactly this can­not be ensured. The algorithm also does not adapt to the user and some­times deliv­ers inap­pro­pri­ate res­ults. When search­ing for the keyword "Friendly AI", for example, the 12 most rel­ev­ant res­ults returned for this keyword ranged from new devel­op­ments in arti­fi­cial intel­li­gence research to an art­icle on football.

Usage in science communication

Basic­ally, JECT.AI is well suited for use in sci­ence com­mu­nic­a­tion. It can facil­it­ate research pro­cesses, estab­lish con­tacts and, in the pro­cess, ensure that sci­ence com­mu­nic­a­tion becomes a little more diverse and, above all, less pre­sup­posed. The­or­et­ic­ally, users do not even need a basic sci­entif­ic under­stand­ing of the tool's focus top­ics (AI, vac­cin­a­tions, cli­mate change), but can use JECT.AI to learn the basics of the neces­sary con­tent, as well as to get help with com­mon for­mu­la­tions with­in a them­at­ic dis­course. In the sci­entif­ic field of "cli­mate", for example, one finds the meta­phor of (rain)forests as the "lungs of the earth". JECT.AI explains the con­text of the meta­phor and how it can be used as accur­ately as pos­sible in an art­icle or post. At the same time, keywords related to a spe­cif­ic top­ic are also provided, which can be used to col­lect and dis­play fur­ther information.

The net­work­ing poten­tial for journ­al­ists and sci­ent­ists that the applic­a­tion holds is also quite extensive.

The lack of flex­ib­il­ity and breadth of the research areas covered (so far only the three focus top­ics men­tioned above are covered) lim­its the actu­al poten­tial of the tool. For young journ­al­ists and new­comers to sci­ence com­mu­nic­a­tion, how­ever, JECT.AI can be a good start­ing point — albeit exclus­ively for writ­ten sci­ence com­mu­nic­a­tion, since JECT.AI does not offer more than tex­tu­al con­tent sup­port. The tool com­pletely excludes cre­at­ive ele­ments and their com­mu­nic­at­ive poten­tial. The same applies to dif­fer­ent media and chan­nels of writ­ten sci­ence com­mu­nic­a­tion. There is no pos­sib­il­ity to tail­or the res­ults and fea­tures to wheth­er, for example, writ­ing is done for news­pa­pers, social media or blogs.

Wrap-up

The idea behind JECT.AI to sim­pli­fy the pro­duc­tion of sci­ence journ­al­ism, provide net­work­ing oppor­tun­it­ies and map diversity in research are use­ful start­ing points for sci­ence com­mu­nic­a­tion work. The tool offers resources and ideas that can help journ­al­ists write dif­fer­en­ti­ated texts, even if they do not have a sci­entif­ic back­ground them­selves or are new to a top­ic. The tool itself lacks pol­ish, flex­ib­il­ity, indi­vidu­al­ity adapt­ab­il­ity and con­stant fur­ther devel­op­ment of the tools with­in the tool and the algorithm. Nev­er­the­less, it is a good aid for research­ing and pro­du­cing sci­ence journ­al­ism, but still leaves the main work to the writers. It is suit­able for any­one who com­mu­nic­ates or is inter­ested in sci­ence, regard­less of age or level of knowledge.