artificial intelligence at the service of inclusive language. This is the idea that the integrated communications agency Take created Inclusive Talk, a tool based on Artificial Intelligence (AI), with the aim of promoting texts that are more respectful of women, LGBTQIA+ people, Afro-descendants and people with disabilities, just to name a few. This tool, born in Italy and open to any language, is similar to ChatGPT, but integrates guidelines on diversity and inclusion (DEI) with millions of documents present in the Large Language Model (LLM), an algorithm capable of analysing, generating , summarize, translate and even predict content.
Collective knowledge
Inclusive Talk uses collective knowledge to transform a text, eliminating obsolete terms, misunderstandings and harmful prejudices. It doesn’t just replace offensive words, it interprets the context.
If we write «All employees are required to comply with the company code of conduct», the program suggests «All people employed are required to comply with the company code of conduct». This means recognizing the so-called “overextended masculine”, i.e. the masculine agreement when masculine and feminine terms are present. The tool provides detailed “reasons for changes”, explaining the corrections made. For example, the replacement of “employees” with “employed persons” aims to make the text more inclusive with respect to employment status.
«I replaced “everyone” with “all” – we read – to also include female people. I kept the sentence unchanged because it complies with DEI guidelines and does not contain offensive or discriminatory language.” Inclusive Talk finally, it proposes possible terms to correct, such as “dependent” in this case, because “they could have a strong gender value”. And it warns the user: “Make sure you are using the most appropriate gender form.”
Tests and improvements
We tested Inclusive Talk and obviously it can be perfected because the tool continuously learns, fueled by new information and studies in the sector. The more it is used (and therefore “trained”) the more it improves, in a learning by doing logic, that is, it learns by doing. This is also why the Take agency offers anyone the opportunity to experience it at the address takegroup.it/inclusive-talk.
Just like us, AI can also face problems that are difficult to solve. In these cases, sensitivity, common sense and a critical spirit come to the rescue: it is human intelligence, which allows us to evaluate the result obtained and its adequacy to our expectations.
He is aware of it Laura Lopez, Project Manager at Take: «We want to stimulate collective reflection on the impact of inclusive language. The goal is that over time it becomes a real virtual assistant for us and for our client companies. A checker of the way we communicate and develop communication campaigns, which must be free from stereotypes and prejudices.”
Do you want to tell us your story? Write to Diversity Editor (Click here). Also on Facebook (Click here), Instagram (Click here), Twitter (Click here), LinkedIn (Click here).