The rapid development of generative models of artificial intelligence (AI) opens up new opportunities for journalism, but also creates new risks.
The increasingly widespread use of special software for the creation of texts and images is changing the media sphere and the work of journalists, as well as causing discussions about the ethical use of these tools. Technological progress requires revision of editorial practices and approaches to work.
The Commission on Journalistic Ethics notes new ethical problems that have arisen or may emerge against the backdrop of the growing popularity of contemporary AI models. The biggest debates arise around language models (ChatGPT, Bing, Bard, etc.). Some foreign media have already developed and are constantly updating the rules for its use in daily editorial practices — alongside the improvement of AI.
It is important that the policy of using artificial intelligence during the creation of materials should become part of the editorial policy of the Ukrainian media.
The Commission on Journalistic Ethics recommends that journalists and editors proceed from the following principles:
- Journalists should study the advantages and new opportunities offered by the new-generation artificial intelligence models. These models can not only simplify some editorial processes and help journalists in their work, but also change the media environment. Media professionals need to be prepared to adapt to these changes while remaining committed to their professional ethical standards.
- It is necessary to take into account the limitations of generative AI models. They may create texts that are very similar to such texts in other media. However, they are not able to distinguish truth from lies, plagiarism from the original text; they are not able to check facts, think critically, and avoid prejudices.
- The responsibility for journalistic material rests with the author and the editors, even if the author sought help from AI during the preparation of this material. The use of artificial intelligence does not exempt journalists from the obligation to verify facts in reliable sources, check hyperlinks to AI-generated sources to make sure they exist and are indeed relevant to the subject matter, be accurate in citations, provide context, clearly separate facts from assessments, maintain a balance of opinions, and avoid discrimination. If necessary, admit and correct mistakes.
- There is no uniform approach in foreign media that have already implemented a policy of using artificial intelligence. In some newsrooms, journalists are allowed to use AI to generate ideas or headlines, but not texts. In other media, staff can experiment more liberally with generative AI models as part of article preparation. However, in all cases, the final decision on the distribution of material is made by a person, not a machine. The Commission on Journalistic Ethics considers it unacceptable when AI-generated materials appear in the media automatically, without any verification or editing.
- The audience should be informed of materials (texts, infographics, static and dynamic images, videos etc.) where authors used AI for partial or full generation of the text, graphic materials, videos. The explanation can look, for instance, like this: “This material is partially written with the use of artificial intelligence followed by a thorough verification of all facts by a journalist.” Or “The text of this material has been generated by AI and edited by a newsroom journalist.” Or: “Photo / image generated by AI for the purpose of protection and security of the source.”
- The Commission advises against the use of AI-generated images to illustrate current news, even if there is a disclaimer. Implementation of algorithms should be done cautiously when it comes to selecting headlines or videos, avoiding manipulative techniques to drive traffic. For instance, machine-generated headlines containing “hate speech” provide more exposure.
- AI-generated text that is part of the journalistic material must be checked for uniqueness. Generative models can lift ideas and entire phrases from authors, and journalists who use this text may come under fire for plagiarism.
- The models and systems we create and teach are a reflection of ourselves. It is therefore unsurprising that AI may also contain gender stereotypes. According to Harvard Business Review, there have been multiple cases of AI picking up gender bias from people. Harvard Business Review cites the example of natural language processing (NLP), which is present in Amazon’s Alexa and Apple’s Siri. Therefore, journalists and editors should also check the texts generated by AI for gender stereotypes and discriminatory statements.
- Media professionals should avoid painting a picture of AI as a reputable source or the ultimate expert (headlines such as “What ChatGPT thinks about…”). Such materials are for entertainment only and cannot replace fact-checking or evidence-based reasoning.
- Journalists should pay more attention to uncovering fakes and manipulations, which have reached a new level due to the use of AI. This is particularly important for Ukraine, which is undergoing a full-scale Russian invasion and attacks by Russian propaganda.
- During a dialogue between a journalist and a generative model, there is a danger that sensitive data may be leaked. Journalists cannot know how their request is processed, who has access to the data contained in the request, and how that data will be stored in the future. Therefore, the Commission advises against sharing any information that contains personal data with AI — for example, personal information of people you are writing about.
Useful experience of foreign media:
What Reuters Is Telling Its Journalists About Using Artificial Intelligence
Open, Responsible, Transparent — DPA’s Recommendations on AI (in German)
The ethics of using generative AI to create journalism: What we know so far
AP https://blog.ap.org/standards-around-generative-ai
BBC https://www.bbc.co.uk/mediacentre/articles/2023/generative-ai-at-the-bbc/
CNET https://www.cnet.com/ai-policy/
WIRED https://www.wired.com/about/generative-ai-policy/
The Guardian https://www.theguardian.com/help/insideguardian/2023/jun/16/the-guardians-approach-to-generative-ai
A global survey of what news organizations are doing with AI
AI and gender equality:
Artificial intelligence and gender equality: key findings of UNESCO’s Global Dialogue https://unesdoc.unesco.org/ark:/48223/pf0000374174
The need to represent: How AI can help counter gender disparity in the news https://blogs.lse.ac.uk/polis/2022/03/16/the-need-to-represent-how-ai-can-help-counter-gender-disparity-in-the-news/
AI to tackle gender bias in media https://worldpressinstitute.org/ai-to-tackle-gender-bias-in-media-journalism/