By Jagpreet Kaur Maker, Borealis Blog editor
Science communication plays a pivotal role in bridging the gap between scientific advancements and the public. In recent years, the integration of artificial intelligence (AI) into research, education, and communication activities has revolutionized various aspects of society, including science communication. In Canada, a nation known for its rich scientific contributions, AI is having an increasing impact on the creation, communication, and understanding of scientific information.
But before we delve into these complex issues, let’s look at what artificial intelligence is.
What is artificial intelligence?
AI is a fascinating field that blends computer science and creativity, shaping the future of how machines interact with the world. It uses massive computing capacity to interrogate large, robust data sets. Based on in-depth analyses of these datasets, AI can make predictions about complex systems and synthesize coherent, intelligible text, audio (voices), and images. AI uses iterative processing to create an “intellectual framework” that helps it better understand and work with new data input.
AI is already part of our lives. From virtual assistants like Siri to virtual fitting rooms, it is making our lives easier and more efficient.
AI-Powered content creation
Content creators use a combination of traditional and AI tools to create appealing content. Image: Maddie Schultz.
One of the notable impacts of AI on science communication is the emergence of AI-powered content creation tools. These tools generate coherent and contextually relevant content using natural language processing algorithms. Canadian scientific communicators, including researchers, social media managers, academics, and students, have recognized the potential of AI for enhancing content creation. Science communicators have begun to use AI driven platforms to produce articles, press releases, and other informational materials. Content creators like ASAPScience use a combination of traditional and AI tools to create content on topics ranging from the periodic table to vaccine development.
OpenAI’s GPT-3, a recently launched platform, is helping scientists craft accessible and engaging content by increasing efficiency in processes such as analyzing and interpreting literature and improving writing styles. A simpler version of this AI, ChatGPT, can help writers select topics, conduct literature searches, organize outlines, add details to subtopics, and improve writing style. Additionally, for non-native English speakers, ChatGPT can promote better grammar, sentence structure, vocabulary selection, and even translation.
Using AI to personalize content
AI has also played a crucial role in tailoring online content to individual preferences, allowing personalized content delivery. Various platforms and applications use AI algorithms to analyze user data and provide customized scientific content that appeals to them. These data may include the visitor’s location, topics of previous searches, or demographics (if available) to create more targeted content. This approach allows the content to address the diverse interests and knowledge levels of the Canadian population.
The use of AI for personalized content has gone beyond the online frontier. ExhibitXplorer is a service designed to revolutionize museum experiences by integrating advanced information such as geofencing, artificial intelligence, and microservices with the visitor’s interests and preferences to create personalized content. By analyzing visitors’ behavior and self-identifications, ExhibitXplorer tailors content to their demographic—researcher, student, casual visitor, or child.
Using contextual geofencing and AI, ExhibitXplorer proactively generates personalized content as visitors approach exhibits. Push notifications are employed to deliver personalized content when visitors enter specific areas or geofences. For example, ExhibitXplorer will describe the Mona Lisa differently depending on the person viewing it. For a visually impaired visitor, the Mona Lisa is described as a portrait of a woman with an enigmatic smile, delicate brushstrokes, and subtle shading, seated against a backdrop of rivers and hills. For a student, it is described as a famous painting by da Vinci, depicting a woman with a mysterious smile, wearing a dark dress and headscarf.
This technology has not yet been used for science communication or in science museums but has the potential to make science exhibits more accessible to different audiences.
Disadvantages of AI use for writing
Articles written by AI present several disadvantages—limited creativity, understanding of context and tone, grasp of nuances and emotions, and research and fact-checking. AI lacks the ability to produce truly original content, often resulting in repetitive or plagiarized material. AI also struggles to comprehend the subtleties of language and emotion, leading to inaccuracies and inappropriate responses.
The advent of AI in traditional journalism has impacted science communication in Canada. Automated journalism, where AI algorithms generate news articles based on data and information inputs, has become more prevalent. While this enhances the speed at which scientific news is disseminated, it raises questions about the role of human journalists in interpreting and contextualizing complex scientific findings. Achieving a balance between AI and human input in content creation is essential, as human creativity and expertise remain indispensable in delivering compelling and impactful writing.
Challenges in ethical AI use
While the integration of AI into science communication brings numerous benefits, it also raises ethical concerns that cannot be ignored. Canada, which is at the forefront of AI research and development, faces the challenge of ensuring the responsible and ethical use of these technologies. Issues such as bias in AI algorithms, data ownership, data privacy, and the potential for misuse must be carefully addressed to maintain the trust and confidence of the Canadian public.
AI may use biased input data which hinders the development of diverse perspectives. Image: IE Foundation.
AI may perpetuate biases (both positive and negative) present in its training data, hindering the development of diverse perspectives. A recent example is Google’s AI model, Gemini. Ironically, in a bid to be inclusive, the AI model created historically inaccurate images—ethnically diverse World War II Nazi soldiers and Vikings. Following criticism on social media, Google temporarily halted its AI model from creating images of people while it works to improve its accuracy and bias issues.
As previously stated, AI trains itself by ingesting large datasets. But where do these data come from and who owns them? Thousands of authors, including Margaret Atwood, have demanded compensation from AI companies for using their copyrighted work as AI training datasets. They argued that AI systems incorporating literature without compensation threatens authors’ livelihoods and rights. They wrote an open letter calling for fair compensation for authors whose work is used by AI systems, highlighting the importance of respecting copyright laws, and acknowledging authors’ contributions.
In their report, “Leaps and Boundaries”, the Expert Panel on Artificial Intelligence for Science and Engineering from the Council of Canadian Academies states that AI holds the potential to enhance reproducibility in science, but currently faces challenges due to a lack of transparency in sharing code and data, eroding trust in the accuracy of results. Ethical considerations arise throughout the research process concerning data collection, AI model design, result dissemination, and long-term data management.
Implementation of data stewardship is essential for responsible and ethical data sharing and use, particularly concerning datasets involving individuals’ information shared without their consent. Such data managements frameworks and principles include the FAIR data principles (Findability, Accessibility, Interoperability, and Reuse of digital assets) and TRUST (Transparency, Responsibility, User focus, Sustainability, and Technology).
The impact of AI on science communication in Canada is multifaceted, bringing both opportunities and challenges. AI-powered content creation tools are reshaping the way scientific information is collected, analyzed, and conveyed to the public. The emergence of AI-powered tools has facilitated the production of coherent and contextually relevant scientific content, enhancing accessibility for the public. However, ethical considerations loom large, requiring careful attention to issues such as bias, privacy, and responsible use. These ethical considerations and the preservation of human journalistic values must be at the forefront of AI technological developments. As Canada navigates this evolving landscape, striking a balance between AI-driven innovation and human expertise will be crucial for ensuring the integrity and trustworthiness of science communication.
Feature image: Artificial intelligence is impacting how information is communicated across different fields. Image: Carlow University.
