ChatGPT: Augmenting PR or Replacing Experts?
The evolution of content creation has accelerated with ChatGPT in recent years. This AI-powered chatbot generates content across multiple formats with simple prompts, sparking debates about its impact on jobs and productivity, particularly in the public relations sector. This transition has recently taken a significant step forward. Nowadays, articles, PowerPoint presentations and research material can be easily accessed by simply using a text prompt in a chatbot interface.
How Does Generative AI Work?
Chatbots as a concept have been around for a long time. Eliza, the first program of its kind, was created in 1966 by MIT computer scientist Joseph Weizenbaum to mimic a conversation with a psychotherapist. Surprisingly, the program elicited emotional responses from participants as they interacted with the script. This observed phenomenon gave rise to the term ‘Eliza effect’; which refers to the unconscious tendency of people to associate computer actions with human-like behaviour.
Since then, many generative AI tools have been released for non-profit applications, such as ChatGPT, and for commercial use — Google BARD and Google BERT. Many of these large language models have the ability to create content — such as blog posts, articles, and social media posts — product descriptions and reviews, summarise and rewrite text, translate and proofread content.
Chat GPT-3 was trained on a vast collection of sentences and data to predict subsequent words within a sentence. According to OpenAI, this unsupervised learning process used around 570 GB of data from books, web texts, Wikipedia, articles, and other online resources. The system has 175 billion parameters and has been trained on 300 billion words as input. This training enables the model to generate text that is both coherent and contextually relevant. Although OpenAI has not revealed the inner workings of Chat GPT-4, experts are now thinking that a similar approach has been used.
An academic study led by Google (2) provided a detailed explanation of the mechanisms used by a typical Transformer. Essentially, the language model relies on ‘tokens’ for understanding and generating text. A token can be as short as a single character or as long as a whole word. In this way, the model interprets the user’s text input, by breaking it down into tokens and generating word sequences that anticipate the most accurate response based on its training data.
AI Challenges in Corporate Communications
While ChatGPT has demonstrated an impressive ability to generate human-like text, discussions and predictions have emerged about the impact of ChatGPT on the workforce. According to McKinsey’s research4, generative AI and other technologies could automate 60–70% of tasks, particularly those using natural language, which account for 25% of work time. This technology, in particular, will improve knowledge-based jobs, with a substantial impact on roles that require higher levels of education and compensation.
Meanwhile, the United States and the European Union have advocated for stricter regulations on generative AI, emphasising the need for transparency and data protection. The challenges of integrating technologies such as ChatGPT into corporate communications (5) are therefore more significant. This is an area that requires accuracy, comprehensive research, and appropriate tone of voice, underscoring the hurdles posed by potential data privacy breaches4. Users may unknowingly disclose sensitive information to chatbots, which raises concerns about the storage and reuse of information and how it may feed into ChatGPT’s learning database. This uncertainty extends to third-party developers who might have access to private data.
The reliability of ChatGPT’s outputs is also under scrutiny. Since the AI’s knowledge is derived from pre-existing data patterns, it can inadvertently produce inaccurate or misleading information, necessitating rigorous fact-checking, especially for data-centric analysis. Samantha Floreani, a digital rights activist, raises concerns in The Guardian online, about the potential degradation of journalistic standards due to the proliferation of AI-generated content. The prospect of “mutant news,” arising from AI models training on each other’s outputs, underscores the critical need for vigilance and ethical responsibility in leveraging AI for content creation.
ChatGPT’s dependency on historical data limits its ability to process real-time information or private internal company details, resulting in responses that may be outdated or missing vital context. There’s also the added risk of plagiarism, as AI might generate content that closely mirrors copyrighted material without clear attribution.
Inherent biases in the AI’s training material may colour its responses, reflecting the biases embedded in its source texts. In addition, ChatGPT’s tendency towards verbosity and repetition can undermine the authenticity and precision of its responses, often lacking the emotional resonance and personal touch typical of human communication.
How Companies are using Generative AI
A survey conducted by the Boston Consulting Group (5) shows cautious optimism among executives about AI, with the majority predicting that it will take at least two years to get “beyond the hype”. Despite the slow adoption, with only around 12% of American companies claiming to be using AI, the landscape of generative AI in business is rich and varied, signalling a turning point in the way organisations approach innovation and efficiency.
For example, Pfizer, the world’s largest pharmaceutical company, has launched Charlie, a generative AI platform named after its co-founder to significantly expand Pfizer’s content creation capabilities and communications efficiency. The platform is designed to generate digital content, draft medical articles tailored to specific audiences, as well as fact-checking and legal review. Charlie is integrated with media analytics for the company’s brands, insights on key competitors, and data from various websites. According to Bill Worple (6), Pfizer’s VP of Customer Engagement Platforms and Technology, “The whole idea is how do we speed up our content creation to actually create messaging that resonates both with the healthcare providers as well as our patients”.
Generative AI also has great potential to transform the banking sector, as evidenced by JPMorgan Chase’s deployment of over 300 AI use cases. Similarly, Capgemini’s initiative to develop over 500 industry use cases using Google Cloud’s generative AI, and Bayer’s claim of over 700 use cases, highlight the broad applicability and scalability of generative AI solutions.
According to Forbes (7), publications such as the New York Times, Guardian, BBC, Bloomberg and Reuters are a part of a growing list of media outlets that have used generative AI to produce articles. In addition, Netflix and Google are using the technology to speed up content creation, from personalised movie trailers to automated news reports.
This wave of innovation is being fuelled by a new breed of toolmakers, such as Synthesia and Writesonic, which allow for the creation of video and written content with unparalleled efficiency and customisation. In addition, popular writing assistants, including Grammarly, Jasper, Hemingway App, and ProWritingAid have been also transformed by AI. These tools have progressed beyond basic grammar and style checks, now including capabilities that adapt to and even mimic users’ individual writing styles.
AI’s impact on PR
The contemporary landscape of public relations is undergoing a modest but expected move toward integrating artificial intelligence9. A Prowly poll 8 of 303 PR professionals in May 2023 found a positive sentiment, with 63% in favour of incorporating AI into their workflow, indicating a preparedness to manage the digital revolution.
According to the Chartered Institute of Public Relations (CIPR)’s comprehensive research (9) “Humans Needed More Than Ever”, AI currently supports approximately 40% of PR tasks, enhancing efficiency in areas such as transcribing, media monitoring, and press release dissemination. AI help for diverse jobs spans from 20% to 60%, with data analytics and social media management receiving the highest average support from AI technologies (53.4% and 53.7%, respectively). In comparison, partnership management receives much less AI support (an average of 13.4%). Notably, no task has been fully automated by AI.
Another CIPR discussion paper on the potential of AI for the PR practice (10) identified five key technology categories: simplification tools that streamline PR processes; media and social media listening and monitoring tools; automation of routine tasks; AI applications for structured data analysis; and AI for unstructured data interpretation. These advancements will allow for more in-depth analysis of vast amounts of data, leading to informed decision-making.
For instance, AI can quickly analyse extensive comments from surveys and social media, identifying themes within hours. Such technologies also have the potential for real-time sentiment analysis, which borrows techniques from media and advertising to personalise the content based on audience reactions. AI and automation can also help in measuring communications effectiveness, such as designing questions, monitoring and interpreting data, and linking communications results to larger corporate objectives. While technology is important in the advancement of communications practices, a human touch to establish goals and match them with corporate plans is still necessary.
The Road Ahead: Faster to Crafting
Augmenting PR or replacing experts? Sam Garg, the CEO of Writesonic, a generative content creation platform, says, “The idea is to augment humans rather than replace them.”
One certainly cannot rely on ChatGPT to deliver the final product. But when used as a tool to overcome writer’s block and generate fresh ideas, it can work wonders. According to Daniel Jörg, Chief Innovations Officer at Farner Communications in Zurich, says: “ChatGPT can get us to the crafting stage faster, which is where all the magic of communication lies.”
Instead of ChatGPT being seen as a threat to communications professionals, it should rather be considered as an enhancement to their work. No one criticises an accountant’s decision to use Excel formulas instead of manually calculating hundreds of lines. So, why should the situation be any different for communications professionals using ChatGPT? By delegating routine tasks to AI, professionals can now focus on the strategic, creative and higher-level facets of public relations. In critical situations, such as crises, the human touch is irreplaceable. AI cannot replicate the depth of relationships that seasoned PR professionals build with senior management and the media, nor can it match the business acumen of a seasoned professional.
Key Takeaways:
- You can never be sure that what you receive from ChatGPT is original or accurate.
- Building relationships, understanding human emotions and empathising with the audiences are all important components of public relations. While they can increase efficiency, machines are still unable to fully understand and mimic human emotions. It cannot authentically engage with audiences or relate to journalists and senior management.
- Humans have a natural capacity for creativity, which is influenced by our emotions, experiences and imagination. It requires a degree of creativity and uniqueness that artificial intelligence cannot match.
References
(1) Scientific Diagram, Research Gate
https://www.researchgate.net/figure/Example-of-ELIZA-ELIZA-a-chatbot-was-designed-by-Joseph-Weizenbaum-to-imitate-a_fig1_348306833
Academic paper from research scientists at OpenAI.’ Language Models are Few-Shot Learners’
https://proceedings.neurips.cc/paper/2020/file/1457c0d6bfcb4967418bfb8ac142f64a-Paper.pdf
(2) Academic paper from research scientists at Google. ‘Attention is all you need’ https://proceedings.neurips.cc/paper_files/paper/2017/file/3f5ee243547dee91fbd053c1c4a845aa-Paper.pdf
Stephen Wolfram. ‘What Is ChatGPT Doing and Why Does It Work?’
writings.stephenwolfram.com/2023/02/what-is-chatgpt-doing-and-why-does-it-work/
Open AI. GPTs are GPTs: An early look at the labour market impact potential of large language models https://openai.com/research/gpts-are-gpts
(3) McKinsey, Comparative Industry Service, Oxford Economics. The economic potential of generative AI, June 2023, McKinsey
(4) Academic research from Partha Pratim Ray. ‘ChatGPT: A comprehensive review on background, applications, key challenges, bias, ethics, limitations and future scope’, Science Direct https://www.sciencedirect.com/science/article/pii/S266734522300024X
Gina Nowicki. The Impact of ChatGPT on PR: Our Tests & Outlook https://prowly.com/magazine/chatgpt-for-pr
(5) The Economist. How businesses are actually using generative AI
https://www.economist.com/business/2024/02/29/how-businesses-are-actually-using-generative-ai
(6) Marty Swant. With ‘Charlie,’ Pfizer is building a new generative AI platform for pharma marketing, Digiday https://digiday.com/marketing/with-charlie-pfizer-is-building-a-new-generative-ai-platform-for-pharma-marketing/
(7) Forbes. Generative AI and the future of content creation https://www.forbes.com/sites/bernardmarr/2023/11/30/generative-ai-and-the-future-of-content-creation/
Academic research from Xinyi Xu. A study on the application of Chat-GPT in media production and communications. Proceedings of the International Conference on Global Politics and Socio-Humanities, Research Gate https://www.researchgate.net/publication/375767861_A_Study_on_the_Application_of_Chat-GPT_in_Media_Production_and_Communication
PR and AI. How artificial intelligence will impact public relations, October 2023, Thumos
Academic research from Mike. S. Schäfer. Science communication in the age of artificial intelligence, Journal of Science Communication https://jcom.sissa.it/article/pubid/JCOM_2202_2023_Y02/
Brenda Gratas. ’50 ChatGPT Statistics and Facts You Need to Know’, Invgate https://blog.invgate.com/chatgpt-statistics#:~:text=It%20was%20trained%20on%20a,12.
A guest post from ChatGPT in response to Brendon Craigie’s article.’ Generative AI and human originality’, Tyto PR https://tytopr.com/generative-ai-human-originality/
(8) PR Academy, The trust dilemma — Managing content in the age of generative AI, Guest Author: Zarrion Walker, pracademy.co.uk/insights/the-trust-dilemma/
The AI Journal, Why do PR professionals need to gain a better understanding of AI tools? aijourn.com/why-do-pr-professionals-need-to-gain-a-better-understanding-of-ai-tools/
(9) CIPR. Humans needed more than ever (2023). The world’s first comprehensive analysis of the use of AI in PR and its impact on public relations work. Authors: Professor Anne Gregory, University of Huddersfield, Dr Swati Virmani, De Montfort University, Jean Valin Hon FCIPR https://www.cipr.co.uk/CIPR/Our_work/Policy/AI_in_PR_/AI_in_PR_guides.aspx
CIPR Review, Artificial Intelligence (AI) tools and the impact on public relations practice, www.cipr.co.uk/CIPR/Our_work/Policy/AI_in_PR.aspx
PR Academy, Briefing AI in PR, Author: Richard Bailey, pracademy.co.uk/insights/briefing-ai-in-pr/
(10) PR Academy. Kevin Ruck, Exploring Internal Communication: Towards Informed Employee Voice. Chapter 17: Automation and artificial intelligence: The reinvention of practice
