Share this
by Raime Merriman on Oct 9, 2023 3:27:21 PM
Our recent webinar featured so many great questions about artificial intelligence (AI) and its effect on public relations. While we responded to many of them live, we weren’t able to answer them all!
In this blog, we answer the questions we didn’t have time to get to plus we provide sample prompts and a list of AI tools mentioned during the broadcast. We hope this resource further educates you about generative AI and how it can help you do your job better.
If you want to learn even more, watch the full webinar on demand.
AI and PR: Accuracy and Authenticity
How are PR pros addressing the concerns that our work could be seen as plagiarism? Or is that not a concern?
When it comes to this issue, PR pros need to educate themselves and act as a resource for educating others. This includes engaging in open dialogue - including conducting webinars, Q&A sessions, and panels with experts on AI issues. They can also create and share case studies of instances where AI was used ethically and responsibly.
Companies should regularly communicate with stakeholders, make their policies and guidelines public, and consider getting external audits for AI algorithms and data practices to show the company’s commitment to responsible AI.
Finally, they should be transparent about the use of AI and identify if any content was created either wholly or in-part with the use of AI.
As suggested in the PR Council Guidelines for Generative AI Tools, always check for inadvertent plagiarism, copyright infringement or trademark infringement in AI-generated output.
Do we foresee a time when it will have to be disclosed that AI was used to write a particular article or statement?
Yes, absolutely. As we move into the next phase of generative AI, the concept of provenance will become increasingly important.
Generative AI is crossing a yet-to-be-defined border between statistical models and human creativity, and while this is something we should welcome, we can also ask that it identify itself when it does so.
If you’d like to learn more about what is being done in this arena check out the Coalition for Content Provenance and Authenticity (or C2PA for short). This group is an alliance between the software industry, newsrooms, and non-profit organizations to design and implement technical standards to combat disinformation online.
I've validated that ChatGPT has made up peer review journal citations that don't exist, falsified court case precedents and responded with incorrect data when posed certain technical questions. As ChatGPT can't be trusted to write or research, what are you recommending that it can do?
As you would do in any research project, we recommend you use a variety of different tools in conducting research and validating data and checking sources. Both Google Bard and Bing AI tools will provide citations and links that can be verified.
You can also ask in your research prompts for the specific sources to be cited with links.
What are examples of tools you use to detect plagiarism?
For AI, there are no tools that can guarantee detection with 100% certainty as of yet, but there are several tools available. One of the most popular is AI content detector Copyleaks. If you are still unsure whether a piece of content is AI generated, you can always ask the author or publisher for clarification.
When using a service that helps you create an AI press release, how does the service ensure that what AI has culled is accurate?
For the GlobeNewswire AI Press Release Generator, you provide a summary of the details you want it to take into consideration including a headline idea, storylines and topics to avoid mentioning – if relevant.
To ensure accuracy, a human touch is required. Once your draft content is generated, always review and refine your press release including the copy and contact details.
AI and PR: Ethics and Security
How can you safely use AI when you are working with confidential content?
Make sure that the providers of the tools you’re using are taking extra steps to secure the information you provide. Not all generative-AI technology will ensure that non-public information is kept private.
If you’re not sure whether this is the case, read the relevant documentation or ask the technology provider for more information.
For example, ChatGPT offers an opt-out to keep your data out of their future trainings, so the default is that your data IS being used. You need to read the fine print for each of these tools.
What are the legal pitfalls that we should be aware of when using an AI platform?
As a best practice, it’s important for brands and agencies to develop clear policies around usage of AI and evolve and update them as needed. This ensures the organization is maintaining its privacy and data security, as well as making clear what is acceptable and unacceptable use of the AI tools.
PR professionals can also build trust by publicizing the company's guidelines and policies regarding AI usage, data handling and other associated ethical concerns.
How do you address the inherent bias in AI? It can shape stories and ideas that perpetuate the status quo. For example: highlighting straight, white, male, voices and similar non-diverse voices and perspectives and needs.
AI reflects and magnifies our society because of the material it was trained on. It requires awareness just as it did before the mass acceleration of AI.
When selecting tools, it’s important to vet biases thoroughly, have oversight processes, test outputs and follow codes of conduct around transparency and objectivity.
What are safe ways to prepare all these great types of content for PR? (concerned about leaked IP, proprietary info, etc.) For example, prepping for a product launch or media interview - prepping content BEFORE it's in the public realm.
Make sure that the providers of the AI tools that you’re using are securing the information that you input by ensuring that all non-public information is kept private. If you’re not sure whether this is the case, ask the technology provider for more information.
For example, with our AI Press Release Generator, your data does not leave the Notified platform.
What do we need to be aware of in regard to proprietary info or HIPAA concerns for health care communicators?
AI systems must be HIPAA compliant. This means that the AI system must have security measures in place to protect patient health information.
To ensure this, AI systems can only be trained on data that is owned and controlled by the healthcare organization.
In your opinion, is it unethical to use artificial intelligence for drafting AI press releases and messaging for brands?
AI is a virtual assistant. As long as you’re following your company’s guidelines and are disclosing your usage while applying your own common sense and inner compass, using AI to draft a press release should not pose an ethical issue.
How PR Agencies and In-House Pros Can Use AI
Could AI be used for AP Style checking and research?
Yes, you could structure a prompt and then paste the content you want reviewed (or use a plug-in). Note: When pasting in data remember that when using public AI your data is not private and you should not paste in any proprietary information.
Does Google ding you regarding SEO if you use AI-generated content on the web? Does AI-generated content affect website SEO or ranking when posted or embedded into a story on a website or in social media?
In Google Search's 2023 guidance on AI-generated content, they clarify that they are rewarding high-quality content, however it is produced:
“Google's ranking systems aim to reward original, high-quality content that demonstrates qualities of what we call E-E-A-T: expertise, experience, authoritativeness and trustworthiness. Our focus on the quality of content, rather than how content is produced, is a useful guide that has helped us deliver reliable, high-quality results to users for years.”
Note: Google’s algorithm for SEO rankings does indeed change and it’s possible at some point Google could change their position on this.
How do you navigate ChatGPT only having information up to 2021? Since AI is currently only sourcing up until 2021 on the internet, how can you ensure material is current/relevant? And when will AI get past the test phase and source up to current info?
If you need newer information, we recommend you make use of a different tool, like Google Bard or Bing AI which has current information available in its model.
The paid version of ChatGPT, ChatGPT-4, also now has internet access.
Assuming you'd need the paid version of these generative-AI tools to get the most up-to-date information? What's your advice on getting agencies/companies to buy into these tools?
As a part of your business case, we suggest highlighting the many ways that generative AI tools can help public relations teams save time—especially during critical news moments—which allows the team to focus on higher-value strategic work.
Additionally, share how these AI tools can also assist with developing storytelling ideas.
As AI technology changes by leaps and bounds, each version gets more powerful and also is in more demand. The free versions of these tools will lock you out when usage gets too high, and part of the premium offering is guaranteed access without usage limits.
For small PR firms, can you suggest methods for using AI for helping track earned media value if you can't afford expensive software such as Meltwater and Muck Rack?
Using an open source and free AI tool to track media mentions is not recommended.
With AI improving to such a large degree, what are the implications for agencies that rely on the billable hours model? If it takes one hour to do something that used to take five hours, how do agencies remain profitable?
Technology has always held the promise of a reduced work week. Jamie Dimon, CEO of Chase, recently stated that for the next generation of workers, AI will allow workers a 3.5 day workweek.
The reality is that humans will continue to innovate and apply themselves. There is no shortage of work that an agency can be doing on behalf of their clients. AI will just free people up to work on more creative, higher-value and human centered activities.
If a company wants to develop their own AI (like the example with Sephora) what companies can help with this? Are they developing in-house?
For large organizations, we recommend at least 1-2 AI strategists and thought leaders who are focused on constantly scanning the AI landscape, experimenting and bringing recommendations on the best tools back to the team. This is especially critical for PR agencies.
At smaller organizations, it is a good idea to partner with someone who can help with this.
For building a new tool, a large organization may be well served by hiring data scientists and AI developers. For smaller orgs, working with a consultant or partner is recommended.
Investors and analysts rely on AI to develop some ratings (e.g. ESG ratings), AI scans companies reports and documents. If you don't use the word they expect, you miss an answer and your rating plummets. In drafting a document, how can we check that the words are properly optimized from an AI point of view?
You can simply ask AI what the keywords are, giving as much detail as you can about your industry and your requirements. And don’t stop at just one - use multiple AI tools and do your own research. Don’t be afraid to ask AI how to best utilize it, as we saw from the text to image example.
What will be the future role of content writers when it comes to AI technology?
Content writers will always have a role, they will just become more strategic. AI will help ease the drudgery of hammering out the first draft or by-passing the issue of writer's block.
Prompt Writing for AI Tools
How do I develop prompts to use AI for? I find it a challenge to tell the AI machine what to do?
To develop the strongest prompts for AI, tell the tool how you want it to behave including persona that it’s writing from and/or to, formatting, sources, etc. Creative and strategic thinking is critical for getting good results.
It's helpful to have someone on the team who is responsible for ensuring the AI is trained for your brand voice and goals, or those of your clients. AI requires a human touch and expertise to keep content accurate and engaging.
The field is evolving, and it also requires rolling up your sleeves and doing trial and error with prompts hands on.
Kelly mentioned that she used a number of AI tools to generate her sample pitch. Was one tool better than another to use? Or what's the strategy to use multiple AI tools for a singular prompt?
Currently, the best tool depends on the format that you’d like your pitch in, and the story you’re telling. Stay tuned by subscribing to the Notified blog - we have something in the works!
With press releases, how do you ask the right questions and prompts to make the release strong?
For a strong press release, we recommend starting with as much content as possible and writing up a pre-draft or framework.
Let AI create a baseline draft of that content and then work from there. High-quality input in is critical for high quality output out.
Will PR degree programs eventually include AI prompt writing as part of the curriculum?
It’s very likely that generative AI prompt writing will soon be a critical part of PR and communications curriculum. Like the growth of social media over the past 20 years, these tools and resulting platforms will likely continue to transform communications.
AI Prompt Example (Shared During the Webinar):
Write five draft media pitch emails to send to writers at top technology publications about a software launch by Notified for public relations professionals. Organize each email to include a subject line, recipients, personalized greetings, two paragraphs of one-to-two sentences of relevant and intriguing copy and an enticing closing signature from the Notified communications team.
AI and Visual Content
How do you copyright materials that include visual AI content?
We recommend watermarking and Digital Rights Management (DRM) solutions. We also suggest agencies and in-house PR pros advocate for and publicize the use of these tools to protect intellectual property.
What are some specific prompts for using AI with images?
The more details you provide the AI the better. Calibrate the image with the specific style that you want. Most people are looking for photo-image quality when using AI and those prompts are typically described as “hyper realistic,” “photorealistic” or even “magazine cover.”
When in doubt, a simple hack is to ask the AI what the best prompts are to accomplish the task.
For example, if I ask AI: “Give me an example of a prompt that will get the best results for photo quality of a woman driving a car,” Google Bard might answer with something along the lines of: “Create a photorealistic image of a young woman with short brown hair and blue eyes driving a red Ferrari 488 Pista on a winding coastal road at sunset, with the sun reflecting on the ocean waves in the background.”
Additional AI Tools Mentioned During the Webinar
Subscribe to our blog for the latest PR product updates and webinars!
