(MENAFN- PRovoke)
There has been a lot of discussion about how artificial intelligence is changing the way public relations professionals operate. There are apparently many who believe AI is still not“ready for primetime,” while others are most concerned about the automation of traditionally labor intensive tasks such as writing and reporting, and others focus on AI's creative capabilities.
But recent meetings with leading global agencies demonstrated how forward-thinking firms are already deploying AI in a strategic role that impacts planning and strategy, message development and more, all the way through to measurement and evaluation. And it uncovered some basic themes regarding AI as a tool for better, smarter PR rather than a labor-saving device, and especially the need to combine AI with good old-fashioned human intelligence.
I invited representatives of six firms on the leading edge of AI usage to talk about how AI is already impacting corporate communications. Participants included:
Brian Buchwald , global chair, AI and product at Edelman
Wendy Carhart, chief communications, culture and purpose officer at Real Chemistry
Imad Lahad , head of AI communications lab at APCO
Chad Latz , chief innovation officer, Burson
Chris Perry , chairman, Futures, at Weber Shandwick
Brigitte von Haacke , parter and CEO, Europe, at FGS
The following discussion was conducted via email and has been condensed for publication.
Paul Holmes , founder of PRovoke Media: Let's start with a basic question: what is the biggest impact of AI on the practice of public relations as of right now?
Brian Buchwald : It's discovering what is automatable to make us more efficient versus what is truly value producing to drive differentiation. We spend a lot of time building AI agents that can make our work faster. But simultaneously we are investing in tools and intellectual property that enable us to be more creative, more thought-provoking and a better advisors to our clients.
Chris Perry :
The greatest impact I've seen is less on what we can do more efficiently (like using GenAI to write press releases), and more on what we can do to better, like using GenAI to understand how information now travels, making sense of cultural chaos, crafting resonant stories, and identifying others than help translate and tell them. The ultimate value is being faster and better at what we do. Not replacing jobs or reducing costs.
But a word of caution about impact. Talk of tools and efficiencies is delaying the impact GenAI can bring to organizations. As billions in VC funding are being poured into large and small models, the real question is how to find the best tools to solve pressing problems-a problem-solving versus tools-led mindset. Elon Musk says the biggest mistake engineers make is "optimizing a thing that shouldn't exist.” There's a lot of discussion optimizing known things versus creating new things that elevate our work.
Chad Latz : The underlying mechanics of how we deliver some of the core tasks of public relations are changing before our eyes, whether that's automating content production, optimizing media materials and outreach, or mining intelligence from data. As we navigate this transformation, there is a need to balance efficiency with consultancy since the speed by which communications can be produced requires greater diligence and craft to ensure the integrity of the output.
For all the obsession with generative AI, we are seeing the greatest impact coming from cognitive AI, or specifically our ability to train models on how various audiences think in order to predict behavioral outcomes, consult clients accordingly, and produce communications to maximize the result.
Imad Lahad : AI's impact is a tsunami-but not because it's just the one final large wave you see crashing to shore. Instead, it's made of various waves of various sizes and strength coming together to change everything.
By revolutionizing data and analysis, AI enables us to anticipate issues and opportunities, rapidly convert vast amounts of diverse data into actionable insights, detect patterns with advanced algorithms for predictive insights, and necessitates a shift in how we perceive and utilize data.
PH: What is your approach to preparing your agency for AI? Are you doing intensive training of a select few or trying to make everyone at least competent? What aspects of AI are you focusing on?
BB: Our strategy consists of three main components: a dedicated AI team, a bespoke workflow creation process, and a scaled change management program. The AI team drives AI ecosystem development, vendor partnerships, adoption, and enablement. They also oversee Trust Stream and Archie, our custom trust platform and large language model. The workflow creation process helps us identify and solve pain points, optimize efficiency, and enhance quality. The change management program ensures that we recruit, educate, and comply with the ethical and responsible use of AI.
We also provide our staff with foundational and role-specific toolkits that enable them to use AI in their daily work. These toolkits today include ChatGPT and Microsoft's Co-Pilot, general-purpose AI tools, and Adobe's Firefly and Stability, which are role-specific tools.
In short, we are training and enabling our staff to use AI in their specific roles and functions, while also building a core team of experts who can drive the AI transformation across the firm.
Wendy Carhart : Our AI strategy democratizes AI knowledge, ensuring every employee is equipped to use AI solutions tailored to the diverse and specialized needs of our clients.
We started by creating our own, privacy-safe version of ChatGPT, called RCIS Workspace, which is available to all employees. We then created a Guild model, sharing knowledge across the organization through ongoing learning sessions where employees share examples, use cases and learn from one another. This approach ensures that everyone, regardless of role, gains AI competency and is empowered to use in their daily workflows.
To ideate new solutions, we created the Skunkworks program, which pairs one of our AI experts with a business expert to create new solutions that directly benefit our clients. We're seeing tangible outcomes from this program in the first few months – things like creation of a media planning and optimization assistant.
CL : We have access to WPP Open, an AI-powered marketing operating system which delivers capabilities across marketing processes including creative, media, production, communications, and commerce, offering a range of outputs and tools tailored to specific needs of clients.
Additionally, we have customized AI solutions bespoke to specific communications applications. For each, we've provided training as part of our curriculum and are assessing utilization to identify where additional learning or change management support may be required.
Brigitte von Haacke : Our AI Taskforce is one example for our thinking. Established in early 2023, the taskforce rolled out our internal GDPR-compliant AI chatbot“Fergus” within six weeks, equipping our workforce around the globe with prompts for everyday tasks. From summarizing complex texts to creative ideation, we use Fergus for all kinds of challenges.
We develop solutions, both internally and externally, to drive our transition to AI-driven analyses of large data sets. This includes media coverage, social media postings, analyst reports, annual reports, and political speeches, ensuring comprehensive insights for our clients and teams. For this purpose, we have built a special team trained in areas like natural language processing, machine learning for pattern recognition, and generative AI.
IL : We have a tiered approach.
We are selecting 60 advocates from each office to serve as the front lines of our AI transformation. These advocates have multiple responsibilities: mastering our AI approach and training others, promoting AI's potential to inspire innovative applications, and curating top case studies and thought leadership.
Additionally, we are providing comprehensive AI knowledge to everyone through ongoing training sessions, an open-office policy with AI leads and flexible office hours, and a knowledge-driven video series.
CP : It's a business and change management priority firm wide. It starts with setting a clear agenda for what we plan to do, why and how. This step, often overlooked, is critical to address the fear factor people have about AI in the workplace. We've been clear from the start we see AI as means to elevate our work not automate it.
Our Futures team is an accelerant to make our intentions real and credible. We have capacity to do research, build AI partnerships, and develop client pilots. This team spends a lot of time with clients on needs to prioritize what we build. Between lab and client work we've created new lines of business with incredible potential.
In parallel, we've been intentional about building cross-agency pilot teams, giving them tools and safe space to determine what our genAI capabilities can and can't do.
PH: How are you making decisions about what to buy and what to build? What kind of proprietary tools are you building that build or improve on what technology companies are developing?
BB : W starts with the clients and their evolving needs. We seek to partner first where the market has capability. We will only build where the market has a clear need and Edelman has a proprietary benefit to offer. This is the case with our Edelman Data Lake, Trust Stream and our homegrown trust LLM, Archie. These offerings build off one another.
The data lake ensures we have a massive repository of digital data at our fingertips and available to be analyzed. Archie enriches that data to quantify trust and to make it measurable and actionable. Trust Stream empowers our teams and clients to leverage the enriched data. This becomes a virtuous cycle ...
IL : Since the inception of our AI Comms Lab in 2018, our approach is to do extensive R&D on various tools and technologies to see what fits more with what our clients' problems are.
We have tested and continuously testing 100s of tools and platforms and updating our infrastructure and solutions accordingly. We are not a technology company and we do not want to reinvent the wheel.
BvH : While we use out-of-the-box third-party solutions for some needs, we build on existing technology to create our own solutions where we see a market gap that we are uniquely positioned to address with our expertise.
A prime example of our approach is our IQ Suite. More and more clients focus on replacing gut feeling with fully data-driven, measurable and comprehensive reputation management. To enable this transformation, we launched ReputationIQ and MediaIQ. They combine real-time stakeholder reputation tracking with AI-driven media analysis. Which provide detailed insights to understand how and where to engage stakeholders, and to measure progress.
CL : Our Innovation Group has a dedicated team that is constantly evaluating new technologies. At any given time this includes assessing more than 20 LLMs as well as a significant portfolio of SaaS products. We look at investment in two ways, first are embedded technologies, which include AI-powered commercial grade SaaS solutions that are commonly available. Second are created technologies, which are bespoke solutions that we build that serve unique needs for clients.
We've built several suites of proprietary solutions. Some of these are created exclusively by Burson, while others are built in partnership with other technology companies. It's important to be able to effectively scan the horizon, anticipate opportunities and create solutions that won't otherwise be immediately replicated or obsolete.
CP : We focus on problems to be solved with genAI, not genAIs in search of a problem.
For us, the question has three starter threads: What can the tool solve? Are there enterprise-grade tools to solve them? Are task-solution tools both available and sustainable given how fast model options change?
PH : How should clients be using AI in the planning process? How can it lead to smarter insights, and predict the effectiveness of different strategies?
BB : As ChatGPT, Co-Pilot, and Stability APIs are being integrated into nearly every analytics, research, and creative platform, the opportunity to create new workflows and processes for the planning process is easily accessible. This empowers strategists and their colleagues to custom build workflows by role specific AI tools to get to insights and next best actions faster and more effectively.
It also means that the work can both be grounded in simulated optimizations and the results can be leveraged to drive empirically better results in the future.
IL : Clients should view AI as an essential partner in the planning process, not a replacement for human expertise. AI, powered by clean, validated data, can uncover hidden insights and predict outcomes that would be challenging to identify manually. This is especially valuable in risk-sensitive sectors, where AI can simulate various scenarios and inform strategic decisions.
However, the most successful companies don't rely solely on AI; they maintain a human-in-the-loop approach, ensuring nuanced judgment and adaptability in the face of evolving circumstances.
At APCO, Margy-an adaptive intelligence platform-offers real-time insights into brand reputation, issue management, and policy landscapes. By treating AI as a colleague and partner, providing context and integrating its insights with human expertise, clients can leverage AI's analytical power while ensuring that strategies remain aligned with broader business goals and ethical considerations.
BvH : Clients should leverage the vast amounts of stakeholder data that the latest generation of AI provides. Having learned from publicly available communications, generative AI can help identify previously overlooked stakeholder groups, analyze and understand their motivations, personalize messages and tailor them across different channels, regions and languages.
The tools we build allow clients to quantify the impact and perception of campaigns, announcements, extraordinary events, or leadership changes. Clients can then use these insights for comprehensive analyses, support CEO positioning with unique themes and stories, and benchmark against their peers.
In any case, working with AI should always be a dynamic back and forth between human communication professionals and the machine, involving continuous questioning, answering, and adaptation. We pursue this approach in our internal processes and when advising our clients.
CL : Clients are realizing tremendous value in our proprietary cognitive AI solutions, with the ability to use our models to predict the behavioral outcomes of any audience for any communication in any format, delivered by any spokesperson or KOL. For one global client we were able to test our communications strategy and approach by predicting the outcome of five communications scenarios, with five different core messages per scenario, delivered by four different spokespersons, all within hours.
This approach allows us to deliver more precise insights, strategies and messages.
WC : Successfully leveraging AI in planning for healthcare is nuanced and requires two things that cannot be built overnight with just a license to a general purpose LLM.
First, structured datasets are essential. Having a complete ecosystem of healthcare datasets such as claims data, social media, clinical trials, PubMed, social determinants of health and more must be included for planning purposes. Second, machine learning models need to be built on top of those healthcare datasets to accurately predict which healthcare providers have likely undiagnosed or misdiagnosed patients for rare diseases so that we know who to engage. And, we can predict how specific, influential reporters and analysts are likely to react to content and issues.
PH: What about the ability of AI to create different personas and predict how different groups might react to messaging? How should this be used in issues management or crisis response?
BvH : In scenario planning, such as crisis preparation or public affairs strategy, AI can create scenarios that range from probable and plausible to creative and almost absurd. Both are helpful in closing blind spots and stimulating human creativity.
Business leaders and communicators probably want nothing more than to predict the future. And this may be possible in part with AI-based models and scenarios. We and many of our clients, use generative AI as an initial "advocatus diaboli." We let AI impersonate a critical journalist or stakeholder persona to stress-test messages. This can help to develop a wider variety of scenarios and close blind spots.
BB : Our clients are increasingly thinking about all their key decisions and actions through the lens of multiple stakeholders. I say increasingly not because it hasn't always been a part of the process, but because now there is an actionability component that was much more time intensive and costly previously. For instance, with our Archie Trust LLM, Edelman can break down any article or social post or even a prospective press release and score it for overall trust. But we can also break it down to understand and explain the likely impact on investors and company employees.
This also contains a critical human in the loop element. We have experts who need to both participate in how we train proprietary and external models. But we also need their voices in how we prosecute the work on behalf of our clients. We can simulate a crisis and create an array of responses. But there is no substitute for a veteran of our crisis team o to leverage their experience and judgment to make final calls.
WC : There is a necessary balance between human and artificial intelligence when it comes to issues and crisis planning and response. The gut and guess approach on how stakeholders respond is only as strong as the relationships and experiences you already have with those stakeholders.
By the nature of a crisis, there is almost always something unexpected, nuanced or unique. Clients then have to make a choice at an inflection point: mirror the best that has been done to date or lead forward with an innovative resolution.
IL : LLMs' proficiency in mimicking human behavior and public sentiment across diverse contexts, make them invaluable for issues management and crisis response. However, AI models are only as good as the data they digest and can fall behind if they're not tuned to real-time events. While they're great for spotting patterns, they need constant updates to stay sharp and accurate.
CL : General LLMs can be prompted to assume the personas of different architypes, and while this might be valuable for creative stimulus, it does little to develop a reliable synthetic audience segment that can accurately produce a prediction (within a reasonable margin of error) that would be valuable to a client.
For this reason, we are constantly building high-value segments with specialized and custom training data so that our AI models that can accurately predict an outcome for any audience. The data and output from the AI are validated for accuracy and the margin of error is transparently displayed to our consultants.
BvH : Beyond what we consider to be the everyday use of AI in communication these days, we are a little skeptical. The more complex and distant the scenario, the less accurate the prediction. Messages often develop their full impact within specific contexts, which can be challenging for AI to replicate accurately using historic data.
PH : What about metrics? How is AI going to enhane our ability to measure the real world impact of what we do, to improve on the fly or to demonstrate changes in attitude and behaviour?
BB : My team is absolutely obsessed technically with proving and improving the business value of communications. It simply changes the perceived value equation of how we contribute to our clients' businesses.
Edelman already does this work on a bespoke basis through attribution and media mix models that we build for clients via our Predictive DXI team. But we now are developing a scaled automated platform to define the direct relationship between communications activities and trust and business KPIs. We aren't at the finish line. Yet we are well on the way.
BvH : Ultimately, what stakeholders perceive is what matters most. We already rely on surveys to gauge changes in attitudes and thus the impact of our work. AI enables media sentiment analysis on another level. Modern LLMs consider context, making analysis straightforward and accurate. Unlike earlier technologies that struggled with nuances like sarcasm or irony, today's AI can interpret these subtleties, providing richer insights.
Our media analysis offering, MediaIQ, and our internal media analysis tool build on the latest AI technology stack, helping us identify patterns in the data we analyze. We are also working on integrating these capabilities into ReputationIQ, allowing us to present survey data in human-readable language, making changes in public perception even more graspable for our clients.
CL : Measuring impact and optimization is one of the most powerful cases for AI. We aggregate the full spectrum of communications, media, behavioral and attitudinal data and then use AI to determine relationships, correlation, and opportunities for optimization. ANS we're able to validate the cognitive AI forecasts we've made earlier in the process.
By way of an example, for a large retail client, our cognitive AI models accurately forecasted shifts in key brand metrics and reputation drivers, based on emerging signals across news media and social. By being able to predict attitude and action of different stakeholders, we can refine communications strategy and quickly optimize messages for desired outcomes.
Story image created by Co-pilot
MENAFN12082024000219011063ID1108542925