Earlier this year I have been interviewed by Inpuncto, the newsletter of PilotProjekt, about corporate communications and AI. This was right before the results of #PR2025 were released and after my contribution to AMEC’s Measurement Month on measuring VR (virtual reality) and AR (augmented reality) for PR.
The interview has been translated in German. The English version is below.
INPUNCTO: According to our research results to date, the use of AI in corporate communication concentrates on the areas of marketing as well as product and sales PR. Other fields of PR such as media relations, human relations, public affairs, crisis PR or corporate identity do not (yet) seem to be the focus. Does this statement apply, or do you have any hints or examples for the use of AI in these fields?
Ana Adi: Marketing and advertising are more keen to use AI or better said Machine Learning Systems and talk about it. It helps position themselves as innovative, cool, groundbreaking and different. That is mainly because they do have a B2C outlook. With PR/corporate communication having multiple stakeholders to consider, the incorporation and use of AI/machine learning depends on the relevant stakeholders.
Moreover, machine learning (big data aggregation and analysis, warning systems) are certainly being used by organizations, especially by big organizations (big data means also bigger budgets) but as they are there to provide them with a business advantage, organizations will be less likely to publicly admit to using them.
Think of banking, health care, politics. Their business success and licence to operate are highly dependent on such discourses which is why such organizations would be prone to embedding extensive monitoring dashboards. This is insight, data driven insight and it is almost live, instant. You would not be going about to tell your competitors that you are using that tool. Machine learning that is supervised is being used already, more than probably people will admit to.
INPUNCTO: Do you know of PR campaigns that have been designed and implemented with the help of AI? If so, can you give examples and evaluate their success
Ana Adi: I do not think you can make that clear distinction, that very clear cut between advertising and PR in that context anymore. Let´s take an example. One of the most recent examples that we have seen is Lexus, using IBM Watson and a team of creative people. IBM Watson served to write a script for an advertising spot and a very extensive team of creative people to bring that script to life. The output of this effort is an ad, so something that is aimed at driving sales, if you take the very strict definition of advertising. When watching the video as a potential car buyer you would not know it has been scripted by IBM Watson. But you and I know about it, and that is PR work: all the actors involved and featured in the project spoke about their collaboration in order to position themselves as innovative. This is a reputational dimension.
Most of the materials produced (and the coverage thereof) portray them as innovators, risk takers, creatives being on top of the curve; all that is PR. Equally, if you look at the output as a result of the coverage and discussions in online forums about Lexus’ use of AI, this is not happening in the consumer driven areas (your regular YouTube, Facebook and the like) and most certainly is not about “Cool. This ad makes me wanna buy a Lexus”. The discussion about Lexus’ use of Ai takes place in specialist forums, groups of communicators and creatives. That is PR.
Another example is AP. They have been using machine learning to write sport articles and financial reports for quite some time. We are thus talking about AP using AI. But who do you think was telling us that story? It was not the journalist or the IT person! It was a person from the communication department who highlighted AP’s innovation. That means AP is at the top of the curve. It´s again a case of reputation and positive awareness. It is about positioning AP in peoples’ minds as an organization that is not afraid to explore, and an organization therefore worthy to collaborate with. This emphasis on innovation and exploration has an additional objective: citizenship. AP is contributing to the teaching of the society.
To sum up, businesses in organizations will embrace AI in an area that will provide them with a business and reputational benefit, that will support their core activity. But the communication of that embracing that technology is done and mediated by the communication department.
INPUNCTO: How does AI change corporate communication and PR in particular? Are they getting better or more effective?
Ana Adi: There are two elements. I think it is very important to go back and define AI. I more like to use the term machine learning more. AI is more in line with Hollywood movies, of sentient machines. But we are not there yet. We are talking about tasks that are handling a lot of Data. ML is not a scary thing. It is about analysis, planning and prediction. Some of those systems work unsupervised, a lot more worked supervised.
Machine Learning has the potential of improving someone’s job. Provided that they deal with groups of stakeholders and with environments that are complex. If I am working for a small shop in the corner of the street, I might not necessarily need to be worrying about machine learning because truly the audience and the people that I care for are not that far away. But if I work for a bigger organization, machine learning might be useful in various ways. It would be a great research and insight resource. If an organization is big enough then machine learning, understood in a way of either aggregation collection or analysis of Data, can only support communication professionals in giving them more specific but also more contextual insight into various stakeholders.
The challenge in this case is that communicators need to get their head around accepting that communication could not only planned but also made specific to different stakeholders.
PR is either seen as an art in which case is perceived as something intuitive, or at the opposite end, as persuasive communication in which case the associations with manipulation, spin, evil are not too far. The fact is that all organizations would want to persuade stakeholders to support them because that would make business better and bring increasing loyalty and revenue. But that doesn’t mean they go about it without any consideration of the others they engage with.
On the contrary, relationship building requires communication and it implicitly involves persuasion. Persuasion requires listening and understanding of the interlocutor.
It is here that technology can help. PR is there to support organizations to understand the demands of their multitude of stakeholders. And it is time to make it obvious that some demands are incompatible and therefore cannot be addressed either values-wise, ethically or financially. Technology can help provide these insights. And that is a good thing.
Communicators need to stop fearing that technology is either going to take over their work or is going to make them obsolete. Even if technology can take over about 15 or 20 per cent of the work (remember data gathering, planning, automating tasks) that means that there is still a lot of interpretation left for communicators to do, a lot of analysis, a lot of putting insights into context.
If you are of the opinion that machine learning increases the quality of corporate communication, is it a competition-relevant factor that provides considerable advantages for companies with a stronger budget?
Ana Adi: Potentially yes. It helps to listen, understand and then relay what various stakeholders expect from (and how are impacted and how they can impact and) organizations. The more successful, the clearer the guidelines from the communication department will be, the more tailored the messages coming from an organization to particular stakeholders. That would have a particularly positive influence on the organization. For big business as for politics, it makes sense to understand the variances within the stakeholder group, to understand the supporters, distractors, time bombs and so on, and communicate with them in a way that bring the organization closer to them in a collaborative manner.
If you have a good monitoring system that you can pick up information that is sensitive (changes in public discourse and mood) it would be useful to be able to address it. An organization would be monitoring various influencers as well. As a communicator it is very useful to know what they are writing about and also have the context in which they are writing (so not only limited to the organization). It is the context that allows communicators to engage with them and see what whether and how collaborations are possible.
The more you know how to ask for Data (clustering, classifying, combining), the better and the easier it is to work with Data. It’s almost like playing with Lego blocks: understanding and knowing what data elements are available, one can create a different story at different times with a different tonality which would resonate differently with different stakeholders. This, in the long run, can help organizations.
Let’s also speak about readability and comprehensibility of texts, so how quickly you can one read a text and how easy it is to grasp its content.as these are also elements prone to machine learning.
A simple example: The PR department wants to write a digital story on a given topic. So an employee writes the text. Through such automated an analysis, the text is then examined for its readability and comprehensibility before it is published. Are the sentences too long or the words too complicated? Then they can be improved accordingly. Once the number of texts optimized in such a way is large enough, the algorithm could be then trained to make suggestions for changes: Shorten this sentence, replace this word with a synonym, write this passage more understandable. This sort of technology already exists but it is, to my knowledge, used little by communicators. It has however great potential to improve the work of communicators. Of course, this is only part of the work. The communicators needs to also ensure that they track the impact of these texts (whether completion of reading, aided/unaided recall, understanding of the message).
INPUNCTO: Machine learning is being used with increasing success in investment, tax and legal consulting, and deep learning methods are also already showing considerable success in diagnostics. Is it only a matter of time before a virtual PR assistant becomes the most important or influential “conversation partner” of a company’s chief communicator?
Ana Adi: There are already virtual assistants like Alexa, Siri and Google. However, they still solve quite simple tasks at the moment. What a PR department needs is a technology that could handle more complex input. Companies, for example, are still very weak at using the data they sit on. If I remember correctly, a recent study pointed out that perhaps only about 10 percent of a company’s communicative output is based on stories told by employees in the communications department. That means there is another 90 percent of data in the company, in various forms (from meeting notes to reports, all of them available somewhere) that are not used in any way. So it’s very difficult for the communicators to find out in time what stories are lying dormant in the company, which have a high information value and a high utility value in order to be able to prepare and communicate them promptly. A solution where information would be flagged up at an early stage would be useful. Provided it has the necessary parameters to recognize which information, which story is interesting and can be of use to the company. But for such a system to work, one needs to first know what and how to supply it and how to classify the information. We’re speaking here of taxonomies, but built into data design and not just as an afterthought.
A virtual PR assistant does not yet exist. The technology still lacks a level of interpretation and context reference. There is no doubt that machine learning can be used to relieve communicators of many routine tasks. But beyond that, the current state of the art cannot yet achieve anything. Human communicators will not be to be replaced soon.
INPUNCTO: How will or must the job profile of corporate communicators change in the near future? Will they also have to be IT and data experts in the future?
Ana Adi: It is interesting that you ask that. I am currently running a study called PR2025. It is a Delphi method study whose aim is to identify the trends, competences and solutions that communicators should consider in the mid-term in order to maintain their relevance and confidence in the profession and the future. One of the questions communicators are asked refers to the technological competence that communicators should have, considering that technology is playing an increasingly important role in our daily life. Some of the responses from the 69 participants include
- Digital literacy
- Moderation skills
- An understanding of coding
- An understanding of big data
- An understanding of content production (including video and advanced graphics, both online and offline)
- The ability to create data models for machine learning
- The ability to operate on a multitude of devices
- The ability to operate community software and tools
- The ability implement agile working methods
PR2025 will be launched at the European Communication Summit in May this year and the second, the quantitative round, is currently running.
So, to come to an end: Communicators will certainly not have to become IT and programming experts. This also includes the use of platforms on which they can combine different media in such a way that the optimum variant of a story is created for each stakeholder group. Now one can ask how they can achieve all this without neglecting their previous tasks? This will be a question of training. We will have to make sure that they learn to understand and use modern technologies as an important component of their job.
What developments do you think will be realistic in corporate communications over the next three years, and what do you think are false expectations or fears?
Ana Adi: First of all it will be important to integrate the ethical component into the discussion. The binary approach is characterized by the choice between black or white, yes or no, inside or outside. If algorithms are to make decisions, they must be designed in such a way that ethical aspects are also taken into account, depending on the complexity of the question and the consequences of their decision. There is likely to be an increase in the use of automated data analysis. Monitored machine learning will continue to make progress and enable improved monitoring of social media. Appropriate organisational and analytical processes will help to produce better content. The platforms will become more elaborate and enable a kind of data content management system, whose various modules resemble the principle of Lego building blocks. In all of this, however, it is also important to help communicators decide which degree of relevance each technical option they are presented with has with their stakeholders and, more important than anything, have them decide which objectives can these technologies truly support.