Public Affairs & AI: Guidelines for Responsible Use in Digital Lobbying

Introduction

Promoting and applying the principles of Digital Lobbying in Public Affairs requires constant study and an in-depth knowledge of the tools available to professionals in the field. This approach becomes even more essential and urgent in light of the significant and rapid growth of Artificial Intelligence (AI).

Among the many sectors affected by AI, lobbying and Public Affairs shows particularly widespread adoption: over 75% of Public Affairs professionals report using AI tools for daily tasks such as research, editing, content creation and translation¹.

As highlighted in our guide Generative Artificial Intelligence in Public Affairs: Instructions for Use, published in summer 2023, and as reiterated in the Digital Lobbying methodology, at ADL Consulting we conceive of and use AI tools from a human & tech perspective of augmented intelligence. This safeguards the quality of our results and analysis, allowing us to seize the most significant advantages while remaining aware of the risks involved.

However, AI adoption is not without obstacles. The success of AI in Public Affairs depends not only on its technical capabilities, but also on people’s willingness to embrace it and use it effectively. Distrust of AI is widespread: according to Forbes Advisor, 77% of Americans fear AI could reduce employment over the next 12 months, while 80% are concerned about the security of their personal data. Furthermore, a YouGov survey found that almost half of Americans believe AI could eventually pose a direct threat to humanity.

In this context, responsibility and awareness are the cornerstones of AI use in Public Affairs. Drawing on daily best practices and recent research by ADL Consulting, this article outlines the opportunities and risks of using AI, and then offers practical solutions to promote its safe and responsible use in our strategic consulting sector.

Why is there distrust of Artificial Intelligence?

In an article published in Harvard Business Review Italia in October 2024, Julian De Freitas, Assistant Professor at Harvard Business School, examined why many people are reluctant to adopt AI tools. The main concerns relate to the risk of job losses, the security of personal data and even apocalyptic fears of a future in which AI could pose a threat to humanity.

Research has identified five psychological barriers that hinder AI adoption:

• Lack of transparency: algorithms are perceived as mysterious and incomprehensible, which breeds distrust.
• Lack of emotion: AI is seen as impersonal and lacking empathy in interactions.
• Rigid decision-making: AI systems can appear inflexible compared to human adaptability.
• Fear of losing control: people worry they will be unable to manage decisions made by AI.
• Preference for human interaction: most people prefer dealing with human beings rather than robots or chatbots, especially in customer service.

People perceive AI as an opaque and unpredictable entity, unable to grasp emotional and social context, and often associate it with risks to data security and labour market stability. Resistance to AI also varies across cultures: in Japan, for example, technology is more widely accepted thanks to a worldview that attributes a soul to objects, whereas distrust prevails in Western countries.

To overcome these barriers, the most innovative companies take a gradual approach: improving algorithmic transparency, integrating human support into AI use, and tailoring its application to the specific features of their sector and corporate culture. Effective communication about the benefits of the technology, combined with an augmented intelligence model in which AI supports – but does not replace – human expertise, is essential to build trust and acceptance of innovation.

Opportunities

By way of example, and in brief, the opportunities offered by AI that we identified – especially while writing Generative Artificial Intelligence in Public Affairs: Instructions for Use – fall into four key areas for Public Affairs professionals:

1. Efficiency in analysis and text production: automation speeds up repetitive activities such as parliamentary and media monitoring, report writing and stakeholder mapping. AI and text mining tools make it possible to analyse large volumes of data in less time, providing insights into trends such as public opinion or sentiment. Chatbots and generative AI also make it easier to draft press releases and policy briefs.
2. Data-driven analysis and forecasting: AI can process data at scale, detecting political and regulatory trends with greater precision. Predictive tools help plan future scenarios, providing a clearer strategic vision in complex regulatory environments.
3. Communication and advocacy: AI tools, especially generative ones, enable more targeted, personalised and effective communication activities and projects with and for stakeholders, making relationships more effective and strengthening ties with key interlocutors.
4. Decision-making and strategic support: AI makes it possible to simulate different scenarios before implementation, reducing the margin of error and improving adaptability to political and institutional dynamics. Predictive models help assess the effectiveness of lobbying strategies and anticipate stakeholder reactions.

Risks

Being open to the opportunities enabled by AI does not mean overlooking the risks of innovation, which must be managed carefully. Although many of these risks are general and apply across sectors, they are especially relevant in our field, given the sensitive nature of the information handled and its political implications. Below, we explore three key risks:

1. Data protection and privacy: entering sensitive data into AI tools can pose a risk, especially for confidential political or corporate information. Data on political decisions and processes in particular must be handled with the utmost care, as must personal data, which must be protected as required by the GDPR. Improper use or excessive data collection by AI tools could breach these regulations, exposing Public Affairs firms and their clients to threats and even legal penalties. Moreover, if AI models or applications are compromised, there is a risk of breaches of confidential data, such as political strategies, internal communications or stakeholder information. As outlined below, protecting this data is essential to avoid reputational damage or the loss of competitive advantage.
2. Bias and inaccurate outputs: another significant risk lies in the biases that may be present in AI models as a result of their training processes. If the data used to train these models is partial or distorted, the resulting analysis and insights may be too, leading to incorrect or one-sided conclusions. This risk is particularly critical in Public Affairs, where accurate information is essential for strategic decisions. Precisely because of these inherent risks, it is essential to avoid overreliance on AI tools. Relying too heavily on their outputs reduces the weight of Public Affairs professionals’ oversight and judgement in decision-making. While AI outputs are valuable, the complexity of the political and regulatory context requires a balance between technology and human strategic intuition, to avoid short-sighted choices or choices lacking the sensitivity that delicate contexts demand.
3. Lack of transparency and control: many of the algorithms that power AI function as “black boxes”, making it difficult to understand the decision-making process behind their analysis. This can generate distrust and hinder the acceptance of AI in lobbying and advocacy processes.

Solutions & tips for responsible AI use

Having clarified the opportunities and, above all, the risks, the following section aims to provide practical guidelines for Public Affairs professionals.

Ensuring data security

• Disable data collection. Most AI tools offer options to opt out of data collection, although data collection is often enabled by default. Finding and activating these options is a fundamental first step in protecting sensitive data in Public Affairs operations. In ChatGPT, for example, model training on your conversations can be switched off in the settings, so that the information shared is not used to further train the model.

%%block_image_and_text1%%

• Limit data storage. Building on the previous point, where data storage is necessary for long-term processes, it is still advisable to limit data harvesting. It is essential to reduce the amount of information stored and limit retention periods, to minimise the risk of breaches and ensure the organisation always complies with data protection laws.
• Use encrypted tools. Finally, it is good practice to favour AI tools that offer built-in encryption, robust privacy policies and compliance with regulations such as the GDPR.

Human oversight of processes and outputs

• Initial oversight of processes. The involvement and initiative of Public Affairs professionals is crucial from the very first stages of any process. Clearly defining AI parameters and objectives, selecting sources and setting analysis criteria helps prevent errors at the input stage.
• Promote an augmented intelligence approach. As set out in our first AI guide, the approach must be one of collaboration between humans and machines. In other words, Public Affairs professionals should foster a collaborative environment in which AI supports, but does not replace, human decision-making.
• Human review and validation of all decisions. All recommendations or analyses produced by AI must be reviewed and validated by Public Affairs professionals to limit the risks associated with bias. While AI can be especially useful for brainstorming and outlining strategies – as described in the section on opportunities – human oversight, and therefore final responsibility, remains essential in the definition and execution phases.

Conclusions

The solutions we have identified are based on greater awareness and knowledge of AI tools, which at ADL Consulting we consider essential to benefit from the innovation these technologies offer.

Alongside these guidelines for everyday tasks, Public Affairs professionals need ongoing training and study of how AI tools work. Only knowledge that combines the technical, regulatory and operational principles of AI makes it possible to reap its benefits while limiting its risks – and, ultimately, to further promote the augmented intelligence approach championed and applied in ADL Consulting’s Digital Lobbying.

———————————————————————————————————————–

¹ BestinBrussels 2024/25, 2024, https://www.bestinbrussels.eu/wp-content/uploads/2024/06/BestinBrussels-2024_25-1.pdf