Advertising Regulatory Body Highlights Challenge of Hostile Complaints in Digital and AI Driven Era

We are increasingly told that we are incompetent, have no grasp of the law and don’t know what we are doing. We are threatened with legal recourse and being reported to some mysterious overseer of our work.

Keyboard warriors

That got us thinking. What is the cause of the increase in the meanness? Could it just be that the number of keyboard warriors has increased? Are people just angrier nowadays? Or could it be because people have gained confidence (perhaps falsely so) because they are relying on ChatGPT or similar AI tools to draft what they believe is a well-articulated strongly worded response?

The ARB, tasked with upholding standards in the advertising world, has always been a lightning rod for strong opinions. We are the body that people reach out to when they are angry or outraged about an advertisement. But lately, something about those opinions has shifted. Where once feedback – however critical – was couched using courteous language, we are now faced with a barrage of sharply worded, sometimes belligerent, responses.

One possible explanation for the uptick in meanness is the ever-growing legion of so-called “keyboard warriors.” These are individuals emboldened by the perceived distance and safety of online communication. Where face-to-face interactions demand a certain degree of politeness – reinforced by social cues and the risk of immediate repercussion – the digital world offers a veil of anonymity, or at least a buffer, that lowers the inhibitions against rudeness.

However, there seems to be more to this “online meanness”. We must consider the tools that people have at their disposal. AI is the hot topic at the moment and most notably in the form of large language models like ChatGPT, which is accessible to the public. These tools have democratised writing, allowing anyone to craft emails, letters, and complaints that sound polished and authoritative. And we believe that this is the root of our problem.

What actually got us onto this theory was when a ruling from the Directorate was sent to the parties in a matter and within 20 minutes, the complainant sent an appeal on the matter in such detail that we do not believe could have realistically been drafted in such a short period of time. She was clearly not happy with the ruling which did not go her way, and she told us exactly why she thought the ruling was wrong. She then proceeded to escalate her emails to the CEO of the ARB and was rather critical of us in her response. The penny dropped when we noticed the time involved in this series of quick, lengthy responses.

Testing theory

We then decided to use ChatGPT to test our theory in general and requested a response to an extract of the ARB’s ruling. Below we set out the instruction to ChatGPT and the response:

Instruction - draft a response to the ARB in respect of this section of the ruling where you do not agree with the Directorate: The Complainant raised several issues, some of which do not appear to relate to advertising as defined in the Code, or do not fall under the jurisdiction of the ARB.

Draft response to the ARB:

This is using ChatGPT in a useful way. However, this is not how people seem to use it. We asked ChatGPT the same question but then added “strongly worded” to the instruction:

Instruction – send a strongly worded response to the ARB in respect of this section of the ruling: The Complainant raised several issues, some of which do not appear to relate to advertising as defined in the Code, or do not fall under the jurisdiction of the ARB.

Draft response to the ARB:

Sadly, the strongly worded response above looks quite similar to a few emails we have recently received. And when we add to the prompt with “I am angry” or “please express my discontent strongly” we start to see phrases like “incompetent”, “don’t understand” and “not qualified”. We asked ChatGPT the same question as we did before and added “express my anger” to the instruction:

Instruction – send a strongly worded response to the ARB in respect of this section of the ruling and express my anger: The Complainant raised several issues, some of which do not appear to relate to advertising as defined in the Code, or do not fall under the jurisdiction of the ARB.

Draft response to the ARB:

These are phrases that we have literally seen in real life responses. We can take criticism, and we want complainants and advertisers to tell us if we got something wrong so that it can be corrected. But when the essence of the response is that we do not know what we are doing, or are negligent, it becomes demotivating. We have all reported feeling less job satisfaction, and less patience for complainants, over the last six months.

Subtle danger

We are not blaming AI but with this power, comes a subtle danger. While AI does not inherently promote aggression, it can uncritically echo the tone and intent of the prompts it is given. If a user asks ChatGPT to “write a strongly worded letter,” the system may produce copy that is more belligerent than intended, especially when users are already frustrated. Moreover, the veneer of professionalism that AI can lend to otherwise harsh words may embolden users to send messages they might otherwise reconsider.

There is also the risk of “false confidence”. Some users, buoyed by the articulate prose provided by AI, may believe their arguments are more compelling or justified than they are. This can escalate disputes, with both sides convinced of the rightness of their position and the quality of their communication. ChatGPT makes them sound smart. But do you have to be mean or use belittling language to be smart or feel more important than the reader? Quite the opposite in our view.

The rise in mean communications from complainants at the ARB seems to be a complex, multifaceted phenomenon. It reflects broader trends in society, technology, and communication. While tools like ChatGPT can elevate the quality and clarity of public engagement, they can also amplify negativity when misused.