Confessions   //   June 7, 2023

‘Mix of excitement and guilt’: Confessions of a covert AI user in the workplace

In the tranquility of their home, an aura of curiosity and excitement filled the air. A remote worker sat at their desk, contemplating a clandestine mission. Unbeknownst to their colleagues, this individual was about to embark on a hidden collaboration with artificial intelligence.  

Motivated by the desire to make a tedious and laborious task easier, they secretly used the generative AI tool ChatGPT to write their part of the company’s annual report. With a few keystrokes, they turned the once burdensome task into a breeze. 

While some businesses have openly encouraged their workforces to experiment with generative AI tools, others have taken the opposite stance. For this WorkLife Confessions installment, where identity is exchanged for candor, we spoke to a senior ad executive at a publicly listed advertising company, which hasn’t approved the use of generative AI at work, on how they’ve used ChatGPT to speed up tasks covertly.

Answers were edited for clarity and flow.

Had you considered using ChatGPT before to complete work?

I had read about the growing influence of AI in the workplace, but I hadn’t seriously considered using it for my work until the annual report landed on my desk. The sheer magnitude of the task and the knowledge that it would require significant time and effort to complete made me think otherwise. The thought of outsourcing the report to an AI not only piqued my curiosity but also promised a potential way for me to save valuable time and energy. The allure of testing the capabilities of ChatGPT to handle the writing process was too tempting to resist. 

Were you not worried about getting caught? 

Absolutely, the thought of getting caught certainly lingered in the back of my mind. The duplicitous nature of using AI to write the report created an awkward mix of excitement and guilt. I was well aware that if my secret collaboration with ChatGPT were exposed, it could have serious repercussions, both professionally and ethically. However, I took precautions to ensure my actions remained covert, using the AI tool discreetly and avoiding any traceable connections to my work.

Precautions like what?

First and foremost, I made sure to work discreetly from my personal device, separate from the company’s network or any other monitored systems. This helped me maintain a level of anonymity and prevented any direct connection to ChatGPT. Additionally, I used a VPN [Virtual Private Network] to further safeguard my online activities and maintain privacy. I also cleared my browsing history, regularly deleted any AI-related files or conversations and avoided discussing the matter openly especially in digital communications or on work-related platforms. By taking these measures, I tried to minimize the risk of leaving any traceable evidence behind. While these precautions weren’t foolproof, they provided a level of reassurance and allowed me to use the generative AI with discretion. 

How easy was it to get ChatGPT to write the section of the annual report in the manner you wanted?

Surprisingly, it was a relatively straightforward process. I was pleasantly surprised by the AI’s ability to understand the context and generate coherent content. The initial setup involved providing the AI with a clear prompt, outlining the purpose and key points I wanted to convey in the report section. 

That said, some of the responses did need some fine-tuning to align what the AI produced with the tone and style suitable for the company’s annual report. Through trial and error, I gave feedback to ChatGPT and made adjustments to the prompts, guiding the AI to produce the final version. While there was an initial learning curve to optimize the AI’s output, overall, the process was relatively smooth and easy. The AI was a valuable tool, saving me significant time and effort in the writing process. 

Were you not worried about ‘hallucinations’ injecting errors? 

I was definitely concerned about the AI providing incorrect information. To address that I took some precautions. I double-checked the AI generated content against trusted sources and internal data to ensure accuracy. I didn’t blindly accept everything the AI produced. I used my own judgment and fact-checked the information. Before finalizing the report section, I carefully reviewed and edited the AI’s output to make sure it matched the company’s messaging.

The organization hasn’t explicitly embraced or integrated AI into their official workflow or policies. So I didn’t want to open myself up to unnecessary risk. The covert approach was necessary to avoid any potential backlash.

Would you use it secretly again?

Reflecting on my experience, I must admit that the sneaky aspect to using AI secretly did make me feel uncomfortable. While the convenience and efficiency it offered were undeniable, the feeling of conducting a covert operation in my own workplace created a sense of unease. Moving forward, I would weigh the potential benefits and risks more carefully. If the circumstances align, and there is a genuine need to streamline a tedious task while maintaining ethical boundaries, I might consider using AI again. Exploring the possibility of integrating AI tools officially within the company’s workflow, with proper authorization and guardrails would be an ideal solution. This way, the benefits of AI could be harnessed while maintaining integrity and trust within the workplace. 

Ultimately, it’s essential to strike a balance between efficiency, innovation and ethical practices. While the allure of secrecy may be tempting, it is crucial to weigh the ethical implications and consider alternative approaches that align with the values and policies of the organization.