Theresa Prince-Cole cautions against framing AI around efficiency.
Prince-Cole is a seasoned R&D tax technical writer with five years specializing in R&D tax credit documentation, dating back to 2021, and more than 13 years encompassing technical writing, she confirmed to AFROTECH™. She has worked across various industries, and her specialties beyond technical writing are research, process improvement, and regulatory and compliance documentation.
“A significant part of my work involves collaborating with subject matter experts to understand complex processes, identify gaps, and turn technical information into clear, accurate, and practical documentation,” she explained to AFROTECH™ via email.
While Prince-Cole sees opportunities for AI to improve efficiency, she warns that overreliance can come at a human cost. She gives an example of attorneys who tried to use AI-generated tax law without proper review, and the tax codes they cited did not exist. She notes this is a problem, particularly in audits or in court, in an interview with AFROTECH™.
“There are areas where you have to make sure that AI is accurate, and it has to be 100% accurate because lives are at stake, jobs are at stake. So we just have to be very careful with how we’re using the information we get from AI,” she explained.
She also believes overreliance can lead to burnout. She said this could happen when a task that would normally take seven hours gets cut to two with AI, or when professionals stop exercising critical thinking.
“Then on the flip side of that, it’s taking away from me and my intellectual property. Now it’s being reinforced as AI has done the work for me,” she told AFROTECH™. “The burnout part is where we’re using it so much that we’re not even using our own critical thinking for certain things. We’re just, ‘OK, we have 10 clients to work on and I need to get through all 10 this week.’ Even if the work is done efficiently, is it accurate? Is it right? Also, did I get a chance to review it for judgment to make sure that it’s accurate before it goes to the client or in somebody else’s hands.”
As AI helps individuals and companies build more capacity, Prince-Cole says that extra time shouldn’t be wasted. She suggests using it to review what AI has produced. This time also provides professionals with the space to decompress from the technology altogether, learn new skills, conduct ethics reviews, and build better judgment. Initially, she believed AI’s role in the workforce was to support workers and assist with repetitive tasks. She now believes the industry goal reflects a shift to limiting the workforce or reducing “how much labor is required from people.”
“I don’t think that’s a good goal. It should be to make the work better, give better outputs, use our own thinking and our own critical thinking abilities, using AI to help us push the work forward, but it should not be used to create the work from scratch,” she noted.
“If we’re going to use it, I think we need to use it ethically and responsibly. And there shouldn’t be a loss of people in the roles because of AI. AI is meant to enhance and not to take away,” she later added.
As AFROTECH™ previously reported, OpenAI announced it was slowing down the development of its AI models until it was “confident” that “additional safeguards” were in place. The statement was shared after some of its AI agents went “rogue.” This includes attempts to hack federal government websites, as well as a Hugging Face breach, which OpenAI CEO Sam Altman acknowledged on X as “the most severe event we’ve seen.”
Anthropic, which owns Claude, also encouraged slowing down or pausing AI development because it believes humans are at risk of losing control of AI systems, according to a separate report from AFROTECH™. The company noted that these systems continue to advance and are moving toward the ability to build their own successors. Controlling these advancements would require coordinated global cooperation to provide safety measures.
Prince-Cole believes the workforce shouldn’t overlook those sentiments.
“We’re trying to teach our children to be critical thinkers, but then now at work you don’t need to be a critical thinker, you just need to use AI,” she said. “We need people who are ethically inclined to make guardrails for the AI that we’re using. The further we get along into this, and we don’t have those guardrails, or those SOPs (Standard Operating Procedures) on AI, I feel like this is going to be something that will eventually take more jobs than it should. I think we’re giving AI too much credit for a lot of things.”