ChatGPT's Impact Across Disciplines Divides Experts

ChatGPT’s Impact Across Disciplines Divides Experts

The first time I watched ChatGPT write a sonnet about my toaster, I laughed. The second time I watched it draft a legal memo that a junior associate would have been proud of, I stopped laughing. By the third time, I was asking the same question that 43 experts from fields as varied as nursing, tourism, and computer science recently sat down to answer: So what if ChatGPT wrote it?

The question is not rhetorical. It is the title of a sprawling 2023 paper in the International Journal of Information Management, led by Yogesh K. Dwivedi and his colleagues, which collected multidisciplinary perspectives on generative conversational AI. The paper is less a unified verdict than a snapshot of a field in chaos. Some contributors see a productivity revolution. Others see a privacy nightmare. A few see both, at the same time, and cannot decide which feeling should win.

What makes the paper worth reading is not that it settles anything. It is that it shows how deeply unsettled the experts themselves are.

The 43 Voices That Could Not Agree

Dwivedi et al. (2023) did something unusual. Instead of writing a single-author opinion piece or running a small focus group, they invited 43 experts from different disciplines to contribute short essays. The result is a document that reads less like a typical academic paper and more like a panel discussion where nobody gets the last word.

The contributors came from computer science, marketing, information systems, education, policy, hospitality and tourism, management, publishing, and nursing. Each was asked to reflect on the opportunities, challenges, and implications of ChatGPT for their field. The authors then grouped the responses into three thematic areas: knowledge, transparency, and ethics; digital transformation of organizations and societies; and teaching, learning, and scholarly research.

The first thing you notice is the range of enthusiasm. Some contributors sound like early internet evangelists. Others sound like they are writing a warning label.

The Enthusiasts: Productivity Gains Are Real

Several contributors argued that ChatGPT is not just a toy. It is a tool that can meaningfully reduce the time spent on routine cognitive work. In banking, for example, the authors noted that ChatGPT could handle customer inquiries, generate reports, and even assist with compliance checks. In hospitality and tourism, it could draft personalized travel itineraries or respond to guest feedback.

The information technology sector was flagged as a particularly strong candidate for gains. The reasoning is straightforward: a lot of IT work involves writing code, documentation, and troubleshooting scripts. ChatGPT can do all three, albeit imperfectly. The authors found that contributors across multiple fields acknowledged ChatGPT’s capacity to enhance productivity, especially for tasks that involve generating text from templates or patterns.

But here is the catch. The authors also reported that opinion was split on whether ChatGPT’s use should be restricted or legislated. Some contributors wanted strict regulation. Others argued that overregulation would kill innovation before we understood what we had.

The Skeptics: Bias, Misinformation, and the Black Box Problem

The skeptics had their own evidence. The paper documents concerns about biases embedded in training data, the potential for misuse, and the difficulty of verifying the accuracy of generated text. These are not abstract worries. They are grounded in the mechanics of how ChatGPT works.

Large language models are trained on vast datasets scraped from the internet. Those datasets contain human biases, sometimes amplified. The authors noted that contributors raised concerns about generative AI reproducing stereotypes or generating harmful content, especially in sensitive domains like healthcare and law.

Then there is the misinformation problem. ChatGPT writes with confidence, even when it is wrong. It does not say “I am not sure.” It says “Here is the answer.” The authors found that contributors across disciplines flagged the risk of users treating generated text as authoritative without verification. In nursing and healthcare, that could be dangerous. In education, it could undermine learning.

The Ambivalent: We Do Not Know What We Do Not Know

Perhaps the most honest responses came from contributors who admitted they could not predict the long-term impact. The paper identifies several open questions that cut across all three thematic areas:

• ▸What skills, resources, and capabilities are needed to handle generative AI?
• ▸How do we examine biases attributable to training datasets and processes?
• ▸Which business and societal contexts are best suited for generative AI implementation?
• ▸What is the optimal combination of human and generative AI for various tasks?
• ▸How do we assess the accuracy of text produced by generative AI?
• ▸What are the ethical and legal issues across different contexts?

These are not rhetorical questions. The authors explicitly call for further research on each one. That is a rare admission in academic writing. Usually, papers pretend to have more answers than they do. This one does not.

Why the Debate Matters for Your Job

If you work in knowledge work, this debate is not academic. It is about whether your job changes, stays the same, or disappears.

For Managers and Marketers

The paper suggests that ChatGPT can enhance business activities like management and marketing. It can draft emails, generate social media posts, and even analyze customer sentiment. But the authors also note that it cannot replace human judgment. A manager who outsources strategic thinking to ChatGPT is making a mistake. The tool is good at patterns. It is bad at context.

For Educators

Education is where the debate gets heated. Some contributors saw ChatGPT as a threat to academic integrity. Others saw it as an opportunity to rethink assessment. If a student can ask ChatGPT to write an essay, then the traditional essay format may no longer test what we think it tests. The authors found that contributors raised questions about how to redesign assignments, how to detect AI-generated work, and whether detection is even the right goal.

For Healthcare Professionals

Nursing and healthcare contributors were among the most cautious. They acknowledged that ChatGPT could help with documentation and patient education. But they also flagged the risks of incorrect medical information and the lack of accountability. If a chatbot gives bad advice, who is responsible? The developer? The hospital? The nurse who used it?

For Researchers

Scholarly research is another flashpoint. ChatGPT can write literature reviews, summarize papers, and even generate hypotheses. But the authors noted that contributors were divided on whether it should be cited as a co-author, whether its use should be disclosed, and whether it could produce original insights. The paper does not resolve these questions. It just makes them visible.

What the Research Does NOT Prove

This is important. The paper is an opinion piece. It is not a randomized controlled trial. It does not measure the actual impact of ChatGPT on productivity, learning, or ethics. It collects expert opinions and organizes them.

That does not make it useless. Expert opinion is valuable when the evidence base is thin, which it is. But readers should not mistake consensus for proof. The authors themselves note that opinion is split on key issues. That means the paper is a map of disagreement, not a destination.

The paper also does not address long-term effects. ChatGPT has been widely available for only a short time. The contributors were speculating about impacts that may take years to materialize. Some predictions will be wrong. The question is which ones.

The Three Battlegrounds

The authors grouped the open questions into three thematic areas. Each is a battleground where the future of generative AI will be decided.

Knowledge, Transparency, and Ethics

This is the most philosophical area. It asks: What does it mean for a machine to generate knowledge? Can text produced by an AI be considered original? Should it be transparent when it is AI generated? The authors found that contributors raised concerns about transparency, accountability, and the potential for misuse. But they also noted that some contributors argued that transparency requirements could stifle innovation.

Digital Transformation of Organizations and Societies

This is the most practical area. It asks: How will organizations change when they can automate text generation? Which industries will be disrupted? Which will benefit? The authors found that contributors identified banking, hospitality, tourism, and IT as likely winners. But they also warned about job displacement, especially for roles that involve routine writing.

Teaching, Learning, and Scholarly Research

This is the most urgent area. It asks: How should education and research adapt? Should we ban ChatGPT, embrace it, or something in between? The authors found that contributors were deeply divided. Some wanted strict bans. Others wanted to integrate it into the curriculum. A few argued that the debate itself was a distraction from bigger issues like inequality in access to technology.

What This Actually Means

The paper does not give you a checklist. It gives you a set of tensions. Here is what those tensions mean for anyone who uses, builds, or regulates generative AI:

• ▸If you are a manager, do not treat ChatGPT as a replacement for judgment. Use it for drafts, summaries, and pattern recognition. But verify everything. The tool is confident even when wrong, and confidence is not accuracy.

• ▸If you are an educator, rethink assessment before you try to block the tool. Detection software is an arms race, and the authors found that contributors were not confident it would win. Instead, design assignments that require process, not just product. Ask students to explain their reasoning, to critique the AI’s output, to show their work.

• ▸If you are a policymaker, resist the urge to regulate too fast or too slow. The paper shows that experts disagree on whether restriction or legislation is appropriate. That suggests a middle path: monitor, learn, and adjust. Do not lock in rules based on early panic or early hype.

• ▸If you are a researcher, be honest about disclosure. The authors found that contributors were divided on whether ChatGPT should be cited as a co-author. But most agreed that its use should be transparent. If you use it, say so. If you do not, say that too.

• ▸If you are a user, remember that ChatGPT is a mirror. It reflects the data it was trained on, which includes your biases, your stereotypes, your worst impulses. The paper warns about this. The tool is not neutral. It is a product of its training, and its training is a product of us.

Dwivedi et al. (2023) end their paper with a call for further research. That is not a cop out. It is an honest acknowledgment that we are early in this story. The 43 experts could not agree on what ChatGPT means. That is not a failure of the paper. It is a reflection of reality.

The question “So what if ChatGPT wrote it?” does not have a single answer. It has 43. And the debate is only getting started.