ChatGPT’s Impact on Thought Leadership: Time to Raise Your Research Game

Just four months after OpenAI unveiled its artificial intelligence-based writing machine ChatGPT to the world, numerous human writers have publicly wondered whether it will replace them. I imagine many more have the same fear. I’ve seen the prose that pours out of ChatGTP after typing my questions into its question bar, and I’ve seen how it responded to other people’s queries. In short, I’m stunned, like many others.

You may be thinking I shouldn’t be so impressed: So much of what ChatGTP spews out is wrong – the wrong years of people’s birthdates, wrong places where they were born, wrong this and wrong that. (I saw six errors in a client’s bio I recently asked ChatGPT to write as an experiment.) That’s hard to ignore. How can anyone rely on a technology that spits out so many incorrect facts and doesn’t provide sources? How much time will ChatGTP really save if humans must check every fact for accuracy, and every passage for plagiarism?

Nonetheless, I assume that as more people use the system and point out its errors, it will correct many mistakes. Hopefully, the software will also start favoring sources that are right most of the time and ignore those that are consistently wrong. But even if the newest version of ChatGPT (GPT-4) isn’t a big improvement, other generative AI programs have attracted $6 billion in venture capital investments in the last five years, according to tech researcher CB Insights.

My immediate interest in the technology is not about whether writers will have to find other professions. But I do suspect many firms will replace marginal writers with the AI because it writes better, cheaper and much faster prose. But companies will still need people to fact-check, copyright-check and plagiarize-check what comes out of these artificial scribes, and that will cost money.

Instead, since my firm helps B2B companies codify their expertise, develop new expertise and become known for all that, I am interested in something different: How will ChatGPT change the way they develop primary research-based “thought leadership” content?

Most people, myself included, can’t begin to fathom the ultimate impact of ChatGPT and other generative AI technology on the thought leadership discipline. The technology is new; it’s evolving quickly; and its biggest beneficial impacts are likely to happen much later. While I have used ChatGPT and read about other people using it, I don’t have a deep understanding yet of its value in developing primary research-based thought leadership content. But in this early stage, here is what I believe:

• Thought leadership researchers will have to raise their game dramatically. Remember when you couldn’t find a great writer to disguise the fact that your study’s insights weren’t groundbreaking? Everyone can now publish well-written studies with ho-hum findings. ChatGPT makes that possible. However, it won’t create revelatory findings – groundbreaking insights – that determine whose research attracts clients and market influencers. (Of course, you need great marketing, too. But that alone cannot save lame or “me-too” insights.)
• Many ghostwriters must improve their skills and gain new ones. I think ChatGPT will eliminate the need for writers who simply write what the researchers tell them to write – what I see as just a step above stenography services. You will have to write compelling prose – copy that uses analogies, metaphors, little-known facts and other things that ChatGPT won’t easily spit out from its digital brain. But even talented writers will have to raise their game. Creating highly readable prose from input the experts give you will no longer be enough; you must be skilled at pushing the experts’ thinking into rigorous and novel arguments.
• Desk researchers will have to master using ChatGPT to track down hard-to-find data. The system, and other generative AI programs that scour the web, will be a powerful desk research tool.

I’ve come to these initial conclusions after some thought (but not with the help of ChatGPT, I will note): looking at how ChatGPT could help in each step of what we at Buday TLP see as the six-stage process for research-based thought leadership content. (If you want the deep drill on this process, read my book, “Competing on Thought Leadership.“)

My Dissection

Stage by stage, here are my brief thoughts on how ChatGPT and other generative AI programs could change the way companies develop research-based thought leadership content:

Research Design

It will help researchers know what the current thinking is on a topic. That should help them develop issues to research that haven’t been researched before, and questions to ask that haven’t been asked before.

Data Collection

Generative AI will help researchers find companies to study, and statistics from other studies that provide additional support of key assertions.

Data Analysis

My thinking is still developing on this, but I can’t immediately see how generative AI will help researchers identify patterns in voluminous amounts of complex qualitative and quantitative data. Also, since they are entering such data into ChatGPT and asking it to interpret the data, I’d worry about making that data available to other researchers from other companies. To me, it’s not worth that risk.

Argument Development

This is about turning big insights on a topic into a compelling narrative for the research report. If you have read my book or my posts, you are probably familiar with my six-part narrative structure for thought leadership content. This should be the work of your researchers and the writers who turn their findings into prose.

Here again, as in the data analysis stage, I would be reluctant to ask any generative AI program to do this, fearing it will learn from my analysis and hand it to others who are researching the same topic. Another user of ChatGPT will then beat me to the punch.

Concept and Data Viz

I don’t know enough to comment intelligently here, except to say these programs exist – Jasper, DALL-E and many more. I haven’t used them much, but for the topic of this article, I did ask DALL-E to give me an impressionistic rendering of a robot interviewing an executive. It was just OK. But I imagine the problem is not with DALL-E here, but with the person who entered the query — me.

Writing

Generative AI programs like ChatGPT will enable even researchers with mediocre writing abilities to produce solid prose. Just feed the narrative outline into the AI, tell it to write prose from the outline, and see what comes out. But I’ll caution again that they’d need to have an account with ChatGPT that walls off the company’s inputs from the outside world.

The work of writers here will be to a) make the ChatGPT prose more compelling, and b) make sure any extraneous facts that may have found their way into a research report are relevant and true. But if you largely make poorly communicated ideas readable, ChatGPT and other generative AI programs may replace you — unless you quickly become a much better writer. Ethical companies that care about accuracy and avoiding the legal risk of plagiarism may hire you to check their ChatGPT-written copy. But do you want that as a career? The new skill to learn here is about pushing the thinking of subject experts and turning their ideas into powerful arguments. But the argument-development skill is different than the prose-writing skill.

And what does ChatGPT itself think of all its impact on thought leadership content development? After developing the ideas for this article, I asked the AI itself. It looks like we’re mostly in synch – other than the “idea generation” piece.

All in all, I see ChatGPT forcing B2B companies to create bigger insights from their thought leadership research on how best-practice firms are solving the problems at hand. In turn, that will require collecting better evidence of such practices, which in turn will require talking to executives in those best-practice firms, not just getting their one-dimensional answers to structured surveys. It will also require researchers to make better sense of qualitative and quantitative data. I’ve seen this happen more often when data analysis is a collaborative rather than competitive process in a research team. But better analysis begins with better data, which in turn requires better designed research, often by narrowing the scope of a study so researchers can plunge deeper than others have gone.

That’s my early read on ChatGPT’s impact on thought leadership content development. I’d be very interested to hear your thoughts.

(I’ll weigh in at a later date on how I see ChatGPT changing how B2B companies capture and publish the expertise they gain from their client work.)