ISTM #57. ChatGPT versus Me
My first book, Naked Conversations, was written in 2005. It warned business decision-makers to adapt to the accelerating changes being fomented by disruptive technologies such as social media or be left behind.
Over the years, more and more disruptive technologies have assaulted business decision-makers at increasingly faster paces. This presented the opportunity for me to keep writing and talking about them, and I wrote five more books about game-changing technologies.
But then AI came to the forefront, and when I tried to write a book about it, changes came faster than I could write about them, and any book I wrote would be outdated before I could publish it.
That turned out to be a minor problem compared with the bigger one that AI has forced me to face. In all these years of writing about disruptive technology, it never dawned on me that I could be among what would be disrupted.
But that is precisely what happened just after Thanksgiving 2022 when ChatGPT3 (GPT) was introduced without OpenAI, the developers, the developers making much noise about it. They didn’t have to. Word-of-mouth about it spread faster than the Maui wildfire. It had one million downloads in its first week and reached 100 million active users in just two months, making it among the most rapidly adopted software applications ever.
GPT has continued to grow at an astounding rate, as have other large language models (LLMs). These models impact dozens of industries and reshape how businesses, educators, and creators approach natural language processing tasks.
An Uber Future?
I started fiddling with GPT shortly before Christmas 2022, making me about the 50 millionth adopter. I wanted to be among the first business writers to post about it, but I found myself speechless. While I had written about dozens of disruptive technologies over the past 15 years, this one threatened to disrupt me, abruptly changing my attitude toward it.
In short, I was afraid. I wondered if my future might involve driving a car with an Uber sticker on the windshield or delivering orders to tables in the restaurant service industry.
I did not like this vision. After years of writing about digital disruption, I understood that resistance to innovation is futile. I knew I had to adjust course, but how? As I pondered, my business inquiries tapered off, and I felt a sense of urgency that I had not experienced in many years.
I talked with other writers who told me they were facing similar situations, which troubled me even more. If I were doing something wrong, I could fix it. If technology was obsoleting my work in the same way, the village blacksmith was replaced by the automobile. Even the best blacksmiths found they no longer had work, and so it might be the same for writers like me.
Good Enough
I received encouraging news when an old friend contacted me and asked me to do a small writing project for him, but a short while later, he emailed me to say he had decided to use GPT instead of me.
“You may write better,” he told me, “but GPT is faster, cheaper, and good enough.”
So, I started pitching myself as a writer to contract when good enough was not good enough. Friends complimented my cleverness, but I landed no deals. Apparently, the idea was not good enough.
By this time, AI platforms similar to GPT began popping up like fireworks at a bicentennial celebration. The best known are Google Gemini (formerly Bard), Microsoft Bing, Jasper, Co-pilot, Claude, Perplexity, and Blender. Together, they formed a new category that is being called AI large language models, or LLMs for short.
Step Back
At this point, I realized I needed to take a step back and take a longer look at a problem that clearly was not going away. I saw the irony of having built a career writing about disruptive technologies only to find myself being disrupted by new technology—but I was not amused.
I turned to my friends, who were sympathetic but as clueless as me about what I should do. I revisited one of my favorite books from college, The Art of War, which had taught me to know my enemy as well as I knew myself.
I stopped trying to find flaws in GPT and began considering ways to use it to make myself more valuable. I began to regard it as a powerful new tool. I am still working on it, but now LLMs have become an essential power tool for almost everything I write, including this post.
LLMs allow me to produce more content in less time than previously possible. This means I can reduce my client fees while maintaining the same margins and I can accommodate more clients.
In fact, GPT and other LLMs are my new best friends. I don’t conduct writing assignments work without them.
Power Tools
I have tried all the LLMs mentioned above and like each one. Each seems to have one or more unique and useful features, but as a lover of simplicity, I have gravitated to GPT and Gemini. They were the first two I tried, and I have not found anything clearly superior to them, so I will keep using them until something more disruptive comes along.
For most of my work, I use them both simultaneously. The technology is still nascent and flawed and not always accurate, as both products will repeatedly remind you.
So, I cross-check most of my content between them and have stopped fiddling with other entries. At least for now, GPT and Gemini are my two power tools.
Among my everyday uses:
· Review ideas. I feed these two LLMs short paragraphs about what I plan to write. I ask them if they will be useful or interesting to a target audience such as entrepreneurs, investors, educators, or students. I refine the idea until both LLMs agree that my content is a good approach.
· Second drafts. While many users feed outlines or a few sentences to LLMs and ask for a first draft, I have been disappointed with those results. I write my own first drafts and feed them to my LLMs, who edit and revise them, which works better for me—but perfectly. While both are good, I prefer Gemini because it treats my copy the way good developmental editor would. It asks me questions and gives me ideas to conside.
· Editing. I have both my LLMs edit, fact-check, and add comments or questions to my drafts. While they are far from perfect, they are useful in catching things my Grammarly is not capable of, such as clarity, conciseness, context, consistency, and jargon. I find Gemini better than GPT in this case. ChatGPT tries to rewrite my content and sometimes does a poor job of it, while Gemini acts like a content editor, re-arranging paragraphs, suggesting content that will improve clarity, and asking tough questions that remind me of my late friend, Charlie O’Brien, the best newspaper editor I ever had.
· Research. This is the silver bullet for LLMs. It saves me lots of time, providing data I might not have found otherwise. For example, for a book I’m working on, my client told an engaging story that took place in a Silicon Valley bar where engineers gathered after work in the 1970s and shared gossip. I asked Gemini to name this tavern from 50 years ago, and it did so in seconds. It was called the Wagon Wheel and reminded me that its bartender, ‘Tender Wally’ Trabing, was said to know more trade secrets than anyone in Silicon Valley. This is juicy meat for writers like me.
How much time do I save? I really don’t know, but my wildest guess is that it is between 20 and 40 percent. What I can share is that I have not published a significant factual error since I started using the LLMs. When the two that I use disagree, which happens less than ten percent of the time, I try a third LLM, (usually Perplexity) or just delete the questionable content.
The Human Element
Many times in my career, I’ve been required to read McKinsey reports, which are credible, accurate, and universally regarded as accurate analyses of trends and strategic insights from a broad and objective perspective.
I have also found McKinsey reports to be a cure for insomnia. Their flat, balanced, detailed tone is about as exciting as the fine print of terms on a maintenance contract.
Most business writing needs to be more interesting, demonstrating passion, humor, or irony so that readers are compelled to want more of what is being offered.
AI may someday be able to deliver these attributes, but not yet and good business writing will probably need the human touch for a period that is longer than I have to worry about.
Limited Time
Because of LLMs, I can take on more projects, charging less money and completing them in less time. I can also take on more clients, and my profit margins will remain as they were before LLMs began to disrupt my work.
But my current competitive edge may be temporary. According to people more knowledgeable than me, LLMs will eventually eclipse my current advantages, but not for several years.
I asked Chat GPT and Gemini about it. Here’s what GPT told me:
“As of now, while I can attempt to write with emotion, honesty, irony, and tell stories, my output might not always match the subtlety or depth of a human writer.
This gap between AI LLMs and human writers may narrow with time, but for now, we human writers have a competitive advantage. It may be for a limited time only. Me? I’m going to enjoy it for as long as I can.”
+++++
Note: I offer one free hour of consulting to anyone planning to write a business book. I do this because it makes me better at my ghostwriting projects. Contact: shel@shelisrael.com 6504304042.
"In short, I was afraid. I wondered if my future might involve driving a car with an Uber sticker on the windshield or delivering orders to tables in the restaurant service industry."
Interesting. Two professionals already targeted by AI and its partner autonomy. Robotaxis are already running, as you know, and robots at the table take your order in many restaurants; delivery comes next. --Michael Coates
Shel, I think this is exactly the right mindset! Our tools got so much more powerful, but they are still just tools. Clients who think that they can neglect the human in the equation will save money now, but will have bland content later. But AI makes you more productive and that's what will lead to much better content in a shorter period of time.