Chapter 1: An Unexpected Encounter with ChatGPT

Recently, I found myself experimenting with ChatGPT, despite my previous reluctance to engage with it. My curiosity was piqued by a simple question: what does it know about me?

One might assume that, due to my lack of celebrity status, ChatGPT would have little information to share. The AI is designed to protect personal data, like phone numbers or addresses, which are easily searchable elsewhere. However, I had a notable presence online during the early web days. My website, aplawrence.com, garnered a significant audience, attracting over 250,000 unique visitors monthly with its focus on Unix and Linux systems.

Don’t waste your time searching for it now; I sold the site years ago. It’s fallen into disarray, as the new owner has neglected to maintain it, and the content is outdated. Unix has taken a backseat to Linux, making my old site less relevant.

While I may not be a household name, I did experience a fleeting moment of recognition at a tech conference when someone exclaimed, “Wait — are you THAT Tony Lawrence, the Unix guy?” However, such moments are rare, so I wouldn’t call myself famous.

Given my long history of writing online since the 1990s, I wondered if ChatGPT could still recognize me, even without widespread acclaim. I began by asking it about aplawrence.com.

To my surprise, ChatGPT responded:

I’m sorry, but I cannot find specific information about a website called “aplawrence.com,” as it may be personal or non-existent.

Oh, the irony! My past acclaim has seemingly faded into obscurity. Undeterred, I next inquired, “What about Tony Lawrence and Unix?”

This time, the results were more promising.

Tony Lawrence is recognized as a computer consultant and author with expertise in Unix systems administration. He has penned several books on Unix and Linux and maintains a collection of articles and tutorials on his website, aplawrence.com, covering various topics in the field.

While it was nice to see my work acknowledged, I no longer own that website, nor do I provide consultancy services.

Then things took an unexpected turn.

As I pressed for more details, ChatGPT elaborated:

Tony Lawrence has over 30 years of experience with Unix systems, focusing on administration, troubleshooting, and performance tuning. He has authored several respected books on Unix and Linux, including resources on backup and recovery and performance tuning.

While this paints a flattering picture, it was laden with inaccuracies. In truth, while I have been involved in Unix for decades, I did not write those specific books; my contributions are different.

Interestingly, I received a royalty statement from Amazon recently, revealing just a couple of dollars earned from my books. Their sales have dwindled significantly over the years.

Yes, I was well-regarded in the Unix community, especially concerning SCO Unix, but my reputation in Linux was not as pronounced. I have participated in some public speaking and training, but never specifically at a Unix or Linux conference.

So, what’s my overall impression of ChatGPT?

I find it difficult to place trust in its responses, and I advise caution to others as well. When Google launches its own AI tool, I plan to repeat this experiment.

Chapter 2: The Nature of AI Knowledge

This exploration of ChatGPT’s capabilities has shed light on the complexities of AI knowledge and its limitations.