I Finally Played with the AI Robots and Here’s What I Found

The end of human creativity is near if you listen to the buzz around the latest AI innovation, widely known as ChatGPT. AI has been in the news everywhere, generating everything from mild panic, to confident hopefulness to sardonic nihilism for the past couple of years, but ChatGPT seems to be everywhere at the moment.

I’ve long had my own thoughts on anything AI related and my position is the same as it has always been. AI is a useful tool to help with productivity, but it’s not a replacement for any human endeavor. So, when a client of mine asked about it for some technical writing, I gave them the knowledge I had, and undertook a little experimentation.

All in the name of science and keeping a satisfied client, well, satisfied. I created my own ChatGPT account, and I started posing questions to it, testing the limits and the boundaries. Here’s what I’ve found.

Citing Sources is problematic, unless you specify the source you want.

For example, I asked it to tell me about religious fasting practices across the major world religions. It was a wisp of an idea I had for a blog post, which I may or may not use at a later date. Right off, there were a couple of issues. I specified fasting seasons, like Lent, Ramadan, and Yom Kippur, for instance.

Well, there is a Buddhist fasting season, Vesak, but no timeframe was given. It didn’t say spring, fall, winter, or even name months. It did for all the other religions, but not for Buddhism. Nor did it offer an explanation.

The second issue was that it lumped Christianity all together. Now, if you know even an ounce of history where that is concerned, you’ll know that answer is problematic. Not only do you have the Western (Roman) Church, but you have the Eastern (Orthodox) Church. And they follow different calendars and different practices. Orthodox Lent and Easter don’t start on the same days or even have the same feast days as Roman Lent. And then there’s the entire issue of Protestants vs Catholics.

Most Protestants don’t celebrate Lent unless they are from a high church tradition. Some even go so far as to say Lent is somehow “works righteousness” or “too catholic.”

Anyway, I asked a couple more questions about fasting, the practices, and the health benefits of fasting for my tentative blog post and then I asked it to provide source information for the last set of data it gave me.

It couldn’t. The answer is below.

As an AI language model, I do not have personal beliefs or emotions, nor do I have the ability to use sources in the traditional sense. However, I am trained on a large corpus of text data, including books, articles, and other publications, which allows me to provide information on a wide range of topics, including fasting and mental health.When providing information on a specific topic, I rely on the most reliable and up-to-date sources available in my training data, which includes academic publications, medical journals, and reputable online sources. However, it’s important to note that my responses are based on my training data and are not a substitute for professional medical advice. Anyone with questions or concerns about their mental health should seek help from a qualified medical professional.

Ok, fair enough. But if you’re looking for Chat GPT to replace your human writer, then think twice. If you present a bunch of information to your readership, you want some good, reputable sources to back up what you’re saying, especially if you’re intent on stating an opinion or trying to showcase how your company can solve a problem.

Or talking about medical advice in general.

But, there’s good new, on that front. I asked for Chat GPT to generate a basic safety program outline with OSHA references, and it turned up what I would consider gold.

Not only did it give a fairly cogent, albeit too generalized outline, it cited the relevant OSHA references.

So, it can cite sources, but you have to be specific.

You can make it contradict itself, or seem to, if you don’t know Chat GPT’s definitions for some terms.

I used ChatGPT to help me research and brainstorm for my upcoming month of Virgil. Since Virgil isn’t my own strongest point–Homer was once again the favored “must read” from the Greco-Roman world–I figured why not see what it can come up with.

So, I asked a series of questions such as, “Why did Dante imitate Virgil?” and “What makes Virgil’s style unique?” When I asked that last question, it came back with a multi-part answer but the first point was about Virgil’s depictions of emotional depth and internal struggle. It contrasted this with Homer who, supposedly, focused more on external deeds.

Nifty? Well, not so fast….

Going off of this, you might be expected to believe Virgil was somehow revolutionary in this regard. He’s not. Ask Chat GPT about who first depicted the emotional depth and internal world of characters, and it’s not Virgil that pops up. It’s Homer.

Which, since I’ve read both Iliad and Odyssey, I can say is absolutely true. While we don’t see character’s thoughts, we do see their emotions. They aren’t static and they don’t lack depth. So, why the contradictory information?

Because Virgil is not the first to depict emotional depth, nor does he deviate that far from Homer. Rather, his focus is on Aeneas’ virtue and right to establish the banished Trojans on Italian shores. This means there is a certain amount more of character development than, say, Odysseus’ journey home to Ithaca.

So, if you go just off the information from the AI, you might leap to some very wrong conclusions unless you either ask clarifying questions, or, more preferably, you go compare Homer and Virgil for yourself.

Now, here’s where it gets interesting. I asked about inner monologue–as that is also part of portraying emotional depth. ChatGPT came back with Marcel Proust as the first to make real strides in inner monologue. But, there is a caveat. ChatGPT specifically defined inner monologue as streams of consciousness writing.

What’s the difference? Streams of consciousness mimics how our actual thoughts happen. It’s all jumbled together and, most famously, it’s what makes both William Faulkner and James Joyce required reading sometimes. Not that I’m a great fan of either myself.

But is inner monologue strictly streams of consciousness? No, it’s not. Shakespeare’s soliloquies are also a type of inner monologue. Jane Austen even uses it to a certain extent, although in a very third-person narrative form. J. R. R. Tolkien uses inner monologue. None of these are streams of consciousness writers.

The issue is that “inner monologue” is a generic term that was given a very specific definition. If “streams of consciousness” is what I wanted, that is what I would have asked for.

So, again, the information can be misleading unless you try to dig deeper, or, again, go read and research for yourself.

Keep feeding it information to get what you’re looking for.

I attempted a little technical writing using ChatGPT, since that is one of the most common suggested applications for it. That generalized outline I talked about earlier? It was very generalized, and I still had to customize it to the company and fill in the content.

But, getting the gist of the various OSHA standards was easy. For once. I just had to copy and past the specific OSHA reference from one of the earlier answers and ask for an explanation. It provided a lot of good content to use, but I still had to sift through, summarize, and format it.

So, you can ask for complete pieces, but depending on your skill and what’s needed, you will still have to customize, format, edit, and all the other busywork of writing. That being said, the program gave me a lot more to work with than endless hours of research and, of course, that pesky rough draft.

But, it’s not a one-shot and done. Which means you still need a human to operate it, and you still need the knowledge to know what to ask it for.

Ironically, the platform makes you confirm you are human before you use it, too.

Complicated historical events can be more easily broken down so you can narrow your search field.

Setting aside the lack of ability to cite specific sources unless those sources are first defined, ChatGPT is good at summarizing complicated information, especially if there are several factors involved. Also, if you need clarification, the AI will consider your previous questions (or comments) and tailors its answers to you.

The novel I’m writing involves the Yugoslav Wars as part of its backstory. This is an area of history about which I know the basics, but not a lot of details. I was a child as the now former Yugoslavia dissolved into a decade-long civil war and a teenager by the time they ended.

My historical interests as a teenager centered on Western Europe, overlapping with whatever I was reading. Since my reading interests leant towards Beowulf, Jane Austen and Shakespeare, you can guess the history I read. Not to mention that Peter Jackson’s Lord of the Rings movies started making headlines about the time I started high school.

Those interests continued for a good chunk of my adult life until I needed a good backstory for my main character and I couldn’t use any of the history I knew about. My main character lives in the modern world–not in a medieval or Renaissance one. But, I wanted to stay in the European arena.

So, the Balkans were my logical choice. But that meant a LOT of research and because the Balkans are not straightforward, either geographically or historically, that meant I had mounds of research to do and very little time in which to do it.

Enter ChatGPT. I’ll still have to consult my source material, but I have a much better idea of what I’m looking at. Which means I’m not having to read several very thick books of history, just to find the few kernels wherever I find them.

Does this mean you don’t still have to learn and read on your own?

In a word, no. ChatGPT makes learning a little easier because it can give you basic information. But it’s no substitute for learning how to evaluate information for yourself, or for how to think, read, write, or anything else. It’s still a computer program. A sophisticated one, but it still requires a human to operate and a human to give it input.

Which means you still have to have knowledge to input into it. And the only way you get that knowledge is to learn on your own. If you have questions, you must have knowledge to know what questions to ask.

Which means you have to make the personal effort and commitment to learning and reading. You can’t rely on a computer program to tell you everything. Sometimes, you really do have to go read a book. Or several.

That being said, ChatGPT is really good for generating ideas, and for getting enough ideas to really start writing and I will be keeping it as part of my arsenal.

Your email isn’t shared with anyone and you can subscribe at any time by clicking the link in one of the email updates. This is for the blog only. You will not be added to the newsletter distribution list here.