Let't See Whatcha Got, Mr Chatbot

This is the second column in our recently reformed collaboration.

PHIL:  Okay, Jim, so the other day I was ruminating over where all this hubbub over Chat Bots, LLMs, and purported artificial intelligence is really headed. And I realized that I’ve been remiss in not looking more closely at what Open AI might really be delivering with its ChatGPT. Consequently, I thought why not perform an experiment in writing performance — a subject dear to our hearts. Here’s what I did.

At one time in my youth, I club raced an Austin Healey Bugeye Sprite sportscar. I’ve also written about that fairly recently. So I asked ChatGPT4 to write a paragraph about Bugeye Sprites. This is what the ChatBOT produced:

“Bugeye Sprites, also known as "Frogeye" or "Frog-eye" Sprites, are a classic and iconic sports car produced by the British car manufacturer Austin-Healey from 1958 to 1961. These charming little roadsters gained popularity for their distinctive and unique design featuring large, round headlights positioned prominently on top of the bonnet, giving them their endearing "bug-eyed" appearance. The Bugeye Sprite was a lightweight and nimble two-seater, powered by a modest but spirited engine, making it a joy to drive on winding roads. Despite its short production run, the Bugeye Sprite has become a beloved classic among car enthusiasts, embodying the essence of carefree motoring from a bygone era.

Reluctantly, I have to admit this is not too shabby, certainly less shabby than I anticipated. Indeed, it sounds pretty much like a Wikipaedia entry. But that is, I suppose to be expected, since I’m sure Wikipaedia is one of the major sources of “food” for the LLM which is at the heart of ChatGPT. Compare now, if you will, what I wrote abut six years ago about my experience with the Bugeye Sprite:

“I raced my Bugeye in G-production class, which allowed me to make certain modifications to the engine, drivetrain, and chassis.  I had the cylinders of my Sprite's engine bored out to their maximum diameter and fitted with oversize pistons and rings. The crankshaft was hardened by shot peening, lightened and balanced, as were the connecting rods. Intake and exhaust manifolds were polished and ported (passages enlarged and polished to reduce turbulence in the flow). Heavy duty hardened valves, pushrods, and extra heavy-duty double valve springs were fitted (to raise the limiting point of "valve float" from the stock 5000 rpm redline to better than 7000 rpm, with an attendant increase in horsepower production).

“Weber carbs were fitted because they had multiple "barrels" for enhanced fuel/air delivery, and "accelerator" pumps that shot a jolt of raw fuel into the carb barrels when you stepped hard on the accelerator. Both air-intake and exhaust systems were "tuned" for maximum efficiency by taking advantage of the back and forth pulsing of gasses intake mixture coming into the engine and exhaust gasses leaving it

“I also fitted the live solid rear axle with a locking, limited-slip differential that overcame any loss of propulsive power due to an inside rear wheel lifting or losing traction in the tight turns. And for short courses, switched to a differential ring gear that yielded a final gear ration of 5.37:1  (as opposed to the standard 4.22:1), whilst for the longer courses, switching to a ring gear that yielded, if memory serves, a 3.85:1 final drive ratio. Higher ratio number for better acceleration on the short courses, lower ratio number for better top speed on the longer courses.

“The result was that, at the bottom of "The Chute" at Meadowdale, a long, steep downhill straightaway, my diminutive slightly "improved" G-production Bugeye could, in the right circumstances, clock 130 mph.

The Bugeye was truly a roller-skate of a car, with tremendous agility and cornering ability due to an exceptionally low center of gravity and a relatively wide track. And although the Bugeye was not stunning when it came to raw speed, it performed much better than you might expect on a road course. Primarily because you did not have to slow much at all in the turns. And it was not unusual for one to slip past a faster, more powerful car in a tight 90-degree turn or a 180-degree switchback.

The maneuver in a Bugeye went like this. You hung about six inches back from the bigger car's rear inside fender coming up to a tight 90-degree or better turn and waited for the driver of the bigger car to pick his line into the turn. You downshifted, followed him on the same line into the turn and waited for the heavier car to slide somewhat toward the outside, pushed by greater mass and centrifugal force.

“As soon as you saw the slide start, you broke off the bigger car's line and aimed the Bugeye. pedal-to-the-metal, straight toward the apex of the turn. If you did it just right — and if you had good tires and your engine was well-tuned and you picked a gear to give you redlined rpm — you'd slide by the bigger, "faster" car with the Bugeye feeling as though it were on rails. Man, there was no better feeling in the world !!”

(from Life Is Like a Monza Wall, https://portroyalgroup.substack.com/p/life-is-like-a-monza-wall">https://portroyalgroup.substack.com/p/life-is-like-a-monza-wall)

Now, I guess I could have coached ChatGPT4 into revising what it had produced into something a bit less lifeless — but why, in the world, would I do that? The minions at Open AI will soon enough run across my piece on the Internet and feed it into the LLM so that, the next time, ChatGPT can, perhaps, better plagiarize my work. However, I do not believe that the “averaging” performed by ChatGPT will ever enable it to approach good writing. Instead, what we are facing, IMO, is a loss of discerning readership, a general numbing of the mind, and a dulling of the pallet for worthwhile, creative prose.

JIM: I can’t say that this was any sort of an unexpected revelation. What’s wonderful about it is that it makes the point in a crystal clear way that Ai writing programs are not, and probably never really will be a substitute for the ‘human’ experience.

This might explain why, over the past several days, I have been noticing a slight shift in tone in the ridiculous amount of Chatbot advertising. They now seem to be trying to put it in its place as a helper, not a substitute. This is interesting because it is tantamount to an admission of the inadequacies of the various programs when compared to actual human experience

By my own admission, I have never interacted with a Chatbot of any kind. But you don’t have to have personal experience to understand what makes these things work, and you only have to have a rather small bit of intellect to understand their shortcomings vis a vis all the original hype that the companies behind these programs have generated.

Frankly, it seems to me that this whole trend is one colossal jerkoff. So you invest in one of these programs, then you have to train people on how to use them, and then you need to have people around with the talent and skill to be able to ‘fatten up’ the paltry, in many cases derivative, sometimes copyright infringing, flat-as-a-pancake narratives it spits out. A smart businessperson, meaning one who has not been hoodwinked by all the initial hype, would likely conclude that this is tantamount to putting lipstick on a pig.

I have seen a lot of interesting examples of how Ai is revolutionizing certain areas of business, art, science, and medicine. A friend of mine named David Loblaw, who is a gifted visual artist is using it to create amazing images.

In our ballpark, which is writing, I am seriously unimpressed with any of the examples I have seen. And the reason for this comes down to one basic thing, which is nuance, which is a key factor in attracting the readers, drawing them into the experience that is being written about. Your example made that abundantly clear.

A Chatbot is merely a shadow of a human writing. I believe, as do many people smarter than me, that it will never have the ability to create writing that has genuine passion, simply because there is no innate human experience to draw on. The best it will do is oink and try to look sexy.

PHIL: Well, Jimbo, that is precisely the crux of the matter — the imposition and acceptance of so much blandness in the written content produced by Ai (lower case ‘i’ used intentionally).

If a visual artist uses an Ai tool to manipulate an image, the result retains its originality and uniqueness because the Ai manipulation is functioning as a tool in the “hand” of the artist, just like a brush does. However, in the case of ChatBots, the creative direction (action?) is being turned over to the Ai tool, as in, “write me story about the Ocean Gate Titan disaster…”

If the nominal author were using the Ai tool in a manner similar to the visual artist, he or she would be dictating the text that the BOT is assembling. Which, BTW, is what we’ve had for decades now and which goes by the name of computerized voice recognition, dictation, and transcription.

Let me leave you with one last comparison. I asked ChatGPT to write me a Haiku about writing Haikus, and this is what I got back:

“In code's embrace, I

ChatBox weaves poetic charm,

Haikus dance in bytes.”

This is not only bland, it contains a significant grammatical error (the subject is singular but the verb is plural). OMG, if ChatGPT can’t even manage correct grammatical construction, why in anyone’s imagination would we want to trust it to give us correct information or data?

BTW, this particular grammatical error appears endemic to all automated writing programs such as Grammarly, the Word editor, and every one I’ve used on social media publishing platforms. If seems the Great Minds of Ai and LLMs can’t are stumped by English grammar.

For the record, several weeks before I tried this ChatGPT experiment, I wrote a haiku meme on Facebook in response to a post about an Ai BOT for writing haikus. Mine went like this:

“It seems to me a
HaikuBOT would generate

More crap than not.”
I rest my case. Cheers!

JIM: As is generally the case with anything new, or should I say purporting to be new, because these days that’s the best you can do, there are always big lies, exaggerations and highly spun hyperbolic rainbows that proclaim wonderfulness but really only sell a notch or two above mediocrity.

Over the past couple of years I have wound down my advertising business and started writing stories. I had a couple of reasons for doing this. One was to see if I really could string enough words together coherently to tell a whole tale, which meant my brain had not suffered due to my illness.

Although I knew about ChatBots and the like, it never even occurred to me that they would be of any real utility, because the majority of these stories were written more or less in a stream of consciousness. So what could I tell a Chatbot when I didn’t even know myself from one page to the next?

To me, regardless of the literary merit of these stories, the key was the pure adventure of letting the story unfold in a very immediate sense. The idea of of using a source outside my own brain, except for research on location details, felt to me like it was taking the pure joy out of the work and transforming it from the joyful and challenging process of creation to the drudgery of editing.

I can see how these tools would be useful to writers who may be just starting out or people who simply have trouble stringing words together. But fro those of us who have been creating original thoughts and transposing them into haikus, poems, lyrics, stories, novellas and larger works of fiction, the whole idea of Chatbots that can ‘help’ you is somehow not just unnecessary, but unwanted. My feeling is that if you need a crutch like a chatbot to improve your writing, maybe you’re in the wrong game.

So I will leave you with a haiku of my own.

I chat with the Chatbot

And the bot, alas,

Says nothing new to me

PHIL: Well, Jim, I have to say in closing that I’m really enjoying this redux of He Said He Said. And I believe you are as well. Which is not surprising because each of us in his own way styles himself a “writer” — namely as someone who enjoys wordsmithing and sees it as an art form, notwithstanding that a lot of prose is produced about non-arty items and issues.
I suspect that this is, as well, why people like you and I recoil at the idea of asking ChatGPT to write content for our blogs and other activities. The ChatBOT’s chat will always pale next to good writing because it is created by seeking the lowest common denominator of expression. Of course, the content factories will love it, since their universal directive is to churn out as much generic pablum as possible in as little time as possible and at the lowest cost per world possible.

As an aside, not everyone is equanimous about this trend. Indeed, I know of a new organization that’s recently formed, the Association of NotBOTs, with groups on both LinkedIn and Facebook. They self-describe as a gathering of “Creatives and humanists opposed to the indiscriminate adoption of, and reliance upon Ai and robotics….” More power to them. Cheers!

NOTE: This is the second in our 2023 series, you can find the first post at

https://www.linkedin.com/pulse/back-our-blazing-saddles-again-jim-murray/