Or – How many Vogons does it take to change a lightbulb?
Nearly seven months ago, tech firm OpenAI launched ChatGPT, its artificial intelligence (AI)-powered natural language processing (NLP) software. Since then, it’s been impossible to ignore. Its advance is unavoidable. And ChatGPT isn’t the only one — there are many other NLP projects out there, and the race is on for whoever can create the best NLP software.
It’s simultaneously fascinating and terrifying to watch for anyone whose livelihood depends on words, i.e., writers, journalists, copywriters, copyeditors, proofreaders, translators — a proverbial Frankensteinian tech creation, out there, on the loose.
Who knows where it may lurk …
Actually, it might be more accurate to think of the tech as being more like the creator himself, Victor, feverishly cobbling together answers to life’s questions from whatever corners of the internet it can, without discernment of source or ability to credit one (please see footnote).* There have already been dire real-world consequences documented due to its misuse in the legal profession, as this article by Forbes illustrates. This year’s Hollywood writers’ strikes are based in concerns over AI (see this article by the Independent). And then there’s this. The concern is understandable.
From my own experiments with ChatGPT as a research tool, it’s been factually incorrect, emotionally flat, and I’ve had to be prepared to argue with it at length until it backs down and admits it doesn’t know an answer (an exciting way to spend an evening, eh?) to questions such as ‘Is Highgate Cemetery used as a location in XXXXX book?’ (‘XXXXX’ was not the literal info I fed it, but you get the picture).
But… I can see its merits for brainstorming ideas. Even if what it suggests is ludicrous, it might provide the prompt needed to reignite one’s creative writing flow.
Apparently it can also compile lists in an orderly fashion. Now, I will need to add a ‘Portfolio’ page to my website very soon but the sheer number of books I’ve worked on since 2014 on numerous subjects makes collating all of that – publicly available – information feel like an overwhelming pushing-a-boulder-up-a-hill task. However, I will have courage and shall report back on how that bold step into automation goes…
While ChatGPT can give you an answer to any of life’s questions, it still needs the Ultimate Question to offer a half-decent answer.
In Douglas Adam’s novel, Hitchhiker’s Guide to the Galaxy, the Ultimate Answer to the question of life, the universe, and everything is 42. Of course. But to get behind that, Deep Thought needs the Ultimate Question. And, in the book, that question is: ‘What do you get if you multiply six by nine?’
The book’s main character (Arthur Dent) declares that when he heard this, it confirmed his fear that there was some deep fundamental flaw in the universe. (Or perhaps that should be the Metaverse? A virtual Dismaland, which died this year according to this article by Insider.)
So, imagining ChatGPT was Deep Thought, I decided to ask it the following question:
‘How many Vogons does it take to change a lightbulb?’
(Apparently, this question only appears in the 2005 film of the book, which I confess I’ve not seen, so thus cannot verify this to be true or factually correct.)
This is what it said:
‘As a fictional alien species created by Douglas Adams in his book The Hitchhiker’s Guide to the Galaxy, Vogons are notorious for their bureaucratic and unpleasant nature rather than their technical skills. Therefore, the answer to How many Vogons it takes to change a lightbulb? is likely to be excessively inefficient and frustrating.
‘In the humorous spirit of Douglas Adams, a typical answer might be as follows:
It takes three Vogons to change a lightbulb: one to fill out the necessary paperwork in triplicate, one to yell at the lightbulb for being inefficient, and one to eventually screw it in the wrong way and declare the task completed to the dissatisfaction of everyone involved – and anyone who heard about it later.’
Ah…
Natural language processing (NLP) tech as we know it today, i.e., publicly available to anyone with an internet connection and not just tech engineers and programmers, is still in its early stages despite being around since World War II. If it were a human, I’d hazard a guess it’d still be gurgling a bit, able to stand up and hold its own at times but then falling over, and grabbing things it really shouldn’t occasionally. Still learning and evolving, as we all do. I personally wouldn’t trust it in a professional copyediting context — not yet, at least — nor for copywriting for the reasons given in the third paragraph above, as well as the obvious copyright concerns. It would be like letting a toddler drive a car.
Decent, forward-thinking tech companies that create NLP products need the services of expert language specialists like translators, copyeditors and copywriters in order to do better, as the world wakes up to all the possibilities these new advances may present.
The proverbial horse has already bolted. The fourth industrial revolution, that of the internet, started a long time ago. The internet began entering ordinary people’s homes, i.e. those of people who aren’t computer scientists, way back in the 1990s. A lifeline for some; dark magic to others. But without it, I wouldn’t be able to do the work that I do in the way that works best for me. Many others I know will feel the same.
Without Alan Turing’s machine during World War II (1939–1945), encrypted Nazi codes wouldn’t have been cracked so quickly. But that machine was programmed with a specific purpose, by a human being with the integrity to know what was at stake. The Turing machine wasn’t created to streamline work processes so that everyone could go home for tea early and have more time to play golf. It was a matter of life and death.
Speed can sometimes be of the essence and such tech can be a force for good. Who’s to say that AI won’t be our best friend in other battles we face today, our contemporary Entscheidungsproblem – those of false information, climate crises, diseases, and pandemics. As always, it’s a tool, like the internet, or a social media account.
How good or useful it ends up being in the long run for humanity depends on what it learns from the particular human beings teaching, feeding — and using – it.
* Please, please, PLEASE – ChatGPT – cite your (URL) sources. All the Who, What, When, How, and Whys of it all.
Got any questions or comments? Please get in touch.
As with the rest of this website — all words are written by me, Anya Hastwell;
images are created by myself using elements from the Canva design tool, unless otherwise stated.