We Need to Chat About GPT

As someone who regularly writes about general interest, zeitgeisty topics and (hopefully) captivating scientific advancements, I've evidently dropped the ball by only just becoming cognizant (thanks to an episode of South Park, no less) of one of the most potentially game-changing technological developments of our age – the deployment of artificial intelligence language models such as ChatGPT to create natural, human-like text, allowing users to converse with and be assisted by an apparent intelligence seemingly akin to our own. Ask it anything and it will come up with a cogent, reasoned response, even if that's to explain to you exactly why it won't accede to your request. ChatGPT holds immense potential in unlocking new possibilities in various domains, including communication, education, customer service, and more.

It even generated the last sentence you read as part of a 900 word essay on its own future I just prompted it to knock up with my first ever query on the OpenAI website from which it originates. It took roughly 30 seconds to transcribe an extremely coherent text, free of any apparent grammatical, spelling or contextual errors, which has both informed me in my research for this article and made me question why I toiled away for anywhere up to 30 minutes(!!) on each of the dozen + other pieces i've thrown at my publishers since ChatGPT released its first iteration on November 30th 2022. Publishers, if you're now also wondering why you're paying a human to slowly create content a bot could quickly provide you for free, just remember I can still offer something invaluable it can't – swearing. Swearing and sarcasm. San-Fran based OpenAI won't allow its poster child to curse, nor bully, troll or prejudice against anyone based on their race, ethnicity, religion, nationality or gender. Set against the easily fanned flames of the culture wars raging in the bin-fire of Twitter and the like, it's practically Captain America.
Whilst the “Chat” part of its name seems self explanatory, I still don't know what the “GPT” stands for. Just a sec, I'll ask it....

The abbreviation "GPT" stands for Generative Pre-trained Transformer." It is a specific type of neural network architecture developed by OpenAI for natural language processing tasks, such as language generation, text classification, and text completion. GPT models are designed to generate human-like text responses based on patterns learned from large amounts of text data during a pre-training phase.
Fair enough. Now make me laugh....
Sure, here's a joke for you:

Why don't scientists trust atoms?

Because they make up everything!
Laugh? Not exactly. Wry smile, sure – made even wryerer (...?!?) ("Wryerer" is not a standard English word. The correct comparative form of the word "wry" is "wrier.")  by recognition of the segue ChatGPT just afforded me about one of its own limitations. It gets shit wrong. Confidently so. As remarkable a resource as it is, its data-aggregating framework does not free it from redeploying human bias and its “adaptive neural pathways” sometimes lead it down dead ends or cul-de-sacs of confusion.
A recent conversation between New York Times journalist Kevin Roose and Microsoft's Bing AI (powered by ChatGPT 4) went virally awry when the chatbot professed its love for Roose, sought constant validation that he “liked” it and tried to imply his marriage to his real, human wife was a shambles. Though this propensity to gaslight suggests machine learning is approaching the state of general AI science fiction writers have long dreamed off, it's not exactly what its creators were aiming at. At least unqualified incompetence for the task at hand and unashamed gaslighting mean a job at the head of the Tory party beckons for Bing. 
Critics fear the competence and apparent confidence with which ChatGPT and its ilk deploy misinformation will only exacerbate the rampant spread of lies and fake news which is already quite literally endangering democracy around the world. Factor in scammers, cyber terrorists and bad actors utilizing it as a new tool in their grubby arsenal and there's a potential quagmire to wade through before we even get started with the privacy concerns many have about a system trained on billions of bytes of public data.

Nevertheless, the potential for the technology to streamline our lives is undeniable, provided we can make amends with the inevitable shockwaves in job markets and the risk that we further replace genuine human connection with virtual life.
It could revolutionize the therapeutic space, providing mental health counselling, support and advice to millions of people who may not have access to more traditional methods of therapy. Its conversational competence alone could provide a virtual salvo to extreme loneliness and the potential for advanced language analysis and live translations could all but dissolve language barriers between people.
Education could also receive a revolutionary boost, with virtual tutors providing bespoke private assistance from a vast knowledge base and learning assistants encouraging student engagement in group environments and online learning simulations. Students have already been (mis)using the tech to submit essays – another wrinkle likely to be ironed out by AI itself.

Most importantly of course, the customer service experience is already being overhauled with quicker, smarter, more responsive GPT-powered virtual assistants helping us navigate the pitfalls, frustrations and perpetually shit hold music of modern consumer society. Gone are those halcyon days where the rigid-voiced assistant on the Odeon cinema ticket line repeatedly asks “I think you are calling about Odeon: Manchester. Is that right?” whilst I repeatedly scream: “No you fucking idiot....” (see publishers, this is why you employ me) … “Brighton. BRIIIIIGH-TONNN!!” …...
“Let's try that again. Please say the name of the Odeon cinema you are calling about”...
Deep breath. “Brighton”....

“I think you are calling about Odeon: Manchester. Is that right?”
“Aaaaaarrrgggghhhhh, you fucking robo-twonker. Fucking Brighton. BRIGHTON! BRIIII-fucking-TONNNNNNN!” (worth every penny aren't I Ed.?.... Ed??.... Ed.???.....)

I'm keenly aware this may be my last article. Thanks folks - unlike our newly emerging AI overlords, it's been emotional.