ChatGPT: a quick Q&A to get you up to speed

When ChatGPT dropped at the end of November 2022, the internet lost its mind. Google is dead. Goodbye writers. Students rejoice. In short, there was a lot of unrestrained excitement and a lot of media coverage, with the platform being overwhelmed as people the world over signed up to experience it firsthand. To make sense of it all, here’s a quick Q&A to help get you up to speed.

What is ChatGPT?

ChatGPT is a new large language model from OpenAI that is kind of like a sophisticated, knowledgeable and very human-sounding chatbot.

You can more or less ask it any question you like and give it a specific set of instructions and it’ll generate seemingly authoritative and natural-sounding responses. You can even question its output and get it to look at things in a slightly different way – and it will respond accordingly.

As its name implies, it will do this in a conversational manner. When you use ChatGPT, it does indeed feel like you’re talking to someone and not some thing. It’s that distinction that sets it apart from, for example, its predecessor, GPT-3. Although the following is a tad overplayed, there is some truth in the idea that it “proves the gap between computers and humans is rapidly narrowing”, as a couple of hacks at the Telegraph recently put it.

Why the buzz?

Because it’s really good. Because it’s unlike anything we’ve ever seen before. And because it can do pretty much anything within reason and the framework in which this sort of technology exists (it can’t for example, build you a castle or whip together an exquisite meal).

Need it to write code? Sure, no problem. Want a long-form article on a niche subject? Done. Looking for an answer to a complex equation? Here you go (with an explainer to boot).

And it can do all of this at speed. ChatGPT ridiculously fast at generating responses that are, relatively speaking, close to the mark.

Is it a watershed moment?

It certainly feels like it. OpenAI is definitely on its way to becoming a household name and the widespread adoption of large language models seems to be more than inevitable, either directly or directly. Every industry will be affected.

That said, even though the technology has been around for a few years now, it’s not without its faults. Because of the way algorithms like this work – in very simplified terms it follows specific instructions and generates results based on precedent without the nuances of human comprehension – a lot of what it turns around can either be complete mumbo jumbo or full of errors.

OpenAI acknowledges this, which is why it has the following mini disclaimer pop-up prior to your entry into the platform: “The system may occasionally generate incorrect or misleading information and produce offensive or biased content.”

And that’s why human oversight over AI output is always going to be essential. Without it, it’s a recipe for disaster. You don’t, for instance, just leave the machines to it when you manufacture cars – humans are involved throughout because the risks are far too great.

Should Google be scared?

Not really, as it’s not a direct comparison. Google search is about making information online more discoverable – i.e. directing you to quality sources it believes best answer your query. It doesn’t answer the question directly, because that’s not what it’s designed to do.

ChatGPT on the other hand is a chatbot that generates, to all intents and purposes, original responses to queries based on a huge sample of data that it has gobbled up like an insatiable competitive eater. The answers it provides are more personal in tone and more direct in that it isn’t providing you with a range of options to explore, but telling you as it is. Well, as it thinks it is.

What about writers? Is it game over for them?

ChatGPT has editorial chops, I’m not going to lie. It can write “well” enough and adapt its style if instructed to do so. It can also organise content in a structured way and produce copy at a speed that even the most brutal content farms can’t match. All of this suggests that the writing is very much on the wall for writers.

But it’s not. Far from it. We’re always going to need writers to produce original, thoughtful and nuanced content from a human perspective, whether it’s a thought leadership piece, a report or a highly optimised SEO blog. The experience, knowledge, insight and craftsmanship belonging to a seasoned writer and thinker will always deliver the kind of unique value that no machine can match.

Yet, of course, there isn’t always going to be a demand for that. In some instances, the kind of content that is required by businesses, for example, to compete online – such as a large batch of landing pages that require a quick turnaround – will be better suited to alternative solutions.

This includes copy written in collaboration with AI support – which is pretty much already the norm in digital marketing – and content generated by AI, with writers more involved in rewrites, edits, fact checking and quality control.

So, the best writers can rest easy. Their jobs are safe. Mediocre writers, however, well, that’s possibly another story.

What’s next?

A full and more comprehensive version of ChatGPT is likely, one that significantly improves on some of the flaws that OpenAI has acknowledged, such as its tendency to be wordy – it clearly needs to read this gem on plain English – its propensity to deliver “plausible-sounding but incorrect or nonsensical answers” and its inability to interrogate queries that are too vague to warrant a single response (which it will go ahead and do).

Beyond that, if rumours are true, GPT-4 is also on its way. If you thought ChatGPT feels like something of a game-changer, the fanatics will have you believe that your mind will be blown by the power of the latest iteration in text generative artificial intelligence.

GPT-4 is likely to be a major paradigm shift. OpenAI is currently keeping schtum about it all, so other than speculation – which you can strip back to stupendously better than GPT-3 but not that much bigger – there’s very little to go on other than hype. And that’s fine. Hype is good for business.

If you want to know how AI-generated content can help your business, learn about our generative content service or get in touch for a demo.

-

This article was originally published in January 2023

Other Posts

All Posts