Beyond The Chat – One Year On, Is ChatGPT Really ‘All That’?
Almost one year on since ChatGPT took the world by storm, Nick Floyd, Head of Content at Catalyst digital marketing agency explores whether it has lived up to the hype.
It’s hard to believe that it was just only 12 months ago that ChatGPT changed the AI game with its remarkable (though not entirely faultless) ability to synthesise information into human-like content. From writing essays and papers to creating poetry and even passing exams, at one point it seemed there wasn’t much it couldn’t do.
Cue endless debate as to what it means for the future of content creation, marketing and customer services, and whether we are all in danger of our robotic overlords taking over.
In reality, though, as the tool has become less of a novelty and more a part of daily life, users have been quick to realise that for all its clever uses, it does have limitations, too. Now, as it approaches its first birthday, the big question is – does it really live up to the hype?
The answer, from my perspective at least, isn’t a straightforward one. Having used ChatGPT a fair bit since its launch, my experience is that while there are many ways in which it can provide real help in the day-to-day, it can also, if not properly managed, come as a hindrance.
Take, for instance, its ability to retrieve information. ChatGPT is able to extract valuable information from extensive datasets or large articles quickly and with an ease that is unmatched.
If you enter, for example, “What is net zero?” into a traditional search engine, you’ll get a slew of results to sort through. In contrast, ChatGPT is able to provide instant, user-friendly responses, usually in easy bullet points or step-by-step guides. This can be a big help, especially when researching new clients or industries. Still, it’s not Google.
Inherently, all generative AI applications are only as accurate as the information which has been fed into them. But the world and information keep changing. This means that there can be huge scope for inaccuracies and misinformation. The danger is that if people start using ChatGPT as a search engine without looking at a number of sources and views and doing their own supplementary research, it could amplify particular viewpoints. By comparison, Google results are in real-time and include full sources (most of the time).
ChatGPT’s aptitude in the content and communications sphere lends itself to an equally interesting debate. Indeed, many of us who have tried “write an article about (topic)” as a prompt to ChatGPT might have initially been impressed by its ability to respond in a human-like manner. Look a little closer, though, and it’s likely that it will be far from a finished article, and require further prompting and editing before it can be used.
In this way, while it can certainly provide a useful tool in helping to create a basic copy outline, human input, and generally lots of it, is still essential. As a general rule, AI is much better at helping you research a topic than writing the whole thing for you.
This takes me onto another, less documented issue. Inherently, SEO is the holy grail of the digital marketing world, but getting it right requires a carefully articulated programme of original content enriched with relevant keywords.
Imagine a world where all businesses in an industry, say manufacturers, use ChatGPT to draft similar blogs on the ‘latest sustainable manufacturing technique’.
If those operators don’t take the time to customise the blog post, they run the risk of producing near-enough the same blog as everyone else; we’re looking at a world where every business is saying the same things in the same way – a world where content is devoid of emotion and the human touch.
Of course, this risk is relatively minimal at the moment. Most ChatGPT outputs aren’t good enough to be used without editing from a professional. However, it’s worth keeping in mind that AI is in its infancy – this is the worst version of AI we’ll ever have. It remains to be seen how AI will navigate SEO in the future as tech develops.
As a last point, it’s important to consider customer interactions. This time last year, the consensus was that ChatGPT would transform the customer services space with its ability to not just generate responses to customer inquiries but do so in a much more engaging and memorable way. In a word, make existing chatbots more human-like.
Has this happened? No, not really.
Yes, it can be beneficial in its ability to filter enquiries until a human operator is available and provide an automated point of contact 24/7. Still though, if you want to give your customers accurate, consistently good service experiences, then ChatGPT – with its lack of empathy and aptitude for misinformation – probably still isn’t the best route.
ChatGPT is only going to improve – it’s important to continue engaging in generative AI conversations. After all, how many major tech breakthroughs have reached their peak on their first birthday? Even most ‘Apple moments’ were the products of constant reiteration and development. Therefore, if you really want to know if ChatGPT is worth the hype, it might be better to ask me again in five, even ten years from now. But for now, the recommendation is to balance its vast uses with a substantial amount of human oversight.