L'article
One swallow doesn’t make a spring… but it announces it! ChatGPT is to content automation what the swallow is to spring: the herald of a revolution that some already feel in their daily lives.
While the feat of OpenAI, the instigator of ChatGPT, is far from flawless, it does promise a future in which the bot will perhaps write like a human. Perhaps with the same flair, the same sense of formula, the same sensitivity, the same humour. That’s a lot of “maybes”…
In the meantime, ChatGPT manages to do this from time to time, especially in English. While it shines with creativity and an impressive ability to converse and understand instructions, it doesn’t always manage to produce ‘consumable’ text in its current state… except perhaps for SEO purposes. What is ChatGPT worth to content marketing specialists? Here are some answers!
How does ChatGPT work?
Developed by OpenAI, ChatGPT is a language model based on deep learning that analyses millions of texts to learn how to generate content autonomously. Launched on 30 November 2022, this bot like no other reached its first million users in just 5 days! To give you an idea, it took Netflix three years to reach this milestone.
Technically, ChatGPT is trained on vast data sets to predict the ‘best’ word to append to the previous term and create text conversations and content that mimic those of humans (essays, ebooks, articles, computer code, etc.). And because it’s based on deep learning, ChatGPT is capable of continuous improvement: as it ‘consumes’ data, it becomes more efficient in its predictions, producing better quality content and more relevant answers to users’ questions.
Even more ‘frightening’, ChatGPT is able to use the insights it gleans from its existing datasets to perform tasks for which it has not been specifically trained! It can synthesise a text into a comparative table, summarise long texts down to the essentials, correct computer code, create posts for social networks with emojis and hashtags, and so on.
ChatGPT: what’s the impact on content creation?
To say that ChatGPT has already redefined the way we think about content creation is an understatement.
The tool has firmly established itself as a premium assistant for web copywriters, content marketing specialists, community managers, editorialists and advertising professionals. It has even been added to the technological arsenal of marketers and salespeople to :
- Accelerate the creation of very high value-added marketing content, particularly in Account-Based Marketing (ABM) campaigns, including contextualisation according to the target audience’s sector of activity;
- Contribute to the drafting of personalised, hard-hitting and structured prospecting emails;
- Suggesting punchlines for inclusion in sales pitches.
Even better: to everyone’s surprise, ChatGPT has made its official debut in the… scientific literature! At the time of writing, the chatbot has already been cited as a co-author in four scientific articles.
Taken by surprise, Google panicked and hurriedly launched Bard AI, a (belated?) attempt to counter the ChatGPT wave, with mixed success. The Mountain View giant even urgently recalled its two founders to draw up a “code red” to preserve its hegemony. ChatGPT answers our queries in a more user-friendly and sometimes more relevant way than Google’s search engine, without having to discriminate between websites on the results page, without having to click and without having to make the effort of looking for information on the page.
And there’s no mistaking the fact that ChatGPT is often down, precisely because of its success. Google, on the other hand, is accessible 24/7, without interruption. And yet Internet users, both private individuals and professionals, do not hesitate to refresh the ChatGPT page dozens of times before accessing the chatbot.
ChatGPT: The First Limits of an “Ongoing” Revolution
Since we’re dealing with a language model based on deep learning, we thought it would be a good idea to ask him. What are your limits when it comes to content creation, ChatGPT? Here’s his answer:
“As an AI assistant, I am able to produce texts on a wide variety of topics using pre-trained language models. However, it’s important to note that I don’t have a complete and contextual understanding of every topic I cover. As a result, my texts can sometimes lack the nuance, creativity and human perspective that experienced content creators can bring. What’s more, like any language model, I sometimes produce errors or inconsistencies in the texts I generate, even though I always try to maintain the highest possible quality. So it’s important that my texts are always checked and edited by qualified humans before being published or shared.“
That says it all. For all its prowess in content creation, particularly in the field of marketing, ChatGPT is still a little too… robotic, cold, and academic. It is criticised above all for its lack of depth, insight, or as it puts it: a lack of nuance and human perspective. Had he been human, ChatGPT would have drawn on his personal experiences and subjective perceptions to create. This is clearly not the case. At least not yet!
In French, AI still has difficulty understanding idiomatic expressions (phrases with a figurative meaning) such as “every cloud has a silver lining“. The absence of such expressions in a text can also indicate that the content has been generated by a machine.
Last but not least, ChatGPT invents studies, surveys and polls, while presenting them to you without the slightest embarrassment. For example: “Companies that use AI to create content generate up to 50% more leads than others”. This statistic, provided without batting an eyelid by the bot, is a figment of its imagination.