Will ChatGPT Soon Make Us Obsolete?

If Hollywood is anything to go by, we should be deeply suspicious of artificial intelligence (AI). Films such as Terminator, 2001: A Space Odyssey and I, Robot paint bleak visions of the future. Since last November, ChatGPT has made a highly advanced text-based AI freely available for the first time. It is similarly powerful to the likes of Terminator and Co., but instead of waging wars, it produces something far more ordinary: text.

As communications experts, this is naturally a development that has captured our attention. That is why we are taking a look at this innovative tool from a PR perspective. To what extent will it change our day-to-day work, and will we soon have to start worrying about our jobs?

What is ChatGPT, and why is everyone talking about it right now?

Artificial intelligence has long since become part of our everyday lives. We use it when Google Maps calculates the fastest routes for us or when Spotify creates playlists designed to match our musical tastes. The principle behind ChatGPT is nothing new either. On a minimalist-looking website, users can ask the chatbot questions via a text box. The answers ChatGPT provides have sparked enormous enthusiasm online and generated extensive media coverage. This is because the quality of its responses is surprisingly high. The AI is capable of writing entire essays on complex scientific subjects. On request, it can generate poems and song lyrics on a particular topic within seconds. It can write apps and websites in programming languages.

Here is an example: Did your new favourite TV series end far too soon? Simply ask ChatGPT to write a new episode. What is impressive is that there is no need to explain the story so far to the text-based AI. It also knows all the characters and their relationships with one another and generates new content on this basis. This is made possible by the enormous volume of text that developer OpenAI uses to “feed” its latest creation. These texts were originally written by humans and may come from books, newspaper articles, social media or online forums. Using an artificial neural network that imitates the way our brains work, ChatGPT connects information and combines it to create new texts.

The problem with the data

Artificial intelligence essentially creates a “remix” based on the sources it knows. It cannot assess whether the texts it has been fed are false or even illegal. As Der Spiegel reported at the end of January 2023, the system requires human assistance to deal with this. Thousands of contract workers from Africa and Latin America have to pre-screen texts for the AI for an hourly wage of around two dollars, preventing descriptions of child abuse, rape and torture from entering the system. It is also unclear how, or indeed whether, OpenAI prevents fake news from finding its way into the database.

The AI itself has no access to the internet. It can neither verify facts nor refer to current events. There is also always a considerable risk of copyright infringement. What happens if ChatGPT reproduces entire passages from copyrighted texts that users then reuse, potentially exposing themselves to legal consequences? Without sources being provided, there is no way to trace this. A great many legal questions therefore remain unanswered.

The standardised text

But let us return to the key question: How good are the texts produced by this machine “superbrain” really? It is worth taking a closer look at different types of text. ChatGPT’s strengths become apparent in short news reports, for example. In general, the system is good at putting together texts that follow a certain structure. Although the text bot is also willing to try its hand at creative work, this is where its weaknesses become apparent. If it writes a love poem, for example, the result is essentially a cross-section of all the love poems the system knows. The result may be acceptable because it meets our expectations of the genre, but it is equally dull and formulaic. Truly unique content is never produced.

Creativity and abstraction are particularly important on social media. Innovation is essential in the world of communications if content is to stand out from the mass of similar material. Every day, smartphones and the internet expose us to an overwhelming amount of text. In the battle for attention, content has to surprise and move its readers. This requires the ability to think abstractly and develop creative new ideas. Artificial intelligence is not capable of doing this.

Empathy as the key to successful communication

Another human ability that no AI possesses is empathy. It is empathy that enables us, as writers, to understand the needs of our readers. It is particularly important in crisis communications. Insolvencies and restructuring processes, for example, can threaten people’s livelihoods and cause concern among various stakeholders. Especially in these critical situations, PR consultants owe everyone involved clear and honest communication that takes their concerns seriously.

The art of communication lies, not least, in striking the right tone. This is impossible without a feel for the context. Only an appropriate choice of words, tailored to the situation, creates credibility and strengthens stakeholders’ identification with the company. To achieve this, you have to be able to put yourself in someone else’s shoes. A machine cannot do that.

The revolution fails to materialise

It is impressive how rapidly artificial intelligence has evolved. But ChatGPT does not represent a revolution in text creation. It produces average texts that lack abstraction, creativity and emotion. It can be helpful with certain routine tasks that arise on a daily basis. But even here, careful review and expertise are required. It can be useful for research tasks or as a source of inspiration. As long as it remains a supporting tool, it can speed up workflows and create more room for creative processes. To write truly unique texts, however, humans are still needed — at least for the time being.