Robo reality check needed on ChatGPT

You are likely to hear rumours of a new threat to job security. Forget about those dire predictions of a flagging economy, the boss you got on the wrong side of or your failure to keep your skill set current.

The latest perceived threat to jobs is a robot that goes by the name of ChatGPT and is being described by experts as one of the most disruptive technologies ever created.

ChatGPT, or Chat Generative Pre-Trained Transformer in full, was released late last year by artificial intelligence research company OpenAI.

It is a larger-than-life chatbot that can spit out human-like responses on almost any topic in response to word or sentence prompts typed into a text box.

In the workplace, a digital marketer could ask ChatGPT to produce copy for an advertising campaign. A lawyer might instruct it to develop a legal brief, a journalist could prompt it to prepare some background for a story and a small-business owner could ask the chatbot to prepare a summary of key competitors.

ChatGPT makes it possible for users to ask questions or make requests and receive responses that are startlingly clear, seemingly well informed and well-written.

In fact, the chatbot is so well developed that, according to OpenAI, it not only responds to prompts but also answers follow-up questions, admits to mistakes, challenges incorrect premises and rejects inappropriate requests.

ChatGPT constructs sentences, paragraphs, reports, songs, ditties and poems by feasting on billions of words that exist across the internet. The chatbot executes its tasks in an automated manner that seemed impossible just five years ago.

It cleverly strings words together in the best possible way to create polished masterpieces.

But the limitations that come with the current robot revolution also mean ChatGPT is capable of dishing up some first-class trash.

So if talking about whether ChatGPT is going to take your job raises deep concerns, take a Robo-reality check first.

Common imperfections

Generative artificial intelligence (AI) tools like ChatGPT can amaze and astound but are far from perfect and nowhere near sophisticated enough to steal your job – for now.

Even though the chatbot can produce large amounts of digital content fast and cheaply, it is only able to produce material that already exists and has no ability to think ahead, forecast or “read a market”.

When it trawls through masses of existing information to respond to your query, it is not intelligent enough to separate real from fake and it certainly cannot give you a new perspective on a topic or come up with an original idea.

Perhaps the most authoritative source on this AI tool’s potential to steal human jobs comes from ChatGPT itself.

Asked to suggest its limitations, the chatbot responds that it has a “lack of common sense” and though “it has been trained on a very large dataset it does not have an understanding of the world like a human does”.

ChatGPT goes on to say that this means “it may not be able to answer questions or make inferences that require common-sense knowledge”.

If this is not damning enough, when queried further about its shortcomings the chatbot admits that because it operates in a sliding window of text it only has a limited amount of context. This can “lead to confusion or errors if the input text is ambiguous or difficult to understand”.

ChatGPT adds that it is “trained on a dataset that reflects the biases of the people who created it”, which means it can “result in the model producing biased or offensive output if not carefully monitored”.

And as far as creativity is concerned, the AI tool is crystal clear on its deficiency and confirms “it can only generate what it has seen before and cannot come up with creative ideas, original concepts and so on”.

Preparing for the future

At this point, you are probably breathing a huge sigh of relief because of the knowledge that the boss is not about to interview ChatGPT for your job.

After all, ChatGPT is hardly going to be dispatched to investigate allegations of harassment, devise a legal strategy, deal with a challenging customer, produce a report on where the market is headed or conduct an employee performance review.

But it might be used as the first point in a sales process or to draft a dossier or briefing note email or to explain complex workplace topics in simple language.

The reality is that this genie is not going back into the bottle.

Its presence will be felt more and more in the workplace.

There is little doubt chatbots like ChatGPT will become more sophisticated and able to perform tasks that – today – require a high level of human skill and training.

Down the track, this could lead to a displacement of workers in certain industries as organisations look to cut costs by automating processes.

For the time being, ChatGPT is probably best deployed to help workers to be more productive. Consider it a tool to enhance human productivity rather than a replacement of humans.

As to whether ChatGPT will make humans redundant any time soon, the chatbot’s promise and peril are nicely summed up in its own words – “a powerful tool for natural language processing but it is not a replacement for human intelligence or creativity”.

So for now, consider those jobs requiring you to show a little bit of empathy, a touch of fresh thinking, a decent dose of creativity and a sprinkling of human intelligence to be safe from the predatory advances of those increasingly competitive robots.