As journalists, we are always aware that with the rise of technology, our jobs may very soon be rendered obsolete - and hence the strive for always doing better. With systems such as Meena (a large language model developed by Google for chatbots and conversational applications) there is increasingly less of a need for human intervention. However, 2023 might just be the year that technology comes for us with the introduction of ChatGPT, an artificial intelligence (AI) system that can generate human-like text in response to simple keywords put in by users.
This AI system seems to be all that educators, marketers, technologists, journalists and writers are talking about as we excitedly test the platform’s scarily human-like ability to produce well-researched content in seconds, and contemplate what it will mean for the future of many writing-based roles and for the people who rely on them. According to CARMA, a media intelligence company, interestingly, currently netizens are more confident than they are confused about the technology. This may be due to content creators wanting to portray themselves as using the technology effectively.
Hilariously, CARMA states that many netizens are also using ChatGPT to write alternate endings to the popular series Game of Thrones.
Nonetheless, if you are one of the many, openly admitting to be confused about the technology, know that you aren't alone. In this article we break down what exactly is the platform and ask industry players if it can actually revolutionise the way we study and work - and what does it really mean for the world of advertising?
We break it down.
Don't miss: Nippon Paint embraces AI in Paint New Happiness campaign
What is ChatGPT
ChatGPT is a generative AI language model that can create original content in response to a user prompt, according to a report by McKinsey. It is a model that has been trained by OpenAI, an AI research and deployment company, to interact in a conversational way.
“The dialogue format makes it possible for ChatGPT to answer follow-up questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests,” wrote OpenAI on its site. It added that the model is a sister platform to InstructGPT which can perform natural language tasks using text prompts. ChatGPT, unlike InstructGPT though, has a different data collection setup.
In 2019, Microsoft invested $1 billion in OpenAI, the tiny San Francisco company that designed ChatGPT. In the years since, it has quietly invested another $2 billion, according to two people familiar with the investment who requested anonymity because they were not authorized to speak with the media.
How was ChatGPT created?
Complex systems require complex building processes and so did ChatGPT which was launched by OpenAI in November 2022. According to the company, here's how the system came about:
“We trained an initial model using supervised fine-tuning: human AI trainers [who] provided conversations in which they played both sides—the user and an AI assistant. We gave the trainers access to model-written suggestions to help them compose their responses. We mixed this new dialogue dataset with the InstructGPT dataset, which we transformed into a dialogue format."
It continued by saying that they took conversations that these AI trainers had with the chatbot to create a reward model for reinformed learning for the platform.
So, what exactly can ChatGPT do?
Currently, Chat GPT can do a lot. According to McKinsey, it can undertake interaction labour in a way that approximate human behavior closely and, in some cases, imperceptibly. This means that we can expect it to be able to come up with personalised marketing, social media captions, sales content and more, said McKinsey. It can also answer complex questions by pulling from vast amounts of legal documentation and draft and review annual reports.
Does it have any limitations?
With a system this scarily human-like, of course the question on everyone’s mind is what exactly can this system not do. According to experts, one area it still lacks is in credibility of the content thrown out.
According to OpenAI, ChatGPT has the potential to sometimes write “plausible sounding but incorrect or nonsensical answers”. It is also sensitive to tweaks to the input phrasing or attempting the same prompt multiple times. “The model is [also] often excessively verbose and overuses certain phrases, such as restating that it’s a language model trained by OpenAI.”
This essentially means that while it is a very highly intelligent system, it does have its flaws and which will mean that human intervention, at least for now, is still a necessity (phew).
How did ChatGPT get so popular in just under three months?
According to WSJ's podcast The Journal, ChatGPT exploded on to the scene with very minimal marketing. In fact, if you go to the ChatGPT site and attempt to try out the system for yourself, you will very often get a message saying that they system is too busy and to try again another time. This probably comes as no surprise considering that the model hit one million users in less than five days of its launch, according to a tweet by OpenAI's co-founder Sam Altman.
According to social listening company CARMA, social conversation around ChatGPT has remained consistently high since the programme was first launched in early December. This indicates people remain curious about it. There remains so much conversation out there because the demand for ChatGPT content is still high, and netizens are creatively finding new ways to create content around it.
When ChatGPT was first launched, within the first few weeks there was a spike in social conversations that start with “I asked ChatGPT to…” . This was around 7% of all conversations on social media.
Million dollar question: How can marketers make use of ChatGPT?
Despite its significant ability to revolutionise the way we work, market and advertise, industry heavy-weights are still skeptical about the reach of the model and its overall impact.
“AI is the current buzzword, but the reality is that it’s not new and won’t be changing the industry dramatically just yet,” said a spokesperson at PR agency Publicis Groupe. What ChatGPT will do is raise the profile of AI as it highlights how effective AI can be in aiding human behaviour or tasks, and it is the existing solution that will likely benefit the most from increased funding and growth.
"ChatGPT will also encourage the adoption of existing AI-powered tools already used in advertising and hopefully enable more effective and efficient personalisation without intruding on our personal lives. But this is only possible when all the stars align in terms of people, processes, platforms and data,” she shared. The spokesperson went on to add that what might prove useful though is looking up statistics on existing paid services that utilise AI and to see if the recent popularity of ChatGPT has driven up sign ups.
Prashant Kumar, the founder of marketing consultancy Entropia (now part of Accenture Song) said Chat GPT is one big step closer on the journey to sentient machines. It’s ability to not just curate information but also process it, can be transformational for scalability of personalisation, immersive experiences and commerce. Its ability to tap unknown unknowns can invigorate creativity.
Having said that, he added, " We are yet to see Sparrow from Google deep mind’s stable, and a few others who are focused on generative AI. ChatGPT may only be the opening salvo in this new game of mind and machine. Quantum computing and at scale 5G deployment will take AI’s real time applications to a new level."
It’s important still to not confuse this with AGI (artificial general intelligence). We are far far away from that, if ever it were to be possible.
Artificial intelligence: Why does it matter?
How artificial intelligence is transforming advertising
New consumer personalisation platform launches