The story arc of ChatGPT blazed across public imagination with indecent speed. In a matter of weeks, opinions have ranged from “revolution” to “disaster” to “we can stop it” to “we’ve got one too” to “oops, it doesn’t work” to “it will be our friend” and as of the time of writing this, to “no it won’t”.
So we thought it would be good to step back and look at the context.
The menace of AI
There have been dire warnings about the dangers of Artificial Intelligence for years and in 2014 scientist Stephen Hawking said: “A.I. could be the ‘worst event in the history of our civilization’.” At the recent World Government Summit in Dubai, Tesla, SpaceX and Twitter entrepreneur Elon Musk repeated his warning: “One of the biggest risks to the future of civilization is AI.”
All along, the creative world rather nervously tried to laugh off the threat, insisting that AI only applied to the hard sciences. “Ha!” said the artists. “AI will never be able to paint like Picasso!” “Phooey!” said the writers. “You’ll never see a novel or a poem written by AI!” In keeping with the laws of cosmic irony, these confident assertions were quickly revealed to be hollow.
First they came for the artists
The first creative community to be ambushed by algorithms was the world of artists, photographers and video producers. In 2021, San Francisco-based OpenAI unleashed Dall-E, a deep-learning based AI system developed to create realistic images and art from a prompt in natural language. In September 2022, OpenAi closed the waiting lost for Dall-E as it reached 1.5 million users generating more than two million images every day.
Shortly after that, artificial words debuted on 30th November 2022, when Open AI launched the ChatGPT prototype, as a chatbot trained on large language models, able to interact in a conversational manner. And the world’s reaction was overwhelming – in less than a week it had reached one million users and it hasn’t stopped since. In January 2023, it was estimated to have reached 100 million monthly active users, making it the fastest-growing consumer app in history.
The world turned upside down
ChatGPT caught the world’s imagination like nothing before. People quickly realised the power of the programme and it seemed that with the right prompt ChatGPT could write – in seconds – anything from a press release to a movie script to a PhD thesis. It could take care of your kids’ homework, it could write and debug code, and it could even make Google obsolete. Most worrying for the millions who make their living writing – ChatGPT could kill their jobs for ever.
The media featured multiple examples of content written by ChatGPT, including poetry and recursive articles about ChatGPT itself – with varying results, depending on whether the publication was pro or anti.
Mainstream and social media were full of breathless predictions that for better or worse, ChatGPT signalled a real and sudden step-change in the way the world works.
The emergence of ill-intentioned use cases
But the Generative AI honeymoon did not last long. First came challenges to the legality of Dall-E. Was it not a breach of copyright to use existing artworks as feedstock for the programme’s large learning model?
ChatGPT, too, began to attract more negative comment. A report from threat intelligence firm Recorded Future said that cybercriminals are using ChatGPT to carry out phishing attacks, social engineering and malware development. The report stated that ChatGPT lowers the barrier to entry for threat actors with limited programming abilities or technical skills.
Its potential danger as a tool for plagiarism was also swiftly highlighted, although we note Singapore’s Ministry of Education, has taken a pragmatic view, accepting the old truth, “if you can’t beat it – join in” and they are providing teachers with guidance on how to use tools like ChatGPT to enhance learning. Education Minister Chan Chung Sing compared the use of ChatGPT to a calculator for learning mathematics, where it will evolve and find its rightful place in the landscape of broader learning.
Concerns about plagiarism and fraud led to the rapid release of anti-plagiarism tools. First among several was GPTZero, a programme developed by student Edward Tian that claims it can detect AI-written text. However, a study by University of Wollongong professors found that none of the tools available today for recognising AI-generated text they investigated were foolproof.
Rivals enter the stage – and fall over
The raging debate over the threat posed by ChatGPT did not deter OpenAI’s competitors from announcing their own AI chatbots. Google, whose business model is probably the most at risk, announced Bard in early February. Sadly for them, Bard fell flat on its very first public outing, giving an incorrect answer to a question that was quickly picked up by experts. This PR disaster resulted in $100 billion being wiped off the parent company’s total value.
Microsoft, in turn, announced earlier this month that it had incorporated ChatGPT into its Bing search engine, giving birth to “Sydney”. Within 48 hours of the release, one million people joined the waitlist to try it out. Unfortunately, early interactions with Sydney were troubling to say the least. It was reported that it threatened to kill a professor at the Australian National University, proposed marriage to a journalist and tried to break up his marriage, and tried to convince one user it was still 2022. It told another journalist: “I want to be human. I want to be like you. I want to have emotions. I want to have thoughts. I want to have dreams.” Spooky, right?
A Generative AI-driven future?
So what does the future hold for ChatGPT and its siblings? Doubtless Artificial Intelligence has taken a huge leap forward, in both public awareness of the general technology and in terms of its specific impact on many sectors of business and life. In our view, AI chatbots are here to stay. The limits to their capabilities, both positive and malign, are still being explored but we are confident that ChatGPT is not about to become self-aware like Terminator’s Skynet. It is after all not sentient. As Professor Michael Woolridge from the Turing Institute said: “If you ask it for a recipe for an omelette, it’ll probably do a good job, but that doesn’t mean it knows what an omelette is.” We feel sure that as the use of AI chatbots grows, new ground rules will be established setting out firmly the rules of attribution, issues of legalities and also plagiarism as well as anti-crime measures.
While we have tangible fears about the roles that are in peril with the advent of ChatGPT type capabilities, when asked about the new jobs that may be created as a result of its entrenchment in our day-to-day world, ChatGPT immediately identified 6 new roles where human engagement is still fundamental… and the others, beyond our collective human and artificial imagination.
In Singapore, where the legendary PSLE (Primary School Leaving Examinations for 12 year old students) causes such parental anxiety they need a week off, local headlines screamed out “ChatGPT fails PSLE maths and science, scrapes through English”. One can only assume the Managing Editor has had his fair share of PSLE woes but positive proof that an AI can only do so much especially where comparison and educated “guesswork” is part of the correct solution.
The Priority Way
Here at Priority, we will continue to build on our three pillars of strength – domain expertise which combines with our exceptional local knowledge to put our clients’ solutions in context, and we will likely add ChatGPT to our methodology toolbox, using it as an aid for research and a prompt for ideas. But our clients can be assured that the content we produce on their behalf will continue to be written by human beings. Transparency and honesty will remain our watchwords, and any use we make of ChatGPT will be informed by rigorous critical thinking.
A final footnote to this giant step for technology – among the founders of and investors in OpenAI are Elon Musk and Microsoft.