Opinion: Late last year was an inflexion point as ChatGPT-3 (or, more accurately, ChatGPT-3.5), was launched.
The results ChatGPT-3 generates can be stunningly good or they can range across the spectrum to terrible and are often laughed at. Some results even contain invented sources/citations, which have been dismissively called, “AI hallucinations”.
People laughing at the often basic errors is normal and to be expected. New technologies are rarely as good as existing technology. Yet, as history has shown, over time the new technologies often surpass the previous technologies.
Remember, for instance, the arrival of digital cameras? The early mass-produced ones from the early 1990s produced photos that were nowhere near as good as traditional photographs. But as the quality of the digital photos improved and the cost of digital cameras fell, digital cameras slowly replaced analogue Single-Lens Reflex cameras, which are now rarely used.
There are two points of differences, however, with GPT and other AI generative tools and the general trend of new technologies replacing old. First, as the digital camera example shows, the replacement of technologies has traditionally been slow. In contrast, the speed of change in technology with GPT and other AI generative tools is exponential. It took just five days for ChatGPT to hit 1 million downloads and the release of GPT-4 in March 2023 has made ChatGPT-3 in effect obsolete.
Though much of the focus has been on ChatGPT and text generation, there have been a swathe of AI generative tools – covering art, design, video, voice, music and computer code – over the past few months.
Second, normally it is the case of a new technology replacing an existing one, such as digital cameras replacing SLR cameras, cars and trucks replacing horse and carts and so on. However, with GPT and other AI generative tools, human knowledge and skill is being replaced. Indeed, tech industry leaders are calling for a pause on the continued development of AI beyond GPT-4, because of AI’s societal and other impacts.
Of course, some new technologies have replaced human labour. Bulldozers and other machines replaced some physical labour but those were isolated examples and new jobs were created to drive and operate those machines.
So what do the AI generative tools mean for the future of the world of work? First, we’re not talking about the future of work. The future is here. These technologies are already in use. For some advertising agencies, what previously had taken three days and required a range of people, can now be done by one person in under three hours. There is little need for copy editors and graphic artists in that advertising agency. It is easy to understand the predictions that in the US and Europe up to a quarter of current work done by people will be replaced by AI tools and automation.
The advertising agency example does not mean, however, that that “one person” can be anyone. For a person to achieve high quality content for the advertising agency, they must be highly skilled, not just in marketing/advertising but also in the use of a broad array of AI generative tools.
As with the use of all new technologies, there will be winners and losers. The winners are those people and organisations who can learn new skills and strategies and harness AI generative tools. The losers will be the ones who cannot use the new technology effectively. History is littered with examples of skilled people rendered jobless who were not able to adapt to the new technology.
Granted, it will be a painful time for those who cannot change. But no one should be surprised. Over 50 years ago, the futurist Alvin Toffler wrote in Future Shock: “The illiterate of the twenty-first century will not be those who cannot read and write, but those who cannot learn, unlearn, and relearn.”