“It has been a long time since there had been so much enthusiasm for technological innovation. Although ChatGPT and its clones have reignited debates about the role of human labor, beyond the strengths and weaknesses of technology, we can affirm the following: First, that the adoption of ChatGPT will take longer than anticipated. Second, the chatbot business model does not have the levels of profitability demanded by investors today, so it seems fragile in the current environment. Third, the use of Big Data will continue to be instrumental in the adoption of Generative AI technology. And finally, the need to skill and reskill their workers will increase rapidly, as human capital problems (and skills obsolescence) will remain fundamental.”
This is how Jacques-Aurélien Marcireau, co-head of equities at Edmond de Rothschild AM, explains the phenomenon of Artificial Intelligence with the ChatGPT boom.
“We believe there is a clear parallel between today’s ChatGPT and 2015’s autonomous vehicles. Back then, everyone was impressed by the quality of self-driving cars and autopilots developed by Tesla (NASDAQ:TSLA) and Google (NASDAQ:GOOGL). But 7 years later, both car manufacturers and tech companies have come back to earth. The technology will be ready one day, but the human brain still struggles to predict the exact timing of these initiatives,” explains Jacques-Aurélien Marcireau.
For this expert, Chat GPT on an industrial scale is no exception. “Perhaps this advanced conversation tool is the future, but its strategy and development model belong to the previous decade (2010), characterized by cheap money and rising markets. Companies burned money to be first in their market and create favorable scale effects, such as Uber (NYSE:UBER), Airbnb (NASDAQ:ABNB), DoorDash (NYSE:DASH) and Groupon (NASDAQ:GRPN),” he explains.
“ChatGPT searches currently represent a huge cost to OpenAI, the company behind this technology. However, these searches are still free. At this rate, OpenAI would have to spend more than a billion dollars each year simply to satisfy the requests of netizens. To this should be added the expenditure on R+D. A monthly fee of $42 could certainly reduce the cost, but users would expect a higher level of copyright protection and serious efforts to fight misinformation. Unfortunately, this race for prominence means that any notion of energy efficiency is also relegated to the background,” says Marcireau.
“ChatGPT is above all a reminder that access to data is essential for such a large-scale model and the most powerful piece of this value chain, especially with increasingly competitive algorithms, as is already the case. The chatbot is also a reminder that big data will continue to be an important investment topic in the coming years,” adds this analyst.
“We’re excited about enterprise applications backed with enough data to generate verticalized versions—well-defined areas where ChatGPT can have an impact. However, its use will be restricted in areas with fewer routines and less freely accessible historical data. This is why Generative AI is having a hard time getting off the ground in areas such as design and simulation. It should be noted that Getty Images is attempting to create jurisprudence on training on its models by filing a complaint against Stable Diffusion for copyright infringement.”
What about human capital?
“We’re not too worried about a world without jobs. On the contrary, serious labour shortages persist in several areas such as health, services and research. There is no point in talking about the end of human work when we need engineers, technicians and researchers to help fight climate change. If generative AI presents a risk, it lies in the need for highly skilled labor. The arrival on the scene and rapid dissemination of ChatGPT is another example of the acceleration of the obsolescence of human capital,” says Marcireau.
“As a result, governments will have to adapt school curricula and companies will have to rethink vocational training. For companies that are firmly committed to training, this represents an important competitive advantage that will allow them to widen their innovation gap and strengthen their advantage,” he adds.
As an increasing number of tasks are likely to be automated, this should also reduce poor working conditions and free up energy for the challenges we will have to take on. “The economist Alfred Sauvy would have agreed with us. His theory of transfer argued that productivity gains are beneficial, as demand and jobs can shift towards higher value-added tasks.”
“However, this does not mean that the future will be easy. A crucial challenge in the coming decades will be meeting the needs for reclassification, upgrading and retraining, while helping people manage non-linear career paths. We may even need to revise the basic principles of school learning,” concludes Marcireau.