OpenAI’s artificial intelligence arrived on the market on November 30, 2022. In two years, the tool has appeared revolutionary. But the American start-up is now faced with limits.
Follow me
Some say that ChatGPT is not the revolution announced and others assure that it has already become part of customs for eternity. His boss, Sam Altman, was bombarded with nicknames, sometimes even nicknamed “the Oppenheimer of his time” in reference to the father of the atomic bomb.
No need to search from noon to 2 p.m., ChatGPT has turned the job market and our daily lives upside down. While part of the population has adopted the tool, this significant advance in generative AI is considered a major technological revolution which builds on several decades of digital change.
What is the real outcome?
“Interest from organizations, but also from the general public, remains very high. And the subject continues to fascinate,” observes Johan Rochel, researcher at EPFL and philosopher from Valais, co-founder of the Ethix innovation laboratory.
He continues:
“For questions of ethics, these two years have been fascinating because they have made AI tools tangible for everyone. Questions of data quality, explainability, design, responsibility but also the impacts on the professional world were previously interesting, but far from people’s reality. We’ve been there for two years, people can test it. ChatGPT has therefore democratized the ethics of AI.”
A new proximity that the public has seized upon. The media have dissected the benefits and harms of such technology in the hands of ordinary people. The slayers of technology wanted to measure themselves against the machine. The French philosopher Raphaël Enthoven, for example, took up the challenge and beat the machine, achieving a 20/20 in the philosophy baccalaureate test. ChatGPT settled for an 11/20.
This opposition between man and machine has animated the human race for a long time. Now, it is more real than ever and fuels a fear from science fiction: being dominated by digital brains.
Yet the excitement around AI appears to be stagnating. “To revolutionize consumption habits, it’s yes, but for a complete revolution, it’s no. Templates are very strong at generating text and code. I do it every day. But they are not yet capable of learning in the physical world,” informs Raphaël Sculati, technology director of The Computer Firm (TCF).
According to Johan Rochel, “the euphoria and hype maintained by the industry are running out of steam a little,” he concedes.
“It’s tiring to say that every day the revolution takes place. It’s nice that things are calming down.”
Johan Rochel
This observation, according to the researcher, underlines that “everyone realizes that using AI tools requires training and time, that it is not enough to open your program and wait for things to happen. are happening.”
Johan Rochel explains that the demands made by his clients are evolving:
“It is no longer simply a question of learning to use the tool (prompting for example), but of implementing good governance in the organization and preparing the teams to increase their skills. This includes fully understanding the limitations and risks of the tool.”
Generative AI is sweeping the world and the employment sector. Many people write their cover letter thanks to ChatGPT. AI has also found its place in the office. From simplified formulation to planning, the different professions connect to the OpenAI platform to optimize their time. Companies use it, for example, for meeting minutes.
People active in the banking sector confirm using it for certain tasks. Same observation in the medical sector, where doctors tell us to question the chatbot to strengthen their diagnosis. In the special education sector, professionals even tell us they use chatbots to simplify their speech.
Raphaël Sculati is himself convinced by the object: “I chat with an AI to help me in my sports planning and better understand how to manage conflicts.”
In 2023, conversational chatbot technology even entered the political arena, when a Japanese parliamentarian used ChatGPT to formulate questions to Prime Minister Fumio Kishida in the development of a draft amendment relating to the policy to combat the Covid-19 pandemic.
The baby of OpenAI has evolved into an assistant 2.0 for a large part of the world’s population. According to data from Similarweb, more than a quarter of a billion unique visitors visited ChatGPT.com each month between June and August 2024, resulting in an average of 2.5 billion visits per month.
ChatGPT is approaching its limits
But there is always the other side of the coin. Today, some leading AI scientists are talking about the limits of OpenAI’s “bigger is better” philosophy.
Ilya Sutskever, co-founder of AI labs Safe Superintelligence (SSI) and OpenAI, recently told Reuters that new ideas are now needed to take the next step. Raphaël Sculati approves:
“ChatGPT remains limited to date, but the evolution can be exponential”
The skyrocketing rise of OpenAI, which in a short time has become the most capitalized unlisted Western company in history, is in the adaptation phase. The cost of producing such technology is astronomical and billions are needed to run the shop.
From an energy point of view, too, the bill is steep, with environmental consequences. To meet the high demand for energy, coal-fired power plants have had their lifespan extended to feed AI.
A major player in the sector, Microsoft, has also abandoned its own climate objectives to be able to participate in this digital revolution which is only just beginning.
