Artificial intelligence, between opportunities and challenges

The advent of a technology with a thousand promises

From science fiction to tangible reality, Artificial Intelligence has made spectacular headway over the last few decades. This technology has invaded our daily lives, shaking up the global economy, the job market and our private lives.

Against this backdrop, 2023 has been dubbed the "Year of Artificial Intelligence" for France and Europe. In fact, France has been a real driving force on a European scale , with 35% more companies adopting AI than on the continent as a whole (32%).

In fact, on March 13, 2024, the European Union officially became the first political power in the world to establish a legal framework to regulate Artificial Intelligence.

Indeed, while the algorithms at the heart of this technology promise many advantages - task automation, efficient data processing, decision support... - they also bring with them new issues and risks.

Jobs and positions under threat, the limits of machine learning and deep learning, security and liability... Critics have plenty of arguments. On the other hand, the pro-IA side is quick to point out the possibilities and opportunities that are opening up as technology advances.

Between enthusiasm and apprehension, only one thing is certain: the last few months have heralded the arrival of a new era that will mark multiple upheavals on an economic, political and social level.

So what can we expect from this new era? What changes are we facing? What challenges do we face? Let's decipher the impact of a revolutionary technology.

A disruptive technology

History and birth of AI 

Artificial Intelligence has been on everyone's lips for some years now, and especially in recent months. While it may seem new to some, the concept dates back to antiquity. The term AI, on the other hand, was born in the middle of the 20th century, with the work of pioneers such as Alan Turing and John Von Neumann.

In the 1950s, Turing proposed a test to evaluate a machine's ability to convince a human that it is human, by means of an online conversation. The Turing test and all his ideas laid the foundations for AI research. The publication of the first mathematical models and the first neural network computer in 1960 marked a real turning point for artificial intelligence.

But although the birth of AI raised many expectations, the craze died down in the 60s due to the difficulties encountered. This was the first winter of Artificial IntelligenceDuring this phase, scientists, researchers and public authorities began to distance themselves from the technology.

It wasn't until the 70s and 80s that the excitement began to build again, thanks to the arrival of new technologies such as microprocessors and expert systems. Even then, however, AI lost popularity to the point where the term itself became almost taboo, replaced by more modest terms such as "advanced computing". Despite Deep Blue's victory over Garry Kasparov in 1997, funding for Artificial Intelligence remained low, and it wasn't until the 2010s that massive data and graphics cards gave it a new lease of life.

So why are we talking so much about Artificial Intelligence today? Because its use has been democratized, notably with the June 2020 launch of GPT-3 by OpenAI, followed by the emergence of tools such as DALL-E. This new era of AI is not just revolutionizing the world of technology: it's making its mark on our economic, professional and social reality.

Types of Artificial Intelligence

Although the term "Artificial Intelligence" is well known to everyone, it's also widely misunderstood, not least because the scientific community can't agree on a real definition. As a result, the word covers a wide range of possibilities, adding to its complexity.

We can nevertheless give a fairly broad definition to ensure that we approach this technology correctly: it's a discipline whose aim is to achieve have a machine imitate a human's cognitive abilities.

Once we've said that, it's necessary to differentiate between the three main types of Artificial Intelligence: weak AI; strong AI and super AI.

Weak AI (also known as narrow AI)

This is a form of AI designed to perform a specific task, often better than humans, but limited to that particular task. For example, voice or facial recognition systems. Today, this is the only type of AI that can be implemented.

Strong AI (also known as general or deep AI)

Unlike weak AI, strong AI is capable of understanding, learning and performing any intellectual task that a human can do. This includes the ability to reason, learn from experience, understand natural language and so on. This AI does not exist, but it is the focus of many researchers' hopes.

The super AI

This AI would be self-aware and would far surpass human cognitive abilities in all areas. Superintelligent AI is still purely hypothetical and remains a subject of debate among researchers.

AI: a major impact on our lives

Widespread fields of application

The impact of Artificial Intelligence is all the greater for the fact that this disruptive technology is present in almost every sector of activity, with a diversification of fields of application over the years. Healthcare, finance, marketing, transport, education, industry, security, customer service, gaming and even human resources: nothing escapes AI.

Between opportunities, added value and cost reduction, even sectors that might have been thought impervious to Artificial Intelligence have finally adopted it, or are in the process of doing so. Education is a good illustration of this mechanic: adaptive learning paths and systems, personalized feedback, automation of administrative tasks... AI promises numerous possibilities for enhancing pedagogy and improving the student experience.

The fear of being replaced by AI

Industry 4.0 benefits directly from the gains of this new technological revolution. It is transforming the professional landscape and the job market, bringing with it a wealth of potential: automation, data analysis, optimization... it's a real "game-changer".

In an article published by Goldman Sachs, it was estimated that generative AIs such as Bard and ChatGPT could automate up to 25% of the entire job market. This figure is much higher if we look at the totality of existing AIs on the market, which obviously concentrates many concerns.

What will be the new practices? Which jobs are likely to disappear? Will my job be replaced by algorithms? How can I train quickly and efficiently?

This is where the big challenge for today's workers begins. Their fear? To be one of the jobs that will disappearbeing dispossessed of their know-how and seeing their value on the job market diminish.

While there is no general opposition to the use of algorithms, the reality on an individual scale, or according to profession, demonstrates real brakes, such as in the legal professions, where almost 32% of working people vote in favor of restricting or banning ChatGPT (GlassDoor study, 2023).

There are also demographic differences, with only 17% of 21-25 year-olds wanting ChatGPT to be regularized, compared with 28% of those aged 45 and over.

So, no matter what promises are made, fears abound and concrete answers are needed in the face of uncertainty. The majority of the scientific community is reassuring: the aim of artificial intelligence is to free employees from low value-added tasks, so that they can concentrate more on more stimulating and profitable activities. They also emphasize that, while algorithms may be capable of replacing humans for certain tasks, they cannot cover all the functions of every job.

Nevertheless, it must be stressed that AI remains a real threat to certain professions, as shown by the "Future of Jobs" report, published in May 2023. It predicts a net loss of 2% of jobs by 2027, due to the destruction of 83 million jobs versus the creation of 69 million.

AI: the emergence of new professions

We're talking about the creation of 69 million new jobs worldwide, and the transformation of many existing positions. To illustrate this phenomenon and reassure the public, many echo the work of economist David Autor, who in 2022 revealed that 60% of today's workers are in occupations that did not exist in 1940.

Faced with these alterations to the world of work, we can either be offended or amazed, but one thing is certain: this reality is not going to change, and we have to jump on the bandwagon of the technological revolution that is underway. For example, a survey conducted by ResumeBuilder in November 2023 revealed that for 96% of companies hiring in 2024, skills and knowledge around AI would be real assets for candidates. So how do you stay competent in a market that is evolving very, sometimes too, rapidly?

We need to acquire AI-related skills related skills, which are increasingly important, sought-after and valued. Training is the key to these professional transitions.

While some professions will evolve, others will emerge bringing with them new perspectives! Here's an example of five new professions coming onto the market: AI data developer, AI engineer, prompt engineer, psydesigner or cognitician and robot coach.

The emergence of new needs directly linked to artificial intelligence, machine learning and algorithms raises questions: are we on the verge of a replacement of man by AI, or rather a beneficial collaboration?

Here, there are two opposing views. For Manuel Davy, CEO of Vekia and head of the Cité de l'AI, the second option is more likely to prevail. Because it's a disruptive technology, it's bound to give rise to new needs, calling for new skills and therefore the emergence of new professions and an influx of training to retrain workers. As he puts it: "AI lacks the critical thinking and ethics of humans, so it will always need them to supervise it".

Impact on the general economy

At the dawn of 2025, the impact of artificial intelligence is spreading into our economic reality. Complex and difficult to identify, their evolution will depend largely on political decisions and arbitrations. Economic growth, territorial inequalities, productivity gains and deflation: the repercussions of industry 4. 0 and the principle of cobotics are taking shape.

The Internet of Things (IoT), cyber-physical systems, massive data processing: the increase in productivity is obvious, whether through automation or process simplification. This means lower costs, higher profits and the emergence of new markets and opportunities.

By way of example, Statista estimated the market for generative AI to be worth 42 billion euros worldwide in 2023, compared with a projection of 1,812 billion in 2030, leading to a global economic upheaval.

However, the benefits of this technology will not be evenly distributed, as the USA and China could account for up to 70% of the economic fallout.

AI: the ethical challenges

Are our societies changing because of AI?

The "Artificial Intelligence" earthquake has not only transformed the economy: society itself has been turned upside down, far beyond the organization of work. This technology is triumphing in the spheres of education and the arts, as well as in justice and healthcare. A double-edged sword, it is clear that today's algorithms will transform and shape tomorrow's representations.

There was a time when we were convinced that certain fields, such as culture and art, were the prerogative of the human species. Today, this is far less obvious: in 2022, for example, it was an image generated by the AI MidJourney that won an art competition in the USA.

Another important aspect is the impact on our consciences and thepossible dumbing-down of our societies. A number of studies show a correlation between the development and adoption of AI and the dumbing-down of human beings, highlighting issues such as critical thinking, the ability to think, decision-making... This raises a real challenge: how do we mitigate this risk? Among other things, this mechanism raises questions about democracy and free will.

But AI's impact on our societies is not just negative. In fact, this technology can make certain products or services more accessible to disabled or less well-off populations.

AI transparency, traceability and explicability 

The widespread use of AI raises legitimate fears about the transparency, traceability and explicability of its results. Many experts highlight concerns such as the repetition of patterns and biases (racist, sexist...) or the standardization of discourse.

This is the case for justice where, according to ProPublica, one of the AIs used in American courts repeated racist biases by predicting twice as many false positives for recidivism on black defendants as on white defendants.

The same deviation can be found in healthcare. On October 20, 2023, a study published in the journal Digital Medicine questioned the choice of AI for doctors' note summaries because of the reproduction and propagation of racist prejudices and misinformation.

In addition to the repetition of these patterns, it is the transparency of the algorithms in their entirety that is under fire. A study carried out by Rishi Boullasani and other researchers at the Institute for Human-Centered Artificial Intelligence revealed that companies such as Meta, Open AI and Google were far from scoring well when it came to the transparency of their "foundation model". transparency of their "foundation model"..

Transparency is a real issue: it protects all stakeholders. It is essential to be able to understand the algorithms and how the model works, both upstream and downstream, and to be able to justify the results obtained. This last point is one of the " top fears" surrounding AI: inaccurate decisions and inconsistencies are feared more than loss of business! As far as machine learning and deep learning are concerned, the challenge of self-fulfilling prophecy also comes to the fore.

Liability and legal aspects of Artificial Intelligence

On the legal front, we're navigating a complex, unregulated landscape in which unprecedented situations are emerging. Whether government institutions, companies or insurance organizations, they all face the pressing need to regulate practices and usages around Artificial Intelligence.

In 2023, the first agreements emerged to support the development of ethical and reliable AI with risk management at four levels (minimal to unacceptable risk). These advances continued and, in February 2024, the European AI Office was created to oversee the application and implementation of AI law with member states.

These early beginnings are far from globalized and the legislative framework is struggling to keep pace with the explosive technological advances. The next few months will be crucial for framing and passing legislation, particularly in a globalized economy.

Privacy and security 

AI is largely driven by massive data. So when we talk about data, we immediately think of its collection, management and respect for privacy.

And here again, trust is out of the question, and with good reason: the CNIL has highlighted the difficulty of protecting AI systems, pointing out that the statistical nature of their construction makes them vulnerable.

This brings us back to the RGPD-IA correspondence. Because the risks associated with the use of AI are not limited to economic issues or biased results: privacy breaches or security failures are just as important and feared. This major challenge must be met if we are to have truly sound and ethical practices.

A world of uncertainty

AI has transformed our daily lives, and will surely continue to do so as it advances. Crystallizing everyone's fears and excitements, it's certainly the uncertainty surrounding the impact of this technology that makes it so difficult to grasp.

In an interview, Laurent Charlin - senior academic member at the Institut québécois d'intelligence artificielle - declared: "To be perfectly honest, when ChatGPT was released in November 2022, I think that the quality of this machine surprised a lot of us.. I don't think anyone knows exactly what can happen. Even in our community, it's debated. Some people see a less rosy future than others."

It's impossible today to define the full impact of artificial intelligence, but one thing is certain: its democratization and evolution will continue to transform the economic, professional and personal landscape.