It's been a busy week for AI from the public authorities' point of view: after the birth of theAI Act the European the AI Commission's report Commission was published on March 13, 2024 and submitted to the French President.
A report that takes stock of the current situation and makes 25 recommendations for the country to take advantage of AI opportunities while controlling the risks.
We already knew: France is lagging behind in the adoption of AI... and that's not good news.
The report therefore aims first and foremost to "de-demonize AI without idealizing it". It stresses that the benefits of AI will not be automatic, but will depend on political choices and collective commitment.
Let's start with the facts:
Significant growth potential
According to the report, AI could have a major economic impact. It could double France's annual growth thanks to the automation of certain tasks. After 10 years, GDP could increase by 250 to 420 billion euros, equivalent to today's industry.
Beyond this transitory effect linked to automation, AI also seems to accelerate innovation in a more perennial way. By facilitating the emergence of new products, services and models, it could induce a permanent increase in the rate of growth.
However, these gains are not guaranteed. Recent history shows that France has benefited little from the digital revolution, unlike the United States. To take advantage of AI, appropriate public policies will be needed, in terms of innovation, industry, competition, training, etc.
French companies lag behind
To date, France and Europe are clearly lagging behind in AI. Investment is 3 to 4 times lower than in the United States, on a comparable wealth basis. Only a handful of European companies are positioned in the AI value chain, and none of them are world leaders.
This delay poses a risk of economic downgrading. On the one hand, France could miss out on the AI economy and see its value captured by other countries. On the other hand, existing companies could lose competitiveness to new players.
To close this gap, the report recommends massively redirecting savings towards innovation, with the creation of a €10 billion "France IA" fund. It also recommends facilitating access to data, particularly personal data, making France a major hub for computing power, and supporting an open ecosystem of AI developers.
Contrasting effects on employment
As far as employment is concerned, the report estimates that AI will have an overall positive effect in France, despite uncertainties. On the one hand, the automation enabled by AI will eliminate some jobs, particularly those consisting of routine tasks. But on the other hand, AI should also create jobs in new professions as well as in existing ones.
An empirical study conducted on French companies shows that those who adopt AI see their total employment increase more than others. This positive effect is explained by the fact that AI replaces tasks, not jobs in their entirety. Only 5% of jobs can be directly replaced by AI.
However, this effect is not uniform. Certain administrative and commercial professions seem more exposed to job cuts. And self-employed workers performing easily automatable tasks could face increased competition from AI.
Beyond the effect on employment volume, AI could also widen inequalities. Companies that adopt AI tend to hire more highly skilled and technical profiles, which are better paid. But conversely, AI also seems to benefit the least skilled or productive workers initially.
To support these transformations, the report stresses the importance of initial and continuing training. It recommends investing in observation and research into the impact of AI on employment. Social dialogue is also seen as essential to building AI uses in a partnership-based way.
Impact on daily life
A technology that's already very present
Beyond the economic sphere, AI is increasingly present in our daily lives. According to a survey, 55% of French people say they are familiar with ChatGPT one year after its launch. But AI applications go far beyond that: facial recognition, translation, content recommendation, voice assistants and more.
This omnipresence arouses both fascination and fear in public opinion. 77% of French people see AI as a real revolution, but 68% are in favor of a pause in its development. This ambivalence is nothing new. In the past, many innovations (trains, electricity, etc.) have aroused fears, sometimes unfounded, sometimes justified.
To promote the acceptability of AI, the report calls for educational work and public debate. It recommends launching a vast plan to raise awareness and train the nation, drawing in particular on education and research.
Increasingly present personal assistants
Among the consumer applications of AI, voice assistants like Siri or Alexa are having a growing impact on our daily lives. They enable many tasks to be carried out without human intervention: listening to music, obtaining information, controlling connected objects, etc.
In the field of customer service, conversational agents are also developing rapidly. They are capable of answering basic questions in a fluid, natural way. Their deployment enables companies to reduce costs and improve service availability.
In the future, personal assistants are set to become increasingly intelligent and autonomous. They could become true everyday companions, capable of learning our preferences and anticipating our needs. Their mode of interaction should also evolve towards more natural, integrated interfaces.
Impacts on mobility and health
Two areas where AI could have a major impact are mobility and healthcare. The development of autonomous vehicles promises to radically transform the way we travel. It could reduce accidents, smooth traffic flow, facilitate parking and even reorganize urban space.
In healthcare, AI is opening up new perspectives in diagnostics, personalized medicine, epidemiology and prevention. Medical decision-support tools are being developed, capable, for example, of detecting cancers on the basis of imagery. Ultimately, AI could enable continuous, personalized monitoring of each patient.
However, these innovations also raise ethical and liability issues. They will require us to adapt our legal and insurance frameworks. The protection of highly sensitive health data will be a major challenge. The report calls for a societal debate on these issues.
Energy-hungry technology
Another challenge for AI is its environmental impact. Training large-scale AI models consumes large amounts of energy. According to one estimate, AI could consume between 85 and 134 TWh of electricity worldwide by 2027, equivalent to the consumption of Sweden.
This consumption is linked to the computing power required, which relies on energy-hungry processors. Their production also has an environmental impact, due to the extraction of rare materials. However, processors dedicated to AI represent only a tiny fraction of global production.
Faced with this challenge, the report calls for France to become a pioneer in sustainable AI. It recommends greater transparency on the environmental impact of models, directing research towards more sober solutions, and mobilizing AI itself to accelerate the ecological transition.
The 25 recommendations and 7 priorities (in blue to speed things up):
- Launch an AI awareness and training plan for the nation to create the conditions for collective ownership of the issues.
- Invest massively in digital companies and business transformation, notably via the creation of a €10 billion "France & AI" fund, to support the French AI ecosystem.
- Make France and Europe a major center of computing power, in the short and medium term.
- Transforming the approach to personal data to continue to protect while facilitating innovation.
- Promote French culture by providing access to cultural content while respecting intellectual property rights.
- Assert the principle of an "AI exception" in public research to boost its attractiveness.
- Structure a coherent diplomatic initiative aimed at founding global AI governance.
- Generalize AI deployment in all higher education courses and acculturate students in secondary schools.
- Invest in continuing vocational training for the workforce and training schemes around AI.
- Make social dialogue and co-construction the cornerstone of AI use.
- Equipping public agents to transform administration with AI.
- Better care thanks to AI by giving more time to care.
- Better education thanks to AI via individualized student support.
- Sovereign computing capabilities.
- Access quality data.
- Attract talent to build the technologies and uses of tomorrow.
- Massively deploy AI in the economy.
- Build the international governance of AI that is lacking today.
- Develop an AI systems evaluation capability in France.
- Avoid dominant competitive positions.
- Facilitate the training of AI models while respecting intellectual property rights.
- Greater transparency on the environmental impact of AI models.
- Directing research towards more sober AI solutions.
- Mobilizing AI itself to accelerate the ecological transition.
- Create a "1% AI" solidarity mechanism for developing countries.
What do we think?
While the report and the definition of the issues surrounding the AI revolution already underway are accurate, the 25 (!) recommendations have an incantatory air that can leave one dreaming...
And unsurprisingly, the temptation to regulate what doesn't yet exist is always present in the subtext, with, let's say... uncertain economic impacts, already seen in the recent past(cuckoo RGPD).
Nevertheless, let's remain optimistic: as the report points out, France has a number of strengths, but must act quickly and decisively if it is not to fall behind.
A certain band of gorillas 🦍 is already at work 😁.