Between fascination and interrogation, AI arouses enthusiasm and business debates.
Capable of upsetting practices by effectively automating certain tasks or accompanying activities hitherto reserved for humans, it also raises major issues: technological maturity, environmental challenges and social responsibilities. Consequently, the sustainable integration of AI into our organizations requires above all an ethical, adaptive and rational approach.
An adoption in the exploratory phase
For organizations, the potential of AI is very promising. However, unlike the predictions of 2022 and the arrival of Chatgpt, the adoption of AI in practices remains marginal. Currently, companies are increasing tests, concept evidence and pilot projects. In project management for example, only 5% of employees use it daily, 36% occasionally and 48% rarely or never.
How to explain such a gap? First, by a lack of skills and knowledge of the uses of AI. For a successful integration, decision -makers must grasp the challenges in order to have a clear strategic vision which will lead to a lasting transformation. Employees must also be supported to prevent brakes on possible changes and resistance.
The technical challenges remain considerable. In a constantly evolving environment, what investments to favor? A homemade solution or a tool on the market? Finally, the question of hallucinations (false or misleading answers) arises, although they are less frequent, requiring specific training models (fine-tuning, prompt tuning, etc.) to be reliable enough.
Major environmental and ethical challenges: a happy medium to find
The potential of the AI and the craze it arouses should not obscure its environmental and social impacts. On the ecological level, the resources necessary for its operation are often pointed out. A single request on Chatgpt, consumes 21 times more energy than a search on Google, it’s colossal. In terms of infrastructure, data centers, which operate AI models, require immense resources in electricity and water for the cooling of servers. We often oppose Green AI (frugal IA) and AI for Green (IA for sustainable development), but it is the alliance of the two which offers the keys to meet environmental challenges.
On the ethical side, data management and social breakage challenge. European regulations (GDPR, AI Act, Data Act, Data Gover Act) supervise AI: data protection, employees, individual freedoms, compliance with intellectual property, digital sovereignty and the conditions for the exercise of “click workers”.
These effects on the planet and the company question: what perimeter of use for AI? How to guarantee respect for the law? How to ensure good quality of training data?
Regulating AI’s activities is fundamental. It is the sovereignty and sustainability of the companies that is at stake!
Expertise: a major issue for maturity and industrialization of sustainable AI solutions
Without expertise, no reliable AI. Ensuring the rise in skills of the various stakeholders involves the creation of a center of excellence on the AI (AI COE). For what purpose? Respond to 3 challenges: acculturation and team training, sharing good practices and incubation of new solutions.
- 1. Train to better control AI.
The teams must master various skills in order to send technical and functional challenges on internal or external subjects. Many functions being involved, the transversal and multidisciplinary approach is essential to better meet customer objectives.
- 2. Share and structure the means
Beyond talents, resource management is crucial: computers equipped with graphic processors, local servers and online. These resources, sometimes expensive, must be shared and prioritized according to the objectives of the company.
- 3. Innovate to keep a step ahead
The experiment is essential: buy an existing tool or develop internal solutions? In the first case, data governance and cybersecurity are central. In the second, the use of the OKR method (Objectives and Key Results) makes it possible to define clear objectives and to measure the effectiveness of the solutions to adjust the strategy continuously.
Gard with alarmist speeches or disproportionate promises, AI must be tackled lucid. As of today, we must structure its uses, invest in skills and work in a team. We have the opportunity to make it a lever for competitiveness and positive impact. It’s up to us to exploit it intelligently!
