A book to understand the challenges of AI
Indecipherable AI?
Artificial intelligence (AI), or should we say machine learning, fascinates us as much as it worries us. This modern-day unknown is the stuff of fantasies. On the borderline between science fiction and augmented reality, AI is an enigma.
Just think of a program that feeds on data and learns from it. Simply magical. What's even more surprising is that it works wonders and that AI is now everywhere. Visual recognition (including facial recognition), autonomous cars, predictive maintenance, precision medicine, chatbots, intelligent assistants, automatic text generation (GPT3) and assisted programming (GitHub Copilot) are just some of the AI-enabled tools we'll soon (or already) be using in our daily lives.
Should AI be regulated to protect businesses?
But what about our protection in the face of risk, excessive surveillance, statistical bias and predictive manipulation?
These are the questions addressed by the collective book "CODE IA: Software versus Hardlaw", the result of reflections by the Manufacture Thinking think tank under the direction of Xavier Comtesse. In the opinion of the computer scientists and legal experts in this book, the rules of work and work itself need to be completely revisited. Code is law.
You can order your copy from Georg éditeur.
Contribution from alp ict
Among the book's authors is our General Secretary Delphine Seitiée, who addresses on page 67 the gap between our daily AI practices and the low level of maturity of companies to implement advanced AI methods, which nevertheless open up new market and development opportunities.
AI can be used to extract meaning, determine better results and enable faster or even automated decision-making. Delphine emphasizes that these benefits are based first and foremost on the careful and extensive harvesting of data. Quality data is essential to create real opportunities.
Finally, the ethical, safety and environmental aspects must be central to the development of AI technologies.