“Entering the new era of artificial intelligence“
Since the launch of ChatGPT(1) in November 2022, generative artificial intelligence (AI) has been integrated into our working methods. It marks the start of a new digital era, on a scale probably equivalent to that of the arrival of the internet. We often read that AI is revolutionising content creation. Above all, what makes it unique is that it allows us to talk to computer systems in natural language. With generative AI, all you need to do is ask a question to get an instant answer. As there is no longer a need to learn technical language, it breaks down the digital barrier, which in turn promotes inclusion. Its second key strength is its ability to reason about sophisticated fields of expertise, for example analysing and processing financial information.
As its use has accelerated around the world, the issue of how it should be managed and used responsibly has come to the fore. In this context, on 2 February 2024, the European Union adopted a regulation to frame its development. In the United States, the National Institute of Standards and Technology has proposed a framework for securing its deployment with a view to ISO/IEC 42001 certification, while the G7 has its Code of Conduct.
15,700
billion US dollars: this is AI’s expected contribution to the global economy by 2030.
Source: PwC, Global Artificial Intelligence study.
At Microsoft, we launched Copilot, an artificial intelligence assistant integrated into all our applications and experiences. Its name says it all: it assists the user, who has the choice of relying on what generative AI produces, while retaining the same critical mindset as usual. We also ensure that we inform and train users in order to foster informed use. Specifically, we must learn to ask generative AI questions, commonly called “prompts”, so as to avoid the risk of generating incomplete or even irrelevant answers.
Many safeguards exist at this time to manage potential risks, such as asking Copilot to explain its responses or providing the generative AI model with verified sources of information from which to draw on and cite. To oversee the responsible use of AI, we have put in place strict governance and a stringent execution framework. In the banking sector, generative AI provides a very wide range of possibilities. With this in mind, we are working with BNP Paribas’ business lines, IT and data teams on the implementation of generative AI solutions. The first stage of this collaboration involves business and process-oriented work on a secure cloud services environment, while the second stage focuses on user-oriented work on Copilot’s ready-to-use solutions for teams.
Microsoft’s responsible AI
In 2017, Microsoft created an advisory committee to advise its customers on the implementation of responsible artificial intelligence. The company also established six key principles for trustworthy AI - accountability, inclusion, reliability and safety, impartiality, transparency, confidentiality and security - with controls on AI systems.
To facilitate the implementation of these principles, Microsoft provides processes and tools which are integrated into its services, and regularly publishes transparency notes on the internet to make its AI technology easier to understand and to help users make the right choices.www.microsoft.com/en-us/ai/responsible-ai