LA TRIBUNE SUNDAY — Some speak of a bubble around generative AI, which ended up disappointing certain companies. For what ?
EMILIE SIDIQIAN — There has been a bubble effect around so-called “co-pilot” generative AIs, such as those from Microsoft and Google, which produce text on demand. Many companies jumped into it without knowing what to do with it, without a strategy, and therefore without a planned return on investment. The co-pilots proved to be disappointing. ChatGPT can organize your vacation but it is less relevant on questions of strategy. Fortunately, we are arriving at a new phase of artificial intelligence, that of agents.
The agents will not be disappointed?
Agents don't just write emails, they suggest decisions and can take action. They sometimes take the form of virtual colleagues who communicate with us based on the data to which they have access. Salesforce is at the heart of this revolution because our platform already brings together a multitude of internal company data. Agents can also take over certain tasks. For example, when a customer calls after-sales service after 10 p.m., an AI agent can answer them. It adapts to the client's tone of voice and accesses their file. I created my own agent myself. Connected to my Slack conversations and my knowledge base, it can respond to customer requests for information.
Generative AI in business, instructions for use
How can we protect ourselves from possible errors by these agents?
The agent doesn't appear out of nowhere. It is the company that decides to create it and predefine its functionalities. In addition, a supervision console allows you to control your actions and gives the reliability rate of a task being carried out. Ours is 95%. When an action is not identified as reliable, an alert allows humans to regain control.
Several social plans have been motivated by AI. How do you respond to this growing risk?
Economic performance cannot be achieved without a social contract. Mass dismissals harm attractiveness, hence the need to know how to modernize while training employees and managers. Training must start at school, with compulsory courses on AI. We also need more history and philosophy courses because it is thanks to these subjects that we will maintain our free will in the face of technologies.
Don't you fear a revolt among employees in the face of this rapid evolution of work?
We announced the creation of a billion agents by the end of 2025. It's true that it's fast, but we have to be pragmatic and help humans work with them. Our first clients did not reduce their workforce, on the contrary they increased the number of tasks processed per employee. They have also improved the quality of their service and acquired market share
Salesforce gets its hands on four AI gems for $2 billion
What will our work life look like in 2030?
Using AI is a matter of life and death for a business. There will be a turnover of talent, but no, not all jobs will be eliminated. White-collar workers will need to adapt to generative AI, and blue-collar workers will experience new advancements thanks to robotics. Our interactions with machines will be simpler. In my opinion, everything will be done by voice and we will no longer have all kinds of applications to manage. The freed up time can be devoted to reflection, creativity and innovation. Technology will remain at the service of humans.
“AI has completely transformed the way our teams work” (Jean-Charles Samuelian, CEO of Alan)
