Generative artificial intelligence (AI) tools will save companies a huge amount of time and money, says Vik Singh, a Microsoft vice president, even though these fledgling applications still have a lot to learn, such as admitting their shortcomings.
“What’s really missing from these models today, to be honest, is that they don’t raise their hand and say, ‘Hey, I’m not sure, I need help,'” he explains in an interview with AFP.
Over the past year, Microsoft, Google and their competitors have been rapidly deploying generative AI applications such as ChatGPT (OpenAI), which produce all sorts of content on demand and give users the illusion of being omniscient.
But despite progress, they sometimes continue to “hallucinate,” that is, to invent answers.
A major problem to solve, according to Vik Singh, in charge of Copilot applications at Microsoft. Because the customers of this AI assistant from the American IT giant – companies – cannot afford for it to make up, even occasionally.
“Very smart people are working on this problem (…) so that the models identify when they don’t know and ask for help,” he promises.
And a more humble model would be no less useful, according to him.
He takes the example of an AI assistant specializing in customer service. “I spoke to a company that was spending $8 per request, each time one of their representatives responded,” he says.
Even if in half the cases the machine had to call on humans to find the solution, the company “would still make huge savings.”
– Productivity –
Having joined Microsoft in January, Vik Singh took over this summer as head of the teams developing “Copilot” (AI assistants) specializing in sales, accounting and services.
Little known to the general public, these commercial applications allow technology groups to justify their massive investments in generative AI, because they are starting to generate revenue.
According to the Windows group, Copilot performs research for salespeople, freeing up time to call customers. Lumen, a telecom company, “saves about $50 million a year this way,” says Mr. Singh.
The start-ups behind the generative AI revolution promise systems so advanced that they will help “elevate humanity,” in the words of Sam Altman, the boss of OpenAI (mainly funded by Microsoft).
But for now, the new technology is mainly used to gain productivity and therefore profitability.
Vik Singh’s teams are working to integrate Copilot directly into the group’s software and make them more autonomous, so that users need to contact them less and less directly.
“For example, you are a sales representative,” suggests the manager.
Two weeks after a conversation with a customer, the AI assistant “can prompt you to follow up with the contact, or, better yet, automatically send an email on your behalf, because the task was approved in advance.”
– Employment –
Before finding a solution to global warming, AI is therefore supposed to rid humanity of boring and repetitive chores.
“This is only the beginning,” recalls Mr. Singh. “We are first focusing on the easiest applications” to implement.
Will all these productivity gains translate into job losses?
Big-company executives like K. Krithivasan, the head of Indian software giant TCS, have said that generative AI will all but eliminate call centers.
But Vik Singh — like many Silicon Valley executives — is counting on the technology to make humans more creative and even create new jobs.
He points to his experience at Yahoo.com in 2008, when a dozen editors were choosing stories for the home page.
“We had this idea of using AI to streamline that process, and people were like, ‘Oh my God, what’s going to happen to the employees?’” Singh says.
The automated system helped refresh content more quickly, increasing the number of clicks on links and the need for new stories. “Ultimately,” the executive says, “we had to hire more editors.”
