Training 300,000 Individuals on GenAI: A Guide

As the debate around generative AI (GenAI) continues to swirl, with some predicting a bubble about to burst and others seeing untapped potential, one company has decided to invest heavily in this technology. Infosys, an IT services enterprise, has taken the bold step of training its entire global staff of 340,000 people on GenAI. 
 
Rafee Tarafdar is Chief Technology Officer at Infosys and was instrumental in implementing this initiative. He recognized that amidst uncertainties around GenAI's long-term viability or usefulness across business sectors, it was crucial for his organization to stay ahead by embracing new technologies early. 
 
"We felt it necessary as part of our own AI-first transformation," says Tarafdar. "Regardless of their level within the company, everyone needs to be familiar with AI." 
 
Infosys began by assessing their existing skills landscape—identifying tasks that could potentially be automated through AI and determining what new skills would be needed in an increasingly AI-driven environment. 
 
Recognizing that not all employees would require the same level of understanding or interaction with GenAI tools led Infosys to adopt a three-tiered approach towards training: 'awareness', 'builders', and'masters'. The first tier aimed at making every employee aware of basic concepts related to generative AI; those who fell into the second category were trained on how they could utilize these tools effectively, such as creating products using generative models; finally, masters are expected to have deep knowledge about safety measures regarding prompt injections or hijacking while also being subject matter experts scrutinizing data sets for quality and utility. 
 
To deliver this varied training curriculum effectively without resorting to a one-size-fits-all methodology, which wouldn’t work here given diverse requirement levels amongst employees across different job roles, Infosys used its internal platform Lex. They created 66 courses tailored specifically according to each identified persona’s learning requirements, covering basics for beginners along with advanced modules suited to more experienced learners, including builders/master categories, respectively. 
 
"About 84% of our employees are now AI-aware. We have a large number that are builders and masters too," says Tarafdar. "We're midway through the transformation; AI-aware training is largely done, but there's more work to do for the builders and masters." 
 
Infosys also took measures to ensure compliance with regulatory standards such as the EU’s AI Act, which mandates against discriminatory impacts or biases in technology applications. A responsible AI framework was incorporated into their training program, covering explainability, ethical considerations, and security aspects related to generative artificial intelligence usage. 
 
In order to track progress within this initiative, Infosys collects metrics around daily users on its platforms along with acceptance rates for code created using GenAI tools. Employees are encouraged to not only flag issues they encounter while interacting with these systems but also correct them themselves whenever possible, thereby helping improve model accuracy over time by providing real-time feedback based on practical experiences during day-to-day operations. 
 
Despite concerns about job losses due to automation technologies like GenAI, Infosys CEO Salil Parekh has assured that no cuts were planned within his organization, attributing any perceived threats purely towards productivity enhancements rather than replacement human workers altogether—a sentiment shared by many other proponents who believe their adoption can bring about significant improvements in efficiency levels across various business processes if implemented correctly without falling prey to mere hype trends prevalent today’s fast-paced digital landscape. 
 
Tarafdar himself remains optimistic about GenAI's future: "The organizations that have taken a strategic approach will deliver value," he concludes confidently.