Over the past couple of years, I’ve noticed a shift in behavior among the people I work with in India. Bolder career moves, pivots to free-lancing, open discussions about burnout as well as more confidence in navigating workplace dynamics. All of this would have been unusual a few years ago.
A recent survey showed that around 50% of young people in India use ChatGPT and similar chatbots when feeling lonely, anxious, or seeking advice. Anecdotally, I know that many people I work with use AI tools for similar reasons.
While I appreciate people being more assertive and mindful about their career choices, this change in behavior is making management challenging. Our managers and HR teams are finding that their skills and tools map to a culture that is now changing very quickly.
LLMs are trained on content from the Internet.
The advice that comes from chatbots reflects American workplace norms and American therapeutic language because most of these tools are made by and for Americans.
When a 20-something software engineer in India asks ChatGPT for help with a difficult supervisor or to navigate their annual performance review, the response is shaped by cultural context thousands of miles away. Is that advice going to be useful? Is it going to be productive?
Consider the manager on the other side. A 40-something who relied on peers, supervisor, and family for career advice and counseling. That advice was rooted in societal norms and loaded with cultural context. Now they are supporting a cohort of engineers whose behavior is being shaped by a completely different culture. Are they equipped to support their team effectively?
The advantages of AI in coding and knowledge work are clear. But there are second-order social consequences that deserve more attention. If we are to mandate the use of AI for work, we must also support managers as they deal with the consequences.
A generation of engineers raised on AI is entering their workforce. Their frames of reference are not coming from the world that they work in.
We speed-ran a version of this experiment with social media twenty years ago. The results have been mixed, at best. I am worried we are doing it again, much faster this time.