Stack Overflow confirms layoffs affecting 28% of workforce

Stack Overflow, the 15-year-old web institution for programmers that is noted for its forums where thorny coding questions are answered by the website’s community, is laying off 28% of its workforce, CEO Prashanth Chandrasekar announced today.

The company did not specify exactly how much of its headcount would be impacted, but its LinkedIn page indicates it has between 501-1,000 employees, and 769 on LinkedIn, so presumably around 215 persons.

In a short blog post, Chandrasekar said that “this year we took many steps to spend less,” without elaborating on what those steps were, but that “Unfortunately, those changes were not enough.”

Chandrasekar also cited “ongoing threats to customer budgets shifting due to the macroeconomic pressures impacting the entire tech industry. ”

What impact did ChatGPT and Github Copilot have?

While Chandrasekar, nor a spokesperson reached by VentureBeat, elaborated on what other pressures might be facing the company, observers on X (formerly Twitter) were quick to point the finger to OpenAI’s ChatGPT, the chatbot based on its large language model (LLM) GPT-3.5 and GPT-4/V.

OpenAI also makes Codex, a programming specific LLM based on GPT but trained on code snippets and designed to help software developers generate and complete code.

Laura Wendel, a startup founder between London and Germany, posted that “This may be the first large layoff directly due to AI,” and cited “people asking ChatGPT instead of Stack Overflow” for answers to their questions, posting a graph showing a precipitous decline in Stack Overflow page views beginning in April 2023, when the chatbot was gaining public momentum.

Earlier in the summer, in June 2023, Google DeepMind researcher Mahesh Sathiamoorthy published a similar graph on X also showing a collapse in Stack Overflow traffic of about 50%.

At the time, fellow Googler François Chollet opined on Sathiamoorthy’s post that Microsfot’s “GitHub Copilot and LLM chatbots are causing a big drop in StackOverflow traffic. What’s fascinating about it is that these systems work largely thanks to the user-generated content from StackOverflow. Is this feedback loop sustainable?”

The big question: how is Overflow AI doing?

In July, just a month after this post and exchange, Stack Overflow announced its own developer tool LLM chatbot product, OverflowAI, that is trained on the corpus of the Stack Overflow public knowledge base.

In Chandrasekar’s blog post, he praised his teams’ “monumental efforts to successfully launch OverflowAI.”

However, the company has not shared much in the way of metrics about how that solution is doing in terms of uptake. But with competing solutions out there and already enjoying immense usage, it is clear that the generative AI age has not been an easy one for Stack Overflow so far.