LLM-based chatbots are solving some hard problems of the world. While many companies are reaping great benefits from them, some are stuck with it, unable to make any move. For instance, Stack Overflow, a community-based discussion platform for developers, is in a great dilemma. It finds itself lodged in a strange place where it’s unable to completely adopt or shun LLMs.
Recently, sensing a loss in users owing to LLM-based chatbots like ChatGPT, Stack Overflow decided to finally ride the generative AI wave and create its own LLM. While this move was important to allow Stack Overflow to stay afloat in the strong generative AI wave, it has risked alienating the community, which makes up the heart of the platform.
After the move, there is considerable backlash from the developer community, leading Stack Overflow to garner somewhat of a negative reputation.
Can LLMs Save Stack Overflow?
What started with banning answers written by ChatGPT has now evolved into using StackOverflow’s considerable corpus of user-generated content to train an LLM. The LLM-based platform will reportedly be trained on the knowledge base of over 50 million questions from Stack Overflow, along with proprietary data from the Stack Overflow for Teams service. While they have mentioned that the users who contributed to the platform will be rewarded for doing so, there is no concrete reward system mentioned.
In a blog post written by Prasanth Chandrasekar, the CEO of Stack Overflow, the platform outlined its plan to adapt to the AI rush. Titled ‘Community is The Future of AI’, the blog speaks about how technology can change existing paradigms, leading into what it means for Stack Overflow. Prasanth stated, “At Stack Overflow, we’ve had to sit down and ask ourselves some hard questions. What role do we have in the software community when users can ask a chatbot for help as easily as they can another person?”
According to the blog, the company has a “dedicated team” working on adding generative AI to Stack Overflow and this comes at a time when every community-based platform is coming up with ways to integrate generative AI.
However, this GenAI twist is set to create a new problem, adding to the existing problems moderators are facing with ChatGPT-generated answers. Going by the tone of the discussions on Stack Exchange thread and Hacker News forum, it appears that the community is not receiving this announcement well.
Communities Don’t Like AI
One only needs to look at competitors like Quora and DeviantArt, both of which integrated generative AI. Quora had Poe, an AI platform that could answer questions quickly and easily, while DeviantArt tried to train a diffusion model called DreamUp. The difference is that Poe was not trained on the writings from the Quora platform directly, while DeviantArt simply scraped all the artwork on its platform without giving any credit to the artists.
As expected, Quora didn’t face any backlash over Poe, while DeviantArt was forced to change the terms of usage for DreamUp within a few days of launch. The main takeaway is that communities don’t like it when their data is used to create an LLM that will eventually make them obsolete, especially for a cornerstone of the developer community like Stack Overflow.
Cory Doctorow, the ex-co-editor of the blog Boing Boing, wrote an insightful piece on how community-based platforms die. In his words, “First, they are good to their users; then they abuse their users to make things better for their business customers; finally, they abuse those business customers to claw back all the value for themselves. Then, they die.”
It seems that Stack Overflow is currently in the second phase of this process, as seen by the response by the community. One user stated that unless Stack Overflow has something that Google and OpenAI don’t, creating an accurate LLM is ‘hopelessly optimistic’. Others have attacked the CEO’s understanding of the platform, pleading to not add generative AI to Stack Overflow. Another user called the move a slap in the face of all the real contributors to the site, and yet another has pointed the finger at the numerous problems Stack Overflow needs to solve first instead of dedicating resources to generative AI.
While there are many facets to the community’s backlash against this move, a statement by a contributor named ‘Andreas detests censorship’ sums up the sentiment. “We’re not interested in fact-checking AI content; we are interested in generating the content ourselves. I find this post very demotivating. I’m not sure if there’s any point in continuing to moderate and curate Stack Overflow.”
While there is still time for the platform to renege on their generative AI dreams, it seems that the damage is slowly being done. In pursuit of creating an LLM product that could be integrated into Stack Overflow for teams, it seems that the platform is alienating what made it great in the first place — its community.
Note: We reached out to Stack Overflow but did not receive a response till the time of publishing this article.