The AI trend was kind of dull. But now, things are changing, and they're changing fast.
Thanks to the generative AI trend, which was kickstarted by OpenAI's ChatGPT, others have jumped into the bandwagon to benefit the technology. And among those, include Stack Overflow.
Considered "the largest, most trusted online community for developers to learn, share their programming knowledge, and build their careers," it was one of the most popular places for developers to gather and brainstorm. It was the community many programmers rely on, and can depend on.
But following the hype of generative AIs, an increasing number of developers and programmers are no longer in needs of such community.
This is because with generative AI products, they can just ask the AIs for help, including aiding them in their coding work.
As a result, Stack Overflow becomes no longer relevant.
But that doesn't mean it cannot put up a fight.
Not just any "fight," because it knows it cannot fight against the mainstream trend. So instead, it's "fighting" by embracing the technology.
To survive in the world where free online data is generated by AIs, the platform is introducing what it calls the 'OverflowAI'.
Announced by Stack Overflow CEO Prashanth Chandrasekar, it's essentially an AI model, trained not only on its public data but on masses of proprietary information.
For all its time, the company's data is mostly arranged in a Q&A format, where coders ask questions online and getting answers from other users, that are voted on by other members of the community. With around 60 million questions and answers, Stack Overflow can certainly boasts a lot of data for its AI train on.
To create OverflowAI, the company fed it with data that includes Stack Overflow's huge list of questions that are already preserved as prompts. The AI model is then trained to make associative connections between the questions and the answers, in order to be able to provide statistical connections between the answers.
Stack Overflow began introducing Overflow AI on July 27, when it started allowing users to get answers for their coding questions from the generative AI.
The system, which was introduced initially as a test version, is designed to create an instant summarized answer to users' queries.
When ChatGPT was first introduced, and soon after a number of popular generative AIs became capable of helping people code, Overflow's CEO, Prashanth Chandrasekar, spotted a worrying trend.
Online traffic to the Q&A website for software coders had begun to slip. In April, traffic was down about 13% from 2022, the company's data showed.
Stack Overflow has been the online community where developers, programmers and everyone in between that work as engineers, to to get tips and suggestions from other fellow coders.
With ChatGPT, Google Bard, Codex, or even GitHub Copilot. those people can get their answers, fast and easy, and free or change.
There's less reason visit Stack Overflow.
Making things worse, the creators of those AIs have been scouting the web for train the AIs, and Stack Overflow has indeed been one of their sources.
"Some of them are very explicit about calling out Stack Overflow as a primary source," said Chandrasekar.
If Stack Overflow does nothing, it can translate to a death by Large Language Models.
According to Chandrasekar:
"People who are leveraging our data for LLM purposes, we took a position several months ago that they should engage with us."
"We should be able to be paid for that data. The large companies have proactively reached out to us, and we're effectively engaged in those conversations at the moment."
Chandrasekar's strategy is to go with the flow, and go with the market's demand.
In other words, if you can't beat them, join them.
Nat Friedman, a former CEO of GitHub through 2021, expects tech companies to pay for training data in the future.
When StackOverflow is fully dead (due to long congenital illness, self-inflicted wounds, and the finishing blow from AI), where will AI labs get their training data?
They can just buy it! Assuming 10k quality answers per week, at $250/answer, that's just $130M/yr. Even at…— Nat Friedman (@natfriedman) July 26, 2023
With OverflowAI, Chandrasekar hopes that it can help newbie users in getting the answers they need.
At Stack Overflow, when less-experienced users asked basic questions that the Stack Overflow community had already solved, the community can be rude, and this can make some people nervous about engaging on the platform.
But not with OverflowAI, because the generative AI is not designed to make fun of people.
Chandrasekar knows that OverflowAI cannot guarantee people to return because after all, people can get answers without ever visiting Stack Overflow or use Overflow AI.
There are way too many AI products that can do the same.
To deal with this bigger challenge, Stack Overflow's survival method is to ensure that Stack Overflow is the place where the high-quality data resides.
"That's a very real thing, which is why you absolutely need really solid high-quality sources of truth like Stack Overflow forever," Chandrasekar said. "If you don't have that, then you're going to run into that situation."
This is why Stack Overflow is planning to charge tech companies that wish to use its data to train their AI, and reinvest the money to create a new "mechanisms" to incentivize human coding experts to continue answering questions.
"We're working on the incentive mechanics of how to make sure those folks get credit, even when a generative-AI answer is delivered to, let's say, a novice programmer who was able to solve their immediate problem," Chandrasekar said. "We are thinking through the details of that incentive system at the moment as we test this with our users over the next couple of months."