"We Hated Them": Could AI Revive Cleveland.com Comments After Toxicity?

The Rebirth of Online Discussions at Cleveland.com

Five years after Cleveland.com shut down its comment section due to persistent toxicity, the news site is taking a cautious step toward reviving online discussions. This time, the approach involves leveraging artificial intelligence (AI) as a potential solution to the challenges that once made the platform unbearable for many readers.

In a recent episode of the "Today in Ohio" podcast, Editor Chris Quinn revealed that he's reconsidering his long-standing stance against bringing back comments. His change of heart came after receiving feedback from 370 readers who expressed interest in AI moderation.

Quinn, who had previously vowed that comments would never return during his tenure, admitted on the podcast, “I’ve said it a lot because I hated them. It was horrible. And no matter what we did, people found workarounds to be horrible.” The decision to remove the comment section in February 2020 came after years of dealing with hate-filled, racist, and vicious statements that overshadowed meaningful conversations.

However, the rapid advancement of AI has given Quinn reason to be cautiously optimistic. He noted, “I am impressed with how effective AI is at reading into things. And while you can always defeat technology, I think this would be a much more effective way.” Quinn envisions a system where AI sets a high standard for content, flagging questionable posts for human review. He believes this process wouldn’t be too burdensome for moderators.

Leila Atassi, a reader opinion expert, highlighted that responses to the idea of AI moderation fell into three main categories. Some readers missed the back-and-forth of comments, appreciating the extra context they provided. Others were skeptical about AI’s ability to understand nuance or worried about potential censorship. A third group was firmly against bringing comments back altogether.

Atassi pointed out that one of the biggest challenges would be determining what constitutes "off-topic" content. For example, if a story focuses on Cleveland schools and someone brings up national politics, is that automatically out of bounds? She questioned, “Once you start deciding where the guardrails are, you’re in the business of defining what counts as legitimate speech. And that’s kind of a slippery job for both humans and machines.”

Despite these concerns, Quinn sees value in the experiment. He emphasized the need for a space where normal people can have conversations about news topics. “They’re being drowned out by the screamers, and could this do it?” he wondered.

The discussion raises broader questions for news organizations everywhere: How can they foster meaningful public dialogue in an era of extreme polarization and algorithmic amplification of outrage? While social media platforms like Facebook offer commenting capabilities, Quinn pointed out that these spaces remain largely unmoderated and “horrible.”

If implemented, Cleveland.com’s approach might require users to be either paying subscribers or registered users. Quinn also suggested that AI could provide feedback to users whose comments are rejected, explaining the specific reasons and allowing them to revise their submissions. This functionality wasn't possible with previous moderation systems. Possibly, the AI could even suggest rewrites of rejected comments that would pass muster.

For a news organization that experienced firsthand how toxic online comments can damage the reader experience, this cautious exploration represents a significant shift. As Atassi concluded, “It still feels like a low stakes experiment worth doing... But I think it’s worth trying.”

The conversation around AI moderation continues to evolve, offering new possibilities for engaging and respectful online discussions. Whether this experiment will succeed remains to be seen, but it marks an important step in the ongoing effort to create a healthier digital public sphere.