The Los Angeles Times just took a step in the right direction by revealing the political bias of its opinion pieces in a new feature called Insights.
As a news organization, the LA Times has a consistent Lean Left bias, and AllSides 100% supports its efforts to become more balanced and transparent. With more balance, the LA Times can better serve Los Angeles, improve audience trust, and better support democracy.
But this small step comes with problems, as many first steps do. While we hope the LA Times continues on this path, a few changes would have made this program more successful. Here are the top areas where this first step fell short and can be improved upon in the future.
Disclosure: AllSides worked with the LA Times from November 2024 to February 2025, sharing our API and providing our experience and expertise, especially on article-level bias analysis using AI tools. Their “Insights” feature is not AllSides technology, but a more limited AI-only, article-level bias ratings system.
News bias is a bigger problem than opinion bias
The LA Times only applied its AI bias rating feature to its opinion pieces. While this is good, and we hope it will encourage the Times to include opinions from a wider diversity of perspectives than it has historically, the place where objectivity and balance are most important is in the news – how and what stories are reported.
AI is biased, and so are the humans who prompt and develop it
AI provides wonderful opportunities to improve the world, but we shouldn’t just hand news analysis over to AI and say “done.” Multiple studies indicate AI is biased to the left, increases bias in humans, and that its biases may pose a risk to democracy.
As expected, LA Times ratings are biased to the left, just as studies have indicated is generally true of AI. AllSides looked at 33 articles rated by the LA Times. Our analysis rated over a third of the 33 articles as being more left than the LA Times rated them.
By rating its left-leaning articles more toward the center, the LA Times makes itself appear to be more balanced than it actually is. This might lead some people to see this as a marketing gimmick rather than the transparency or balancing tool it is intended to be.
AI can be used to enhance or increase speed, but must not be used to replace human insight
Some expressed concern about the lack of human editorial oversight behind Insights, such as LA Times Guild vice chair Matt Hamilton, who reportedly said, “we don’t think this approach — AI-generated analysis unvetted by editorial staff — will do much to enhance trust in the media.”
Many critics pointed out how human oversight would have prevented a problematic Insights note regarding the KKK.
At AllSides, we are excited about how AI can help us make an impact, but our politically balanced Editorial Team and input from people across the political spectrum will always be the most important part of AllSides.
To ensure our tools that use AI, like the AllSides Bias Checker, are trustworthy, we always go to great lengths to ensure results reflect the average judgment of people across the political spectrum. We are extremely transparent about our multipartisan team, our biases, and our methodology, and will always clearly indicate any instance in which we used AI to create content.
The AI-enhanced AllSides Bias Checker was responsibly developed and carefully designed using the patented AllSides Media Bias rating method. It was tested, reviewed and calibrated by our balanced team continuously. This is how we intend to approach any use of AI, and we recommend others follow these same principles.
Bias ratings can fairly reflect society across the political spectrum
Bias is of course subjective – different people will rate news content differently based on their personal biases and experiences. To be trustworthy, bias ratings must incorporate an outside standard — something other than the biases of AI or insiders behind closed doors.
One way to do that: AllSides Media Bias Ratings™, reflecting the average judgment of all Americans. AllSides balances the input of experts and ordinary people across the political spectrum by using multi-partisan Editorial Reviews of trained experts and Blind Bias Surveys™ in which tens of thousands of participants across the country (any country — we do international work, too) rate content without knowing the source. This balanced approach ensures more trustworthy bias ratings than AI alone. While our AI-enhanced tools like our Bias Checker can help us rate more content more quickly, our tools are always designed to accurately and fairly reflect everyone across the political spectrum.
The goal: balanced, objective, trustworthy content
A bias rating alone is not the goal – balanced, trustworthy news content is. That means newsrooms ought to integrate trustworthy systems into the newsroom to improve their news output. AllSides clients like Newsweek, Spectrum News, and Straight Arrow News have done this with fantastic results, all earning Center or balanced ratings, and all succeeding with growing their audiences as a result (especially Newsweek).
To do anything reliable in business, you must measure results. Our news media clients use detailed AllSides Media Bias Audits™ to measure their bias in a very thorough way, receiving not only overall bias ratings of their news content, but data and information about how different segments of their audience react to their content, why, and steps the company can take to make improvements. They use and even integrate into their CMS a variety of balancing tools and resources, including the AllSides Bias Checker, which gives journalists real-time feedback to see unintended bias and make adjustments before publishing — like Grammarly for bias.
With trust in news at all-time lows, transparency and balance are critical. For the leaders at LA Times, please keep making positive, courageous steps. This is just step one in a long, and potentially quite rewarding, journey.
John Gable, CEO and Co-founder of AllSides Technologies