There's a lot to say about AI's integration into journalism—and if you're only asking me, almost all of it would be extremely negative. Well, an engaged reader wouldn't just take my word for it, instead seeking to read widely on the subject to better inform their own opinion. My one request, if I may be so bold, is that your wider reading doesn't begin and end with asking an AI chatbot how it 'feels' about the subject.
In recent weeks, the LA Times has begun book-ending its opinion pieces with AI-generated 'Insights' (via AP News). Clicking on this dropdown tab offers information such as the supposed political alignment of the piece you've just read, a bullet point summary of the piece itself, as well as points offering "different views on the topic." It's a 'both sides' approach by way of Perplexity-powered AI.
Insights was first implemented on March 3, so it remains tricky to get a solid sense of the quality of this still very recent addition. Still, it's notable that some of its features, such as identifying the supposed political alignment of a piece, has so far only been applied to opinion pieces and not news.
At the very least, I appreciate the AI-generated bullet points offer some linked-out citations so you can dig deeper into its claims yourself. Mind you, that's a very journalist thing to say; how many people will investigate the quality of the included citations beyond noting the AI presents them at all? There's also arguably more to this story than mere AI bandwagon-hopping.
First, a brief recap: AP News notes the LA Times was bought back in 2018 by Patrick Soon-Shiong, a transplant surgeon, medical researcher, and investor who has also served as the publication's executive chairman for the last seven years. Interviewed by Fox News last year, Soon-Shiong said, "We've conflated news and opinion," later adding that the LA Times wants "voices from all sides," before going on to say, "If you just have the one side, it’s just going to be an echo chamber."
As such, opinion pieces are very clearly demarcated from news, often labelled as 'editorial' or 'Voices'. Beyond that, the publication also chose not to endorse a specific presidential candidate last year—about two weeks prior to election day—despite an editorial in favour of Kamala Harris being allegedly already prepared. The Los Angeles Times's editorial editor, among other members of the editorial board, resigned in response to this decision. To put it another way, since at least last year, there appears to have been a greater push from upon high to steer the publication more centrally in the name of impartiality.
Right so, with that context in mind, let's take a peek at the LA Times' AI-generated 'Insights' in action. In this opinion piece touching upon recent ICE detainments and deportations, Matt K. Lewis claims, "The point was never really about deporting violent criminals. The point was a warning to anyone who wants to come to America: Don’t come here. Or, if you’re already here, get out."
In response, Insights offers, "Supporters defend enhanced immigration enforcement as necessary to address a declared 'invasion' at the southern border," and "Restricting birthright citizenship and refugee admissions is framed as correcting alleged exploitation of immigration loopholes, with proponents arguing these steps protect American workers and resources."
While the opinion piece's stance is very clear, the AI-generated so-called-insight is comparatively mealy-mouthed, with phrasing like "a declared 'invasion' at the southern border," leaving far too much unchallenged. While it would be far from ideal to descend into a rabbit-warren of AI-versus-human counter arguments, it feels very odd to allow AI the last word. What's most frustrating is the implied assertion that the AI's regurgitated claims are at all equally valid views to be presented alongside the opinion writer's stated, well-sourced horror at ICE's overreach.
As such, I fear Insights may be yet one more far from neutral, bias-reproducing AI, rather than a worthwhile tool that offers valuable context to readers. Insights notes, "The Los Angeles Times editorial staff does not create or edit [this] content," so you can be sure that no pesky journalists were allowed to do their job and give the AI a stern talking to about uncritically repeating hearsay. Naturally, it would be ridiculous to hold the AI accountable for the decisions of the humans steering the ship—I just hope the course correction is swift.
