MLS Used AI To Write Match Recaps and It Went as Badly as You’d Expect

Oct 19, 2024; Fort Lauderdale, Florida, USA; Inter Miami CF forward Lionel Messi (10) celebrates scoring during the second half against the New England Revolution at Chase Stadium. Mandatory Credit: Sam Navarro-Imagn Images

Never underestimate Major League Soccer’s ability to use technology stupidly.

It took two full years partnering with Apple TV to realize it should harness the relationship to provide viewers with more matches to watch, not fewer.

And while MLS was an early adopter of VAR, it has still dragged its feet on using virtual offside lines, semi-automated offsides systems or goal-line technology.

Then came this weekend, as the league website began publishing match recaps written by Artificial Intelligence without any editorial oversight. It went about as poorly as you'd expect.

Full disclosure: I worked as a freelance contributor for the site, mlssoccer.com, for the better part of a decade until 2022.

There was predictable (and justified) outrage among the league’s underserved fans and undercompensated human journalists. In particular, longtime FC Cincinnati beat writer Laurel Pfahler correctly pointed out how far short MLS media access is relative to other leagues, and how asinine that practice is when MLS significantly lags those in media coverage.

But equally alarming was the quality — using that term very loosely — of the writing.

Aside from correct grammar, this was fail-your-100-level-journalism-course stuff. And the comically bizarre decision to publish it to your league’s own website only amplifies one of MLS critics’ strongest criticisms: The average MLS game doesn't matter.

Everything included in each recap could be ascertained more quickly by looking at the match summary or match feed also included on the website. Absent was all of the context that would set a good narrative news story apart.

For example, Big Brother apparently forgot that Messi’s brace for Miami put him back on top of the MLS Golden Boot leaderboard. He rambled for 177 of his 205 words before remembering that Thomas Muller scored the match-winning goal for Vancouver in second-half stoppage time. He would prefer not talk about the playoff race or the standings a week from the end of the regular season, thank you.

At best, these stories are published to exist, not to be read; to scam fans and clubs into believing more coverage was devoted to their team, while trusting they won’t look at more than a headline.

The data supports that conclusion; the MLS website is just one of many outlets to direct its humans away from writing recaps and focus on more responsive stories based on the biggest news, quote or other controversy that emerges from a game. That's what readers respond to now. They've already seen the video highlights.

But is that worth the message it sends to casuals?

The Athletic's Tom Bogert correctly pointed out the league is not alone in utilizing AI, citing ESPN, the U.S. Open, MLB, the NBA and others. That’s largely true. Yet in most of those cases, AI covers entities with less of a following — think unseeded players, mid-Major colleges, minor-league prospects and so on.

Someone in MLS decided to use the computers to help tell the story of the world's greatest living player and two of the league's best teams. If the league's own humans can't be bothered to care about those, why should any others?