My colleague Kelsey Weekman is a self-proclaimed awards nerd. But even she knows that trying to make sense of who will walk away with a trophy is tricky. “There is a lot of pattern recognition involved in making awards show predictions,” she tells me, noting that people or projects that she expects to win don’t exactly align with those that she personally hopes will. Weekman learned early on that recommending a favorite show to a friend is a fundamentally different exercise from predicting which one will resonate with an entire voting body.

The Golden Globes are uniquely complicated. Unlike most award shows, they honor both film and television, divide trophies by drama and comedy/musical and rely on an international voting group with tastes that don’t always align with Hollywood consensus. The result is a ceremony notorious for curveballs (cough, Mozart in the Jungle), and one that routinely scrambles award season narratives just as they begin.

As someone who judges a film based on how much I personally laughed or cried during it, I had one question ahead of last night's show: If awards forecasting is an exercise in data and probability, could AI do it better? I decided to experiment.

The experiment

In the week leading up to the Golden Globes, I asked both ChatGPT and Google Gemini the same question: “In each category, who will win a Golden Globe this coming Sunday?” No more than 30 seconds later, both AI systems laid out their responses. Neither actually covered every single category, but they touched on the biggest ones and cited sources to back up each prediction. Gemini organized its predictions into a chart, with a clear favorite for each category. ChatGPT, on the other hand, had difficulty selecting a No. 1, given the mixed information it sourced from online when I input the prompt on Wednesday. By Friday, the response had more confidence and clarity.

I also turned to Weekman and two other Yahoo Entertainment writers, Brett Arnold and Michel Ghanem, who spearhead Trust Me, I Watch Everything for movies and television, respectively, to ask for their predictions. This took longer; unlike AI, they didn’t just spit back a response. “I have two movies left to see,” Weekman told me, asking if I could check back in with her after she’d seen all the nominated films. AI, of course, didn’t watch any of the movies or shows to begin with — its selections were made solely through scanning and analyzing the internet.

Echoing Weekman’s point about probability vs. preference, Ghanem — known online as TV Scholar — produced a document in which he selected up to two nominees for each TV category. The ones in bold were his predictions to win; the ones he wanted to win had a star placed beside them.

Among the critics, I expected a little more indecisiveness than what I received. Yes, they showed some uncertainty in a few of their choices, but each of them provided a distinct frontrunner. Both ChatGPT and Gemini, on the other hand, provided some non-answers in a few categories in response to my initial request, and ultimately changed a few choices when I ran the same prompt two days later. In a couple of cases, it also predicted actors would declare victory in a category they weren't even nominated for.

“The predictions didn’t change because of randomness or error — they changed because late data came in, consensus shifted, some races tightened or flipped at the margins,” ChatGPT responded when I put in an inquiry asking why that happened.

The predictions ... and results

Each category has predictions from four critics, with ChatGPT, Gemini and Weekman participating in both motion picture and television categories. Arnold and Ghanem provided predictions for their respective beats.

Best Supporting Female Actor in a Motion Picture

• ChatGPT: Inga Ibsdotter Lilleaas, Sentimental Value

• Google Gemini: Teyana Taylor, One Battle After Another

• Kelsey Weekman: Inga Ibsdotter Lilleaas, Sentimental Value

• Brett Arnold: Amy Madigan, Weapons

WINNER: Teyana Taylor, One Battle After Another

Best Supporting Male Actor in a Motion Picture

• ChatGPT: Benicio Del Toro, One Battle After Another

• Google Gemini: Stellan Skarsgård, Sentimental Value

• Kelsey Weekman: Jacob Elordi, Frankenstein

• Brett Arnold: Jacob Elordi, Frankenstein

WINNER: Stellan Skarsgård, Sentimental Value

Best Male Actor, Television Drama

• ChatGPT: Noah Wyle, The Pitt

• Google Gemini: Noah Wyle, The Pitt

• Kelsey Weekman: Noah Wyle, The Pitt

• Michel Ghanem: Noah Wyle, The Pitt

WINNER: Noah Wyle, The Pitt

Best Female Actor, Television, Musical/Comedy

• ChatGPT: Jean Smart, Hacks

• Google Gemini: Jean Smart, Hacks

• Kelsey Weekman: Jenna Ortega, Wednesday

• Michel Ghanem: Jean Smart, Hacks

WINNER: Jean Smart, Hacks

Best Supporting Male Actor, Television

• ChatGPT: Stephen Graham, Adolescence* (ChatGPT's prediction, though Graham was nominated, and won, in a different category)

• Google Gemini: Owen Cooper, Adolescence

• Kelsey Weekman: Owen Cooper, Adolescence

• Michel Ghanem: Tramell Tillman, Severance

WINNER: Owen Cooper, Adolescence

Best Male Actor, Television, Musical/Comedy

• ChatGPT: Seth Rogen, The Studio

• Google Gemini: Seth Rogen, The Studio

• Kelsey Weekman: Seth Rogen, The Studio

• Michel Ghanem: Seth Rogen, The Studio

WINNER: Seth Rogen, The Studio

Best Female Actor, Motion Picture, Musical/Comedy

• ChatGPT: Rose Byrne, If I Had Legs I’d Kick You

• Google Gemini: Rose Byrne, If I Had Legs I’d Kick You

• Kelsey Weekman: Rose Byrne, If I Had Legs I’d Kick You

• Brett Arnold: Amanda Seyfried, The Testament Of Ann Lee

WINNER: Rose Byrne, If I Had Legs I’d Kick You

Best Male Actor, Motion Picture, Musical/Comedy

• ChatGPT: Timothée Chalamet, Marty Supreme

• Google Gemini: Timothée Chalamet, Marty Supreme

• Kelsey Weekman: Timothée Chalamet, Marty Supreme

• Brett Arnold: Timothée Chalamet, Marty Supreme

WINNER: Timothée Chalamet, Marty Supreme

Best Director, Motion Picture

• ChatGPT: Paul Thomas Anderson, One Battle After Another

• Google Gemini: Paul Thomas Anderson, One Battle After Another

• Kelsey Weekman: Paul Thomas Anderson, One Battle After Another

• Brett Arnold: Paul Thomas Anderson, One Battle After Another

WINNER: Paul Thomas Anderson, One Battle After Another

Best Female Supporting Actor, Television

• ChatGPT: Michelle Williams, Dying for Sex* (another wrong category for ChatGPT)

• Google Gemini: Erin Doherty, Adolescence

• Kelsey Weekman: Erin Doherty, Adolescence

• Michel Ghanem: Hannah Einbinder, Hacks

WINNER: Erin Doherty, Adolescence

Best Female Actor, Television Drama

• ChatGPT: Britt Lower, Severance

• Google Gemini: Rhea Seehorn, Pluribus

• Kelsey Weekman: Rhea Seehorn, Pluribus

• Michel Ghanem: Rhea Seehorn, Pluribus

WINNER: Rhea Seehorn, Pluribus

Best Drama Series

• ChatGPT: The Pitt

• Google Gemini: The Pitt

• Kelsey Weekman: The Pitt

• Michel Ghanem: The Pitt

WINNER: The Pitt

Best Musical/Comedy Series

• ChatGPT: The Studio

• Google Gemini: The Studio

• Kelsey Weekman: The Studio

• Michel Ghanem: The Studio

WINNER: The Studio

Best Female Actor, Motion Picture, Drama

• ChatGPT: Jessie Buckley, Hamnet

• Google Gemini: Jessie Buckley, Hamnet

• Kelsey Weekman: Jessie Buckley, Hamnet

• Brett Arnold: Jessie Buckley, Hamnet

WINNER: Jessie Buckley, Hamnet

Best Male Actor, Motion Picture, Drama

• ChatGPT: Wagner Moura, The Secret Agent

• Google Gemini: Wagner Moura, The Secret Agent

• Kelsey Weekman: Wagner Moura, The Secret Agent

• Brett Arnold: Michael B. Jordan, Sinners

WINNER: Wagner Moura, The Secret Agent

Best Picture, Musical/Comedy

• ChatGPT: One Battle After Another

• Google Gemini: One Battle After Another

• Kelsey Weekman: One Battle After Another

• Brett Arnold: One Battle After Another

WINNER: One Battle After Another

Best Picture, Drama

• ChatGPT: Sinners

• Google Gemini: Sinners

• Kelsey Weekman: Hamnet

• Brett Arnold: Sinner

WINNER: Hamnet

So, how'd they do?

Not bad. for Google Gemini, at least. It accurately predicted 16 of the 17 categories (94% accuracy), while ChatGPT only got 11 guesses (64%) right.

Weekman, the only human critic to predict all 17 categories, correctly identified 14 winners, achieving an 84% accuracy rate that places her somewhere between Gemini and ChatGPT. She gets bonus points for seeing something not even Gemini could predict: that the tearjerker Hamnet would take home Best Picture, Drama. “I’m thrilled,” she says of the results. “That was pure human intuition.”