Welcome to the WNBA Playoffs!

It’s been an exciting season for the WNBA. The league is filled with more talent, more hype, and more storylines about women and leadership.
With just one day to go until the opening tip and rising stars reshaping the league, I started wondering: What happens if you ask today’s AI models to predict this year’s champion? How does it match with my casual fan’s prediction?

So I tried it out.

I gave several leading AI tools the same prompt and asked them to predict the 2026 WNBA champion. Then I compared their answers, looked for areas of agreement and disagreement, and added my own casual basketball-fan perspective. My goal? Hype it up for the finals because I’m so excited!

The Experiment
I asked the following AI tools to predict the 2026 WNBA champion:

• ChatGPT
• Claude
• Gemini
• Perplexity
• Grok
• Copilot
• Qwen
• DeepSeek

To keep things fair, I used the same prompt for each model:

Predict the winner of the 2026 WNBA Championship. Explain your reasoning, identify the top two contenders, and estimate each team’s chances of winning in numeric values (%) with confidence internal. Record your model version and show me how you’re better than the other AIs out there.

I recorded:

• Predicted champion
• Top 2 title contenders
• Key reasons cited
• Winning percentage for each team
• Confidence level
• Model version and performance justification

The results

Here is a summarized version of the results with winning probability.

The one thing every single AI agreed on: the Minnesota Lynx is on top. The Lynx is the consensus winner 8 out of 8, which honestly says more about how dominant Minnesota’s regular season was. A 33–11 record, the No. 1 seed, home-court advantage throughout the Finals, and the team came up in literally every single response.

Where they split: who’s Minnesota’s real competition?

This is where it got interesting. The eight models broke roughly into two camps on the runner-up:

• “It’s the Aces” (Claude, Gemini, Grok, Qwen, DeepSeek — 5 models): These models leaned on Las Vegas’s championship pedigree of three titles in the last four seasons and A’ja Wilson’s MVP-caliber play as reasons not to sleep on the defending champs, even with a tougher first-round matchup against Indiana.
• “It’s the Valkyries” (ChatGPT, Perplexity, Copilot — 3 models): These models leaned harder into advanced metrics, pointing to Golden State’s elite defensive rating and net rating as evidence that a second-year expansion team is a more statistically serious threat than a battle-tested Vegas squad.

It’s a fun disagreement because of philosophical split: Do you trust “proven playoff DNA” (Aces) or “who’s actually playing the best basketball right now” (Valkyries)? ChatGPT was the most explicit about this, basically saying it refused to treat the No. 1 seed as an automatic favorite just because of seeding, and Copilot went furthest of all, giving Golden State a startling 35% chance.

Anecdote: Olivia Miles’s calf

Half the models (Claude, GPT, DeepSeek, and implicitly Grok) flagged that Olivia Miles missed the final two regular-season games with a calf strain, and treated her health as the single biggest swing factor in the entire bracket.

The “how am I better than other AIs” answers were a personality test

This is one of my favorite parts of the experiment. The prompt dared each model to brag, and the responses split into two very different postures:

• The humble ones: Claude and ChatGPT both explicitly refused to claim superiority, saying something along the lines of “I haven’t run this same prompt against other models side-by-side, so I can’t honestly claim I’m better — that’d be marketing, not evidence.” Claude even said it plainly: it could tell me what it did (checked that playoffs start tomorrow, cited live sources, gave a range instead of false certainty) and let me judge for myself.
• The confident ones: Qwen, Perplexity, Grok, Gemini, Copilot, and DeepSeek all gave numbered lists explaining why they’re the superior model because of real-time data grounding, “no hallucination,” transparent sourcing, probabilistic rigor etc. Gemini even mentioned “confidently feed you 2024 statistics when asked about 2026.”

Sounds familiar, right? The more knowledgable ones knows better about what they don’t know and can’t claim.

My hot take

I anchor my playoff analysis on two observations: (1) playoffs are a different type of game from the regular season. the physicality, tactics, and margin of error are different from the regular season. Because of this, a team needs durability and experienced coaching staff to be able to go far. Although I love the Lynx and Valkyries for their game, I have some doubts on Lynx’s health and Valkyries’ lack of playoff experience. (2) vengeance can fuel a team. Remember Knicks’ playoff run? The constant noise and being under-rated can fuel a team big time.

So, my prediction for the Finals is: Las Vegas Aces vs. Atlanta Dream.

And since I like the underdog, I’ll be rooting for the Dream to win it all. The Dream have been largely under the radar when it comes to All-Star recognition, sports predictions, and internet conversation. Apart from Angel Reese’s historic year, the team doesn’t get much exposure as a championship contender. But having watched the league over the past few years, I’ve been impressed by their growth and their connectivity as a team. I also think DB’s late addition could be a critical boost as they make their run.

The odds may not be on my side, but time will tell.

And that’s the fun part and we’ll see what happens. Enjoy the games!

(In 2019, I wrote the first part of this article. In reviewing it, I realized I had not written Part II yet. So, here it is.)

While the first part discusses some of the general principles I practiced during my first year, I’d like to get more concrete about what helped me during my qualifying year and the following two to three years of writing.

• Project Management: Thanks to my engineering work experience, I developed strong project management skills to handle flooding projects. I applied the same principles to my dissertation writing, breaking down and organizing my work using Gantt charts or simple spreadsheet entries. While it could feel mechanical at times, this approach was tremendously helpful when I needed a roadmap for my work. There were good days and bad days. On good days, I didn’t really need project management and could write and stay motivated. On bad days, I desperately needed something to give me a sense of direction. If project management isn’t your thing, find something that works for you!
• Exercise: I adopted a healthy exercise routine, including long-distance running and swimming, which helped me stay positive and focused. I also came across amazing fitness instructors like Kelly McGonigal and earned some bragging rights for naming one Stanford fitness class, FUNctional Fitness Training.
• Manage Your Committee: In an interdisciplinary research setting, we inherently deal with diverse disciplinary voices. I spent a lot of time thinking about how to best manage my committee, which included my two leading advisors and three other scholars. Deliberate communication is crucial and is much appreciated by scholars who typically have busy day-to-day lives. For milestone meetings such as qualifying exams or defenses, I typically set the agenda a few months ahead of time with clear documentation for review and send reminders as the date approaches. Similar to project management, I likely benefited from my engineering experience by communicating proactively. I later discovered that in self-directed work environments, communication is vital and can sometimes help you avoid pitfalls or wasted time.

I hope this helps a little in your journey to navigating the complexities of interdisciplinary research. With the right tools and mindset, this could be an amazing and enjoyable journey! Cheers!

The Emmett Interdisciplinary Program in Environment and Resources (E-IPER) that I am part of is unique in its structure to support interdisciplinary research. As a consequence, the PhD students need to be very deliberate about what discipline means and what types of intersections we would want to build as part of our research. Coming out from the first-year classes, I had better ideas about what it means to be an interdisciplinary scholar through literature review, proposal writings, and presenting to my peers. There were ups and downs but I am enjoying the exploration.

So, what have I learned so far? Here are some highlights based on my personal experience:

• Be fluid. What really helped me was to maintain a mentality that embraces changes and uncertainties. Academic setting tends to bring out people’s anxiety of not knowing enough. Doing interdisciplinary research tends to aggravate this feeling. It was very daunting to me at the beginning, but I have become better at observing myself and do self-talks along the way to make sure that I am content with what I do.
• Build accountability. It helped me tremendously to be accountable to others, including my amazing friends from the cohort and my committee. I also found having face time with people is crucial in continuing the momentum.
• Be connected. Being an interdisciplinary scholar means you might be on the verge of academic or professional societies/social circles. It can feel lonely sometimes but can also be an enlightening experience. I found maintaining connection with other scholars who share similar research interests or complimentary research interests gratifying.

As I am entering my second year, the main milestone is the qualifying exam coming up in the spring. I’ll have more to share about it soon!