A Real Test of AI: ChatGPT vs Perplexity vs Claude in an NFL Football Pool

In the grand tradition of sports betting degenerates who suddenly become statistical analysts when money's on the line, I decided to conduct a little experiment this NFL season. Rather than simply throwing my hard-earned cash into the football pool void based on my own questionable judgment, I thought: why not drag artificial intelligence down with me?

The setup was elegantly simple. I joined an NFL pool—one of those weekly affairs where you pick winners for every game from Thursday through Monday night, with the Monday game serving as a tiebreaker based on total points scored. The beauty of pools like these is that everyone becomes an instant expert, armed with half-remembered statistics and an unshakeable confidence that this is the year they crack the code.

But here's where it gets interesting. Instead of just trusting my own football wisdom (accumulated through years of Fantasy Football mediocrity and an unhealthy obsession with NFL RedZone), I decided to pit the three leading AI platforms against each other—and against me. Same prompt, same resources, same Week 1 slate. Let the artificial minds duke it out.

The Experiment

I fed identical prompts to ChatGPT, Perplexity, and Claude, providing them with the full NFL schedule, links to top football analysis sites like The 33rd Team and Football Outsiders, and current Vegas odds. The instructions were clear: don't just follow the betting lines like sheep. Add some nuance, take calculated risks, show some personality in your picks.

What emerged was fascinating, revealing the distinct personalities that these AI systems have developed—and one glaring weakness that surprised me.

The Great Schedule Debacle

First, the embarrassing confession: Perplexity, the AI platform built specifically for research and known for its web-scraping prowess, completely face-planted on the basic task of getting the Week 1 schedule right.

Not once. Not twice. Four separate attempts before it could accurately tell me who was playing whom. This is like asking a sports reporter to cover the Super Bowl and having them show up to the wrong stadium. Both ChatGPT and Claude nailed the schedule on the first try, making Perplexity's struggles all the more puzzling.

For a platform that prides itself on real-time information retrieval, this was like watching a master sommelier confuse wine with grape juice. Perplexity's eventual explanation—something about "incomplete web page previews" and "snippet content"—felt like the AI equivalent of "my dog ate my homework."

The AI Personalities Emerge

Once we got past the basic logistics, the three platforms showed remarkably different approaches to football prognostication.

ChatGPT: The Confident Brainstormer ChatGPT came across as the friend you'd want to bounce ideas off of at a bar. It was conversational, confident, and easy to work with. Its picks showed a willingness to go against the grain—taking the Falcons over the Bucs, backing the Raiders over the Patriots—while providing clear, digestible reasoning. ChatGPT felt like it was actually enjoying the process of making picks, and showed good knowledge of current storylines (like Aaron Rodgers joining Pittsburgh for a revenge game against his former Jets team).

Claude: The Analytical Perfectionist Claude brought the receipts. Its analysis was thorough, nuanced, and impressively detailed. It considered injury reports, coaching changes, and even weather conditions. Claude seemed to actually understand football strategy, discussing offensive line issues, defensive upgrades, and matchup advantages with the sophistication of a beat reporter. It was the AI equivalent of that friend who shows up to your Fantasy draft with color-coded spreadsheets and backup plans for their backup plans.

Perplexity: The Cautious Consensus Builder Once Perplexity finally figured out what games were being played, it produced solid, if unremarkable, picks that largely aligned with conventional wisdom. Its reasoning was sound but safe—the kind of picks that won't win you the pool but probably won't embarrass you either. Perplexity felt like the AI that studied everyone else's homework and then submitted a competent but uninspired version.

The Picks: Where Artificial Minds Diverge

Looking at the predictions, several games revealed the different AI philosophies:

Tampa Bay at Atlanta: Claude went with the Bucs (citing Atlanta's offensive line issues), while ChatGPT and Perplexity backed the Falcons (home field advantage). This felt like the perfect encapsulation of detailed analysis versus surface-level trends.

Miami at Indianapolis: Only Perplexity took the Dolphins, showing some independent thinking after its earlier struggles. The reasoning about Miami's explosive passing game was actually quite sound.

Pittsburgh at New York Jets: This is where the real Aaron Rodgers revenge storyline comes into play—he did sign with Pittsburgh after being released by the Jets, making this Week 1 matchup quite compelling. Claude and ChatGPT both took Pittsburgh, while Perplexity went with the Jets, perhaps giving the home team the benefit of the doubt.

Monday Night Tiebreaker: The total points predictions clustered tight—ChatGPT at 46, Perplexity at 44, Claude at 43, and my own pick at 44. When even the AIs can't agree, you know you're in for a coin flip.

What This Reveals About AI

This experiment illuminated something crucial about how these platforms approach complex, subjective tasks. ChatGPT excelled at the collaborative aspects—it felt like working with a knowledgeable partner who could adapt and brainstorm. Claude brought genuine analytical depth, processing information in ways that felt almost human-expert level. Perplexity, despite its research hiccups, ultimately provided solid consensus thinking.

But here's what struck me most: none of them captured the essential irrationality that makes sports predictions both maddening and compelling. They approached this like a math problem to be solved rather than a beautiful mess of human emotion, random chance, and pure chaos that defines NFL Sundays.

Each AI brought its own biases and blind spots. ChatGPT was good at identifying compelling storylines and showed solid current knowledge. Claude sometimes overthought obvious picks. Perplexity couldn't even get the basic facts straight initially, though it recovered with reasonable analysis.

The Human Element

As I compiled my own picks alongside these digital oracles, I realized something interesting: I was using them not as replacement decision-makers, but as sophisticated research assistants. Each platform highlighted different angles I might not have considered, from injury reports to historical matchups to coaching tendencies.

My final picks represent a blend of their insights with my own football intuitions (such as they are). Did I trust the AIs more than my own judgment? Absolutely. Will that prove to be a mistake? We'll know in exactly one week.

The Verdict (So Far)

If I had to rank them based purely on this first week's performance:

1. Claude wins for analytical depth and genuine football insights
2. ChatGPT takes second for collaborative ease and creative reasoning (even if some of it was fictional)
3. Perplexity brings up the rear, despite eventual solid picks, because of the basic execution failures

But here's the beautiful thing about football—and maybe about AI—none of this might matter come Monday night. The Ravens could blow out the Bills, the 49ers could struggle in Seattle, and my gut feeling about taking Chicago over Minnesota could prove that human irrationality beats artificial intelligence.

That's the real test of AI: not whether it can process information or generate coherent analysis, but whether it can navigate the gloriously unpredictable chaos of real-world events. In one week, we'll see how artificial intelligence fares against the beautiful randomness of grown men in helmets chasing a bouncing ball around a field.

May the best algorithm win. (But I'm still putting my money where my gut is.)

Week 1 Predictions Summary

Results to be updated after Week 1 games conclude.