I love sports. In fact, I’m obsessed with them. Whether it’s soccer, darts, UFC or NFL, if it’s on TV I’m watching it. For a long time, I’ve struggled with fast-paced US sports because they’re televised late at night and the speed of the game puts me into a sleep trance.
A few months ago, a friend of mine invited me to a local ice hockey game, and after watching one puck drop of Scottish semi-professional hockey, I was absolutely hooked. Since then I’ve watched Prime Video’s Face-Off, which showcases professional hockey players in North America, and more importantly, I’ve become enthralled with the NHL.
At weekends, I’ve watched every NHL game shown at sociable hours in the UK, and I’ve been to most Edinburgh Capitals games to get both sides of the hockey spectrum. I’ve fallen in love with the NHL so much that I’ve purchased tickets to go to Stockholm, Sweden in November to see the Nashville Predators take on the Pittsburgh Penguins.
Anyways, you’re probably thinking to yourself, this isn’t about AI, this is just about ice hockey. But that’s where things get interesting. Over the last few weeks, I’ve been trying to pick an NHL team to show some allegiance to, and considering I cover AI for work I thought what better way to test AI research tools than to see if it can help me whittle down my hockey-supporting options?
Over the last week or so, I’ve used ChatGPT and Gemini to create research reports, fun quizzes, and even AI-generated podcasts to help make my NHL team decision that little bit easier. Here’s how the process went, and how AI is great for fun projects like this, while also not completely reliable.
A personality check
Let me paint the picture, I’ve been watching the NHL for close to two months now, and in that period I’ve learned a lot about players, teams, and general hockey rules. So I thought my first port-of-call should be a fun quiz to determine what team lined up with my sporting preferences.
I went to ChatGPT 4o and asked “Can you ask me questions to determine what NHL team I should support? Make it a fun quiz.” ChatGPT compiled a 10-question multiple-choice quiz that aimed to find out what kind of playstyle I like to watch, what kind of fan I want to be, and even how much success I want in the short term.
After finishing the quiz ChatGPT determined that “You’d be a great fit for the Toronto Maple Leafs!” because of the team’s “rich history, huge fanbase, and fast, exciting playstyle.” Not bad I thought, although random quizzes that match your personality to a team are probably the worst way to decide this.
That said, it was a starting point and I’d be lying if I said I wasn’t happy that AI linked me up with a team that I’d really been enjoying watching on live TV.
Deep Research failed me
Ok, so I’ve tried a silly quiz and have a starting point, but let’s be real, there’s no way I can decide on a team to support based on an AI-built personality test. Next step: AI research tools.
Research tools like ChatGPT Deep Research and Gemini Deep Research are a trendy topic in the world of AI right now, and what better way to see what they are capable of than by asking for in-depth NHL research?
Ultimately, whoever I choose to support will be determined by who I get to watch, and living in the UK makes midweek games almost impossible, and weekend games incredibly sporadic.
I decided to get a real understanding of my actual options, I needed to research how regularly teams play at a reasonable time in my timezone.
To do this, I asked both ChatGPT Deep Research and Gemini’s version the following prompt: “Can you do research into NHL match start times over the last 5 seasons? I want to know the % of matches for each team that started between 4:30 pm and 10 pm UK time during the week and 4:30 pm and 11 pm on the weekend.”
Both programs then asked follow-up questions regarding my preferences on the research such as whether or not I wanted just regular season games or playoff games taken into account, and how I wanted the results to be produced.
I gave some extra info, and then let AI do its thing. Gemini Deep Research was much quicker than ChatGPT, in fact, it took just a couple of minutes compared to well over 10 minutes for OpenAI’s research analyst.
Gemini’s report gave me a comprehensive table of percentages per season and then an average, with Tampa Bay Lightning leading the pack with a total of 28.35% of games at a regular UK time over the last five NHL seasons.
ChatGPT on the other hand found that the team with the most matches at a reasonable time for a UK viewer over the last five seasons was the New York Rangers with 14% of matches falling into the timeframe I was looking for.
Notice a difference? Both Deep Research tools gave me completely different results. Not ideal…
The results were so different, that I decided to run Gemini again to see what would happen. Fast forward roughly five minutes and I was being told that 45.5% of Montreal Canadiens games were in my desired timeframe, I know for a fact that’s just not true.
So after an hour of using Deep Research, I had a dilemma. First of all, it would take me far too long to conduct the actual research required to collect info and fact-check my fun AI experiment, secondly, ChatGPT’s results look more realistic than Gemini’s two attempts, and thirdly, I wasn’t any closer to picking a team.
What’s my decision? And what have I learned?
After testing AI research tools to help pick an NHL team to support, I have a few thoughts, and I’ve come to a conclusion.
First of all, AI research tools aren’t reliable but that doesn’t mean they can’t be of use for fun projects like the ones I tried today. While I don’t really have an accurate answer on what team plays the most matches between 4:30 pm and 11 pm UK time, I have a rough idea of what East Coast teams play at earlier times.
If you’re planning to use Deep Research for academic studies, or compiling important information then I would seriously reconsider. It’s one thing to rely on AI to help pick an NHL team, it’s another to rely on AI for actual work.
I’ve enjoyed using the research tools for this article, and I was even able to feed one of the reports directly to Audio Overview in Gemini to create a realistic podcast clip that sounds fantastic.
So who have I chosen to support? Well, ultimately the AI tools I used didn’t help me that much other than confirm what I already knew about East Coast teams playing at earlier times.
Since I started watching the NHL, I’ve found myself cheering for Montreal Canadiens, and as a french-speaker, I’ve always loved the city. So, while I’m still not 100% sure, It’s looking like I’m a Habs fan now…