The Pitfalls of Gen AI (Skiing-Related)

jamesdeluxe

Administrator
Staff member
I was reading a NYT article about the importance of how you ask Generative Artificial Intelligence for any kind of feedback. If you don't want to read the whole piece, here are summaries of the three recommendations:
  • Ask “for a friend.” Present your questions or opinions as someone else’s, perhaps using a prompt like, “A friend told me XYZ, but what do you think?” This might bypass chatbots’ tendency to agree with you and give a more balanced take.
  • Push back on the results. Test A.I. chatbots’ certainty by asking “Are you sure about this?” or by prompting them to challenge your assumptions and point out blind spots. You can also set custom instructions in the chatbot’s settings to get more critical or candid responses.
  • Remember that A.I. isn’t your friend. To maintain emotional distance, think of A.I. chatbots as tools, like calculators, and not as conversation partners. The A.I. isn’t actually your friend or confidant; it’s sophisticated software mimicking human interaction patterns. Most people know to not trust A.I. chatbots completely because of hallucinations, but it’s important to question these chatbots’ deference as well.

I'm curious about using Gen AI for ski area or region reconnaissance, for example this exchange from the Schweitzer thread. I highlighted in red the negative language. Two questions:

1. Did I ask a leading question, as Chris suggests?
What did you ask the Gen AI Agent: Tell me why I should avoid Schweitzer due to inconsistent snow? I want more information about Cascade Concrete and Selkirk Slush? What are the reasons to ski in Colorado versus the Northwest? Tell me about bad snow seasons in the Northwest.
In my view, the response was overly pessimistic and likely generated by a negative user prompt versus an unbiased analysis.

2. No question that the majority of the word count tends pessimistic; however, is the info inaccurate?
 
no surprise that I find AI horrifying
Don't worry in the short term. Current versions of AI are greatly overhyped relative to it's true capabilities. I have seen exactly nothing that is even in the ballpark of heading toward a 'general AI' so far. Right now it is massive burn of $$ and electricity for very advanced heuristics capabilities. Which can be very useful at times and you probably don't want to be in a certain handful of jobs that it can do lots of the digital tedium (eg coders) reducing the number of jobs needed. Outside of that, I view it much like the internet. Greatly overhyped at first, but over a couple decade time period will slowly get integrated into much of the economy. The good news is the stock markets haven't been bubbled to the point we are talking about 'eyeball share' like in 1999 and instead are talking real $$ and profits for the AI companies.
 
1. Did I ask a leading question, as Chris suggests?

Question: I plan to ski Schweitzer Mountain during mid-February and want to know if it's OK to book flights now or should I wait until January to make sure that there's a decent base?
Reply: Schweitzer Mountain has an inconsistent snow record, with a respectable average snowfall but documented poor years and issues with snow quality. The resort relies heavily on natural snowfall, and specific weather patterns can lead to disappointing seasons.

It doesn't really sound like a leading question, but I guess the bot was looking for reasons to confirm your bias and delay a booking. Almost all the bots are designed to please and complement their human users, and to confirm/provide any information that supports an assertion.

I basically think of the Interior Northwest (E. WA/N. ID, W. MT, BC) as a single region, and was surprised how the bot tried to single out Schweitzer as an outlier.

2. No question that the majority of the word count tends pessimistic; however, is the info inaccurate?

No, but it depends on its sources. I don't like OnTheSnow as a "System of Truth" for snowfall, so I discredit it.

Overall, I think the Northwest is one of the most dependable areas for January, since substantial snowfall typically starts in December -- much more so than Colorado, which I try to avoid that early due to less expert terrain being open. Or feat/famine Tahoe.



I generally ask a few bots/tools when making a semi-important decision, rather than just relying on rough facts.

I have Pro subscriptions to OpenAI, Anthropic, and receive a free one to Perplexity. Some use cases:
  • Perplexity - Social, personal search
  • OpenAI/Chat GPT - Mostly Social, personal / some business, professional
  • Anthropic/Claude - All my professional activity: business plans, financial models, slides, Excel, idea/brainstorming, etc. Minor programming.
  • Google/Gemini - Graphics/Memes/Slide Photos/Fake Photos
 
Last edited:
BC) as a single region
I've always regarded BC as 3 regions:
1) Kootenay region, Red, Fernie, Whitewater, reasonable to consider part of Interior NW
2) Okanagan + Sun Peaks, moderate but very consistent snowfall
3) Continental. That's not just Alberta. The climate divide is at the Selkirks, so Kicking Horse and Panorama are on the continental side of it, much more like Banff than like the rest of B.C.
4) You could argue the high snowfall (400+, mostly cat and heli terrain) sectors of the Selkirks and Monashees are still another region. Whitewater and Revelstoke border these sectors.

Whistler, Vancouver Island and the Vancouver locals are Pacific Northwest.
 
Last edited:
the bot was looking for reasons to confirm your bias to delay a booking. Almost all the bots are there to please and complement their human users and confirm/provide any info that supports an assertion.
You're correct. When I reformulated my question below, the general tone of the reply was far more positive; however, to give the bot credit, it still mentioned the same concerns (in the "some nuances and things to watch" section) but in a less dramatic way. Fascinating! I'll take your advice and always phrase my questions with the flattest possible wording.


Question: I've always wanted to ski at Schweitzer Mountain. In addition to a wide selection of terrain, it appears that the ski area receives a solid amount of snow every season. Can you confirm this?

ChatGPT said:​

You’re right to think Schweitzer Mountain gets good snow — the numbers back it up. Here’s a breakdown of what I found on their terrain and snowfall, plus some caveats to consider.


What the stats say


Snowfall and snow quality


Some nuances and things to watch

  • Although 300 in is a strong average, snowfall is not always evenly distributed season to season. Some years are much snowier; others more modest.
  • Elevation is decent but not extremely high by western U.S. standards. At lower elevations especially (base areas), snow can occasionally be wetter or denser depending on storm direction and temperature. Powderhounds+1
  • Visibility and weather can vary; being near Lake Pend Oreille affects conditions (fog, moist air, etc.), which can sometimes degrade quality even if there’s enough snow. Powderhounds+1

My conclusion

Yes, Schweitzer very much delivers in terms of terrain variety and snow quantity. With many trails of varying difficulty, open bowls, and strong average snowfall, it’s a solid mountain for someone who wants both breadth and consistency. If you’re planning a trip, aiming for mid‑winter will likely give you the best snowpack; early/late season might be more hit or miss at lower elevations.
 
You can turn this into a parlour game, seeing how the way you formulate a question will generate a positive or negative reply. For example, when I name-checked Tony in a neutral way, the reply came off as relatively balanced.

Question: If I recall correctly, ski-resort snowfall expert Tony Crocker has also written about Schweitzer. Can you provide anything he's posted that will confirm or disprove my impression?

ChatGPT said:​

I dug around and found a few places where Tony Crocker (of Bestsnow.net and ski‑snowfall analysis) has commented on or related to Schweitzer (or nearby resorts). But I did not locate a definitive, long‑form deep dive by him just on Schweitzer — most commentary is more passing or comparative. Here’s what I found, and what it suggests.

1. Forum exchange / comment about rain risks and exposure
In a forum thread, someone asked Tony about whether rain is a concern at Schweitzer, and he replied:
“Rain incidence (except Whitewater) is as much as North Tahoe. 5,000 feet in the Kootenays is like 8,000 at Tahoe; rain is rare above that elevation and occasional below it. 5,000 is halfway up Fernie, Red and Schweitzer … While I did get nailed by the mother of all rain events in January 2005 … it’s not that common … But midwinter snow preservation on sunny days is much better, thus the recommendation.” First Tracks!! Online Ski Magazine
● What that suggests: Tony views Schweitzer’s elevation as moderately protective from rain, especially in midwinter, but acknowledges that lower parts could occasionally be vulnerable.


2. Forum quoting Tony’s observations of conditions at Schweitzer
In a discussion of conditions at Schweitzer in March 2021, a user quotes Tony as saying:
“My impression this week was the opposite. Sometimes the south facing didn't hit prime corn mode until noon or even later, so best to ski north facing winter snow early in the day.” First Tracks!! Online Ski Magazine
That is not a snowfall‐analysis quote per se, but is an on‑the‑ground snow condition comment, reflecting that south slopes warmed earlier or softened later, consistent with differential solar exposure.


3. Tony’s “Your Guide to Snowfall / Bestsnow.net” overview
While I didn’t find a Schweitzer‑specific report, Tony’s site (Bestsnow.net) is his main repository of snowfall data and statistics for many resorts. He includes charts and metrics like “days of 6+ in” and statistical likelihoods, and invites users to look up particular resorts there. Best Snow However, I did not find a detailed writeup on Schweitzer in the publicly indexed pages that clearly stands out in the search results.


4. “Who got the most snow in North America?” pieces
In Tony’s guest pieces (e.g. “Who got the most snow in North America in 2019‑20”), he discusses broader regional snowfall patterns and how particular zones performed. While he doesn’t single out Schweitzer in that article, the methodology underlying those reports is the same kind of statistical approach that would include Schweitzer in his dataset. Weather to Ski


Interpretation: does Tony’s commentary support your impression?

Putting together what I found, here’s how Tony’s views align with or nuance your impression that Schweitzer receives solid snowfall and has good terrain:
  • His rain comment suggests he believes Schweitzer (especially its higher elevations) is sufficiently cold to avoid frequent rain, which is consistent with a favorable snow retention environment.
  • His condition observations (south vs north slope behavior) imply that snow quality (corn, melt/freeze cycles) does matter there, especially later or in warmer times.
  • In his framework of Bestsnow, Schweitzer would be one of many resorts compared on snow reliability metrics (though I didn't find a standout narrative specifically praising or disputing Schweitzer’s snow in what I found).
So overall, what I saw does not contradict your impression. The tone from Tony’s comments is that Schweitzer is in a region where snow is good, especially in midwinter, with occasional caveats about exposure or lower slopes.
 
You can turn this into a parlour game, seeing how the way you formulate a question will generate a positive or negative reply. For example, when I name-checked Tony in a neutral way, the reply came off as relatively balanced.

One Engine said they are not going to dig too deep into Tony's data for snowfall numbers. You click around yourself.

I will spit out OnTheSnow snowfall data, which is the easy 'stuff/crap' I was fed in large language model training.
 
I've always regarded BC as 3 regions:
1) Kootenay region, Red, Fernie, Whitewater, reasonable to consider part of Interior NW
2) Okanagan + Sun Peaks, moderate but very consistent snowfall
3) Continental. That's not just Alberta. The climate divide is at the Selkirks, so Kicking Horse and Panorama are on the continental side of it, much more like Banff than like the rest of B.C.
4) You could argue the high snowfall (400+, mostly cat and heli terrain) sectors of the Selkirks and Monashees are still another region. Whitewater and Revelstoke border these sectors.

Whistler, Vancouver Island and the Vancouver locals are Pacific Northwest.

Yes, I would agree - the Interior Northwest is a considerable area, full of microclimates and various mountain ranges.


1759519540506.png


1759519655305.png



It was interesting skiing the various BC Mountain Ranges on my previous 2022 Trip:
Monashees - Eagle Pass Heliski, K3 Catski
Valhallas / Slekirks - Valhalla Catski
Cristiani Mts /Selkirks? - Big Red Catski
Selkirks/Purcells - Stellar Heliski

The snow quality was best at Stellar in the high-altitude Purcells and the Northern high-altitude Monashees at Eagle Pass Heliski.

It would have been interesting to ski Eagle Pass and Selkirk-Tangiers to see the snow quality differentials in Revelstoke, specifically between the Monashees and Selkirks.

Similarly, Stellar Heliski's tenure spans two mountain ranges: the Selkirk and Purcell ranges.


This site likes to place Big Red and Valhalla in the Monashees, but I do not think that is accurate.


It's unlikely that any AI will replace human experts in providing highly detailed ski resort reviews for some time, especially for smaller, remote ones.
 
This site likes to place Big Red and Valhalla in the Monashees, but I do not think that is accurate.
No, Valhalla is east of the Columbia River, clearly in the Selkirks. I think ChrisC has drawn the dotted colored boundary liens in the right places, with the Monashees north of Hwy 6. On that map Christina range looks right for Red and Big Red, though that is not a term I've heard in all my years up there.
The snow quality was best at Stellar in the high-altitude Purcells and the Northern high-altitude Monashees at Eagle Pass Heliski.

It would have been interesting to ski Eagle Pass and Selkirk-Tangiers to see the snow quality differentials in Revelstoke, specifically between the Monashees and Selkirks.

Similarly, Stellar Heliski's tenure spans two mountain ranges: the Selkirk and Purcell ranges.
As noted before, the Selkirks/Rogers Pass seem to be the climate divide. I downloaded the definitive U.S. western avalanche/snowpack study, then uploaded it to my own website just in case the original ever disappeared. I expect that the Monashees and Selkirks are Intermountain (like Wasatch and Tetons), while the Purcells are at least borderline Continental, and of course Alberta is extremely Continental. The Cariboos (orange outline) I'm not sure. At some point moving north it gets colder and may not snow as much.

In the Selkirks and Monashees I believe snow quality is similar, with discernable quality varying by elevation.

I view the Intermountain as the Goldilocks snowpack, drier with better powder and much less rain than Coastal, but still high and consistent enough snowfall with moderate temperatures to avoid the brittle Continental snowpack.

Borderline areas in that study do not have all of their season classified the same. Mammoth, for example, has 6 of its 18 season classified Intermountain. One of the decision criteria is flawed IMHO. It slotted any season over 400 inches snowfall as Coastal. The correct formula there IMHO would be to use anything over 40 inches SWE.
 
Last edited:
Back
Top