Fly fishing and AI

[greenghost]

I’ve been fly fishing for five decades. Over that time the sport has seen remarkable changes. From starting out as an eccentric—often viewed as a hoity-toity—way to fish to the faddish boom that “A River Runs Through It” brought, to today’s world, the digital age of angling, where newbies can become incredibly effective in a very short time thanks to the Internet, countless videos, webinars, and advanced products.

It all begs the question, “What’s next?”

AI is the answer. It will take fly fishing, all fishing, to the next level In fact, the commercial fishing industry is already using operational AI in smarter “fish finders” that remember past successes and recommend, based on that data, the best place to fish. It’s a matter of time before it becomes available to recreational anglers. The old adage of “90% of the fish is in 10% of the water”? AI will be very good at finding that 10% of the water based on a wide variety of data input. Generational AI (similar to ChatGPT) will be used to create new fly patterns based on the effectiveness of current ones, stream-specific entomology, material availability, etc.

Is it all good or bad? Will it take some of “art” of the sport away? It’s a shortcut to the work involved in of finding new fishing waters and the creative aspect of fly tying? It’s a matter of opinion.

One thing I do know, as a marketing writer, I’ve been using AI and found it to be a remarkable tool. Not a hindrance but a doer of the more mundane tasks. However, I’ve always considered it still not quite there yet to come up with humanistic, truly creative ideas that resonate. Then today I asked it to write a short poem about flyfishing and what it means to me. This is the result:

Amidst the currents and the dancing flies,
I leave behind mundane ties,
Among the mountains, under the sky,
I find my peace, I learn to fly,
Yes. In this art, I've found my decree,
Fly fishing's become a part of me.

GULP I’m glad I have a 401(k)

 

[sixfootfenwick]

I’ve been fly fishing for five decades. Over that time the sport has seen remarkable changes. From starting out as an eccentric—often viewed as a hoity-toity—way to fish to the faddish boom that “A River Runs Through It” brought, to today’s world, the digital age of angling, where newbies can become incredibly effective in a very short time thanks to the Internet, countless videos, webinars, and advanced products.

It all begs the question, “What’s next?”

AI is the answer. It will take fly fishing, all fishing, to the next level In fact, the commercial fishing industry is already using operational AI in smarter “fish finders” that remember past successes and recommend, based on that data, the best place to fish. It’s a matter of time before it becomes available to recreational anglers. The old adage of “90% of the fish is in 10% of the water”? AI will be very good at finding that 10% of the water based on a wide variety of data input. Generational AI (similar to ChatGPT) will be used to create new fly patterns based on the effectiveness of current ones, stream-specific entomology, material availability, etc.

Is it all good or bad? Will it take some of “art” of the sport away? It’s a shortcut to the work involved in of finding new fishing waters and the creative aspect of fly tying? It’s a matter of opinion.

One thing I do know, as a marketing writer, I’ve been using AI and found it to be a remarkable tool. Not a hindrance but a doer of the more mundane tasks. However, I’ve always considered it still not quite there yet to come up with humanistic, truly creative ideas that resonate. Then today I asked it to write a short poem about flyfishing and what it means to me. This is the result:

Amidst the currents and the dancing flies,
I leave behind mundane ties,
Among the mountains, under the sky,
I find my peace, I learn to fly,
Yes. In this art, I've found my decree,
Fly fishing's become a part of me.

GULP I’m glad I have a 401(k)

IMO, it's dangerous to nearly all facets of our lives.

 

[krayfish2]

Who is Al? Should I know this guy?

 

[troutbert]

I’ve been fly fishing for five decades. Over that time the sport has seen remarkable changes. From starting out as an eccentric—often viewed as a hoity-toity—way to fish to the faddish boom that “A River Runs Through It” brought, to today’s world, the digital age of angling, where newbies can become incredibly effective in a very short time thanks to the Internet, countless videos, webinars, and advanced products.

It all begs the question, “What’s next?”

AI is the answer. It will take fly fishing, all fishing, to the next level In fact, the commercial fishing industry is already using operational AI in smarter “fish finders” that remember past successes and recommend, based on that data, the best place to fish. It’s a matter of time before it becomes available to recreational anglers. The old adage of “90% of the fish is in 10% of the water”? AI will be very good at finding that 10% of the water based on a wide variety of data input. Generational AI (similar to ChatGPT) will be used to create new fly patterns based on the effectiveness of current ones, stream-specific entomology, material availability, etc.

Is it all good or bad? Will it take some of “art” of the sport away? It’s a shortcut to the work involved in of finding new fishing waters and the creative aspect of fly tying? It’s a matter of opinion.

One thing I do know, as a marketing writer, I’ve been using AI and found it to be a remarkable tool. Not a hindrance but a doer of the more mundane tasks. However, I’ve always considered it still not quite there yet to come up with humanistic, truly creative ideas that resonate. Then today I asked it to write a short poem about flyfishing and what it means to me. This is the result:

Amidst the currents and the dancing flies,
I leave behind mundane ties,
Among the mountains, under the sky,
I find my peace, I learn to fly,
Yes. In this art, I've found my decree,
Fly fishing's become a part of me.

GULP I’m glad I have a 401(k)

That sample of AI output is very poor writing.

 

[Silent Ocelot]

That sample of AI output is very poor writing.

I concur. The last line would have a much better cadence if it read "fly fishing's a part of me." Really the last two lines don't sound so hot, but, overall it doesn't read very pleasantly. I'm not a strong believer in all this AI stuff. I doubt an AI could ever write something up like Cormac McCarthy.

Rumor through the grapevine is that you're an author.

 

[albatross]

I'm glad to be closer to the exit of my career than the entrance. AI is definitely going to shake things up.

 

[Mike]

However, I’ve always considered it still not quite there yet

While critiquing the poem, the critics appear to have missed what greenghost said about AI above, making the critiques a bit unnecessary. The refinements are coming though and then watch out. For now I’m impressed with how far they have gotten.

 

[sciflyguy]

Funny you should mention this. Me and a friend had a similar idea a few years back when I was still a bass fisherman.

One of our favorite fisheries was Lake of the Ozarks, a behemoth of a lake in southern MO with >1000 miles of shoreline. At the time, there was a pay-to-view website with daily bass fishing reports, all from Lake of the Ozarks and dating back many years.

The project began with us coding up a web scraper to extract the relevant data (water temp, lure, #'s of fish caught, size of fish caught, what part of the lake fish were caught on, what structure the fish were caught nearest, etc.) from each individual webpage, because each report had to be opened in a new tab/window which made them very slow to view manually. We scraped this data and put all 2,000 reports into a spreadsheet, allowing us to use Excel to quickly sort to find days with similar water temp, time of year, sky conditions, etc. to whatever it was on the day we were going fishing, and then use those functionally similar fishing reports to inform lure choice and location (on such a massive lake even simply points vs coves vs pockets vs brush piles can take days to determine for yourself). This worked pretty well and helped us find the fish a lot faster, but we wanted more.

The subsequent goal was to use machine learning (random forest) to help identify underlying patterns in the data and make optimal predictions for a given input of water temp, sky conditions, air temp, and date. We had to use some basic NLP to further clean up the data, because while some things like air temp and water temp were easy to scrape out directly, other things like lure were much harder to scrape out of the fishing report because they were usually contained in a paragraph of "other information". We got this done well enough, trained the random forest model on "big fish" datapoints, and finally tested it on our test dataset and the results were...lackluster.

In conclusion, I like the idea of applying AI/ML to sport fishing, but I think that data quality/consistency is a major hurdle, and it's likely that even with high quality data, an incredibly large n (>100,000?) would be needed to overcome the fundamental fact that even on a single river/lake, there will be many different lures/flies that can be used to catch large fish, and those large fish can be caught in many different locations

(The main limitations, to be fair, were our expertise (or lack thereof) with NLP and ML optimization. Scikitlearn and the like make it shockingly easy to apply these amazing algorithms, and I know that we picked a suitable algorithm, but when it came to optimization we were pretty at a loss.)

Also, while I mostly talk about ML here while your post is about AI, my point holds: both of these technologies necessarily can only be as good as the data they learn from. Not only is there is a paucity of high quality data for sport fishing, but the data which does exist has properties (>>1 way to catch good fish at any given time) which make it particularly challenging for them to learn from in the context of any reasonable n.

 

[poopdeck]

When did Allen Iverson start fly fishing? I love how Darwinism is at work killing the dumb in their “self driving” teslas. it’s not quite there yet is an understatement.

I think I’ll continue with the old ways of fishing even when the system is perfected. Going out knowing I’m going to catch fish somehow loses the essence of fishing.

 

[DomR]

I'm glad to be closer to the exit of my career than the entrance. AI is definitely going to shake things up.

More like "F" things up! AI is going to turn a whole lot of high paid job titles into fry clerks in the next 10 years.

 

[sixfootfenwick]

More like "F" things up! AI is going to turn a whole lot of high paid job titles into fry clerks in the next 10 years.

Im not a fan either but I also see the irony.

People in high paid positions want automation and I threatens the low scale jobs, especially manufacturing. Now those high paying jobs are at threat because technology has gotten better and invented AI.

 

[tomgamber]

More like "F" things up! AI is going to turn a whole lot of high paid job titles into fry clerks in the next 10 years.

Nah, when the suits still can't get what they think they want they'll have to bring back people to blame stuff on.

 

[caddisflyer]

I think I need AI after posting some of the garbled crap I write on this site and later reread and cringe. I think my brains going bad.

 

[troutbert]

One thing I do know, as a marketing writer, I’ve been using AI and found it to be a remarkable tool. Not a hindrance but a doer of the more mundane tasks. However, I’ve always considered it still not quite there yet to come up with humanistic, truly creative ideas that resonate. Then today I asked it to write a short poem about flyfishing and what it means to me. This is the result:

Amidst the currents and the dancing flies,
I leave behind mundane ties,
Among the mountains, under the sky,
I find my peace, I learn to fly,
Yes. In this art, I've found my decree,
Fly fishing's become a part of me.

That is very bad writing.

Does anyone have examples of good writing by AI?

 

[PennKev]

Generational AI (similar to ChatGPT) will be used to create new fly patterns based on the effectiveness of current ones, stream-specific entomology, material availability, etc.

I can imagine an Alexa voice repeatedly recommending that I tie a fly with pheasant tail and peacock herl...

Uh, yeah, we already do that one Alexa.

 

[Silent Ocelot]

While critiquing the poem, the critics appear to have missed what greenghost said about AI above, making the critiques a bit unnecessary. The refinements are coming though and then watch out. For now I’m impressed with how far they have gotten.

Big difference in not quite there and very bad writing, as Troutbert stated twice.

I am not terribly impressed with AI being involved in fishing. Computers are substantially better than humans at crunching data to anticipate outcomes, but they lack intuition. Sometimes when fishing you just simply know what is going to work by the environmental factors presented and just have a pretty good guess at what will work. A computer, however advanced, will never be able to grasp these concepts. Sure, it can come out with an ideal fly based on a data found on the internet to best match a hatch, but all I have to do is look under a rock...

 

[caddisflyer]

Artificial intelligence is fake intelligence

 

[troutbert]

Artificial intelligence is fake intelligence

"Artificial intelligence" is an inaccurate term.

A more accurate term would be "imitation of intelligence."

These programs are "trained" on massive quantities of the writings of human beings.

So, everything AI writes is derivative of what humans have created.

So, it's just regurgitation, with some re-arrangement.

Epic copyright cases are on the way, because these AI programs are using peoples' copyrighted material without permission.

 

[FarmerDave]

And here I thought Artificial Intelligence was when a blond dyes their hair brown.

 

[tomgamber]

"Artificial intelligence" is an inaccurate term.

A more accurate term would be "imitation of intelligence."

These programs are "trained" on massive quantities of the writings of human beings.

So, everything AI writes is derivative of what humans have created.

So, it's just regurgitation, with some re-arrangement.

Epic copyright cases are on the way, because these AI programs are using peoples' copyrighted material without permission.

It's no a violation of copyright to paraphrase, segment or reference. It's just basically a human cheating by letting a program do one's research and writing. However the applications, for those looking to solve real problems rather than claim they wrote something they didn't, are incredible. Who doesn't Google when they need to know something they don't. Imagine. With the right prompts, being able to access the entire internet (these days that means anything that's ever been written) to solve real problems. Energy, money, medicine, there are no limits. But like alternative energy sources and autonomous machines, some will resist to the point of holding society back from progressing once again.