|
|
Posts: 11
| With quite a bit of chatter in the news these days on AI,. what impact, if any, will this play in the sport of fishing? |
|
| |
|
Posts: 1415
Location: Brighton CO. | You can't replace a weak mind and a strong back! |
|
| |
|
Posts: 11
| chuckski - 4/13/2023 7:05 PM
You can't replace a weak mind and a strong back!
So true. |
|
| |
|
Posts: 386
| https://youtu.be/ro130m-f_yk
Good video if you don't mind some swearing. Discusses "AI"s shortcomings and a bunch of other rambling about big tech and such. |
|
| |
|
Posts: 1288
| You'll probably get into your boat, and ask Siri to take you to the largest muskie in the lake. She'll fire up the Lorgarminbird integrated boat, and take you right to it, and tell you what bait to tie on, where to cast it, and what retrieve to use. Probably scold you for a bad figure 8, as well. |
|
| |
|
Posts: 32887
Location: Rhinelander, Wisconsin | Muskie angling need not be concerned. Anyone who's addicted seems to be without any sort of I, much less worry about AI. |
|
| |
|
Location: Athens, Ohio | I don't think the fish need it, either; they seem smart enuf already. m |
|
| |
|
Location: PA Angler | Just because you watch or hear someone say something like place doesn’t mean you just go out and catch one. |
|
| |
|
Posts: 398
Location: SW Ohio | AI will end up outlawing all hunting and fishing. "It's bad..."... |
|
| |
|
Posts: 530
Location: NW WI | Spose it depends on your definition. Way I figure, it's already here. Is it "True AI" no of course not. But we already have self correcting & self guiding navigation. Algorithms are used to feed you educational and entertaining media.
I'm not versed at all in computer science but I'd speculate it wouldn't take much for a graph company to pick a few handfuls of the major lakes, pay multiple guides a large some of money (not all will bite) to sit down and cover every single variable and develop a tech where your phone or graph says "Hey! Today do this,this,or this" than you and other users input their data that is shared and adapted per success %. |
|
| |
|
Location: Eastern Ontario | I'm much more impacted by old age and a deterating body than AI for younger anglers it will be the new normal. When I started electric motors and flashers were the new hi tech. |
|
| |
|
Posts: 326
| AI is dependent on the underlying 'library'.The muskie library is,well.......tainted at best.:-) |
|
| |
|
Posts: 110
| I am a data scientist and have already done some work here, so I can provide a few "inside" points.
1. As someone mentioned earlier: AI is a data game. If the underlying data is bad or not trustworthy, then your model is going to suck. Does anyone trust the lunge logs outside of the lake they were caught in and the measurement of the fish? It's one of the same reasons people roll their eyes at fishbrain. Everyone knows the entries are not honest, and for good reason. Nobody wants to burn a hole by giving away too much info.
2. Once you've got enough people willing to help participate in providing non-garbage catch information, you still have a problem of scale. People who have fished on West Battle know it fishes differently than Mille Lacs. There are reasons for this: substrate, depth, size, shoreline development, water quality and clarity, forage, etc. What works on one probably won't work on another. I can make you a pretty solid algorithm for, say, Vermilion or something, that sees enough pressure. It gets REAL complicated past that.
Edited by bloatlord 4/16/2023 12:33 PM
|
|
| |
|
Posts: 7047
Location: Northwest Chicago Burbs | bloatlord - 4/16/2023 12:29 PM
I am a data scientist and have already done some work here, so I can provide a few "inside" points.
1. As someone mentioned earlier: AI is a data game. If the underlying data is bad or not trustworthy, then your model is going to suck. Does anyone trust the lunge logs outside of the lake they were caught in and the measurement of the fish? It's one of the same reasons people roll their eyes at fishbrain. Everyone knows the entries are not honest, and for good reason. Nobody wants to burn a hole by giving away too much info.
2. Once you've got enough people willing to help participate in providing non-garbage catch information, you still have a problem of scale. People who have fished on West Battle know it fishes differently than Mille Lacs. There are reasons for this: substrate, depth, size, shoreline development, water quality and clarity, forage, etc. What works on one probably won't work on another. I can make you a pretty solid algorithm for, say, Vermilion or something, that sees enough pressure. It gets REAL complicated past that.
Ok, so would THIS be enough to create a data set that would give any true predictive measures:
-4 boats, 2 fishers each
-Fishing every day from season open to close
-logging each cast (bait, location, speed of retrieve)
-logging weather/wind/barometer/moon for each day/moment
-log every follow, lost fish, caught fish and all of the above within each muskie encounter.
With this data, could you build a model that would give a "real" AI/ML measure?
*sorry, I'm no DS, but I hunt them at work. |
|
| |
|
Posts: 110
| On one lake? Yeah, probably. . Take note that you've pinpointed something that these "predictive" apps have seemingly ignored: what FAILS to catch fish is just as important as what catches it. Thousands of data points are lost. There are plenty of other variables that are easy enough to pull from open weather API to give a more clear weather picture as well as DNR reports on shoreline development, water quality, etc.
Edited by bloatlord 5/5/2023 1:23 PM
|
|
| |
|
Posts: 3869
| bloatlord - 5/5/2023 2:22 PM
On one lake? Yeah, probably. . Take note that you've pinpointed something that these "predictive" apps have seemingly ignored: what FAILS to catch fish is just as important as what catches it. Thousands of data points are lost. There are plenty of other variables that are easy enough to pull from open weather API to give a more clear weather picture as well as DNR reports on shoreline development, water quality, etc.
Rats, you've caught onto my best scouting trick.... asking muskie guys at the landing, "Well, what didn't work today?" |
|
| |
|
Posts: 110
| Ranger - 5/5/2023 2:55 PM
bloatlord - 5/5/2023 2:22 PM
On one lake? Yeah, probably. . Take note that you've pinpointed something that these "predictive" apps have seemingly ignored: what FAILS to catch fish is just as important as what catches it. Thousands of data points are lost. There are plenty of other variables that are easy enough to pull from open weather API to give a more clear weather picture as well as DNR reports on shoreline development, water quality, etc.
Rats, you've caught onto my best scouting trick.... asking muskie guys at the landing, "Well, what didn't work today?"
You laugh, but classification models work this way. In order for an algorithm to classify between two events (catch vs. didn't catch musky), it needs examples of both. Thankfully, I am providing all of the latter one could ever need. |
|
| |
|
Posts: 3869
| From another perspective, I'm reminded of an obscure scifi book I read 30 years ago, "Cemetary World", I think. In that story the earth had become so polluted that people had to move to space. Wealthy people still used the planet for sports, tho, like hunting. A rich couple would land on the planet, set up a picnic and release a couple animatronic dogs to track down, corner and pin a deer or cougar. The couple would then decide to either hike to and shoot the prey, or just send a directive to the dogs to kill and drag the deer/cougar back to camp. 95% of the hunt was comfortably sitting, eating and drinking at camp.
So, why not a fake fish, a little submarine, to search out and spear a target fish and then drag it back. Real time, with video, like a X-Box game. |
|
| |
|
Posts: 110
| I get the critique of technological creep in the sport, but I think line drawing is largely arbitrary. So long as individuals practice effective Catch-Pic-Release and obey relevant laws, the manner in which they enjoy the sport is up to them.
Also, fwiw, Ray Bradbury's "There Will Come Soft Rains" is a really, really good short story that was a critique of technological advancement at the time of the Cold War. It's a ten minute read, tops. Really good stuff. |
|
| |
|
Posts: 3869
| bloatlord - 5/5/2023 4:21 PM
Ranger - 5/5/2023 2:55 PM
bloatlord - 5/5/2023 2:22 PM
On one lake? Yeah, probably. . Take note that you've pinpointed something that these "predictive" apps have seemingly ignored: what FAILS to catch fish is just as important as what catches it. Thousands of data points are lost. There are plenty of other variables that are easy enough to pull from open weather API to give a more clear weather picture as well as DNR reports on shoreline development, water quality, etc.
Rats, you've caught onto my best scouting trick.... asking muskie guys at the landing, "Well, what didn't work today?"
You laugh, but classification models work this way. In order for an algorithm to classify between two events (catch vs. didn't catch musky ), it needs examples of both. Thankfully, I am providing all of the latter one could ever need.
No laughing. I'm a Certified Quality Manager (ASQ), you don't get that without understanding data management. Way back in the day, MBA school, I could do multivariate linear regression problems on a chalk board. |
|
| |
|
Posts: 110
| Doing ANOVAs with more than 4 groups...... just to have prof hand you back a largely marked down paper because you made an arithmetic mistake early. |
|
| |
|
Location: Athens, Ohio | Animatronic Donald Trump vs. Animatronic Joe Biden November, 2124. Be there. m |
|
| |
