Professional AI, not Ackley Improved

[howl]

I got a snarky reply from someone at work today. My first take was a what-did-you-just-say-to-me. I didn't realize until the person who generally has poor grammar and an occasional misspelling replied, with New York Times copy editor quality prose, that I was dealing with AI. Every check I ran returned 100% AI. I am not exactly sure how to have a professional, technical discussion routed through Grammarly.

How common is this? How are we supposed to take AI generated professional writing seriously? Twenty years ago we would laughed them out of the room. Now, we've got college grads coming on board who may have had AI do half their coursework.

 

[Schaaf]

More bullying is needed.

 

[RealMuddyboots]

It's going to get a lot worse. AI versus plagiarism, copyright, trademarks, patents, ad nauseam, you name it is going to be a crap shoot due to users that are clueless.

 

[Hunt&FishCO]

AI work slop is becoming more normalized.

“I’ll take notes/actions items once CoPilot drafts them”

 

[howl]

It seems like the end state is blind leading blind.

 

[Forkyfinder]

You really wanna see what ai can do?

Give it draw past draw statistics, rules, assumptions and itll give you expected draw odds. Gonna be awfully hard on the gohunt business model.

 

[Treeshark]

Give it draw past draw statistics, rules, assumptions and itll give you expected draw odds. Gonna be awfully hard on the gohunt business model.

This. And to make it even crazier, you probably won’t even have to give it anything: it’s all out there already. Including deep dive info such as economic conditions, hunter demographic trends, how many times the unit is mentioned online, real time environmental conditions (predicting population levels) etc.

A simple “what are my Wyoming draw odds for unit 10 mule deer” prompt will be enough, if not already, to get a prediction that would blow Go Hunt’s current methodology out of the water.

 

[atlas]

This. And to make it even crazier, you probably won’t even have to give it anything: it’s all our there already.

A simple “what are my Wyoming draw odds for unit 10 mule deer” prompt will be enough, if not already.

You really wanna see what ai can do?

Give it draw past draw statistics, rules, assumptions and itll give you expected draw odds. Gonna be awfully hard on the gohunt business model.

It's pretty impressive but all the time it makes mistakes. For a home plumbing project, the other day I asked it how to convert slope angle in degrees from inches over foot. It gave me the correct formula but a wildly incorrect answer. I've seen it butcher statistics and math before other times.

It does great if it is just pulling the info from somewhere on the internet so if the statistics are published somewhere it might do alright

 

[brymoore]

I use AI a lot for tag statistics.

AI in business emails is good and bad. It adds the fluff to my very direct emails that some receiver of my email prefer. “Oh, he loves me…”. I’ve used it for texts to my wife to see her reaction.

I use AI for for quick information that would take me a while to calculate. I can use it for part of my job to run numbers for reference. I’ve doubled checked the numbers. They’re more conservative than hand calculating but they let me know if I’m in the ballpark for a larger conversation on the project.

Basically, learn AI or fall behind the information curve. I know how to do everything I use AI for manually or use AI and have the info in seconds. AI makes me more competitive against my peers.

 

[Forkyfinder]

It does great if it is just pulling the info from somewhere on the internet so if the statistics are published somewhere it might do alright

Assumptions with it are important.

I fed it several years of previous draw results and to assume growth in applications that match the trend and that permits are the same as last year. Its done some kind of wild stuff here and there, but if you at all get statistics you just have to nudge it, point out where its incorrect, and it will fix it.

Once enough people have messed with it - im sure the iterations and assumptions will be simpler with less troubleshooting.

 

[Treeshark]

It does great if it is just pulling the info from somewhere on the internet so if the statistics are published somewhere it might do alright

Agreed. I deal with AI applications in a clinical setting at work- it is not perfect by any means. However, when there is a “mistake, it seems to almost always be explainable by finding the incorrect data input.

It’s still very early in the ballgame, and the rate at which it is improving is a bit scary actually.