Anyone Using Artificial Intelligence With Their Gold Detectors?

[Steve Herschbach]

18 hours ago, phrunt said:

Just imagine how many of your big Aussie nuggets are left in the ground as current technology can't get deep enough

Just imagine also that this is often largely myth, as plenty of scrape to detect operations have proven over the years. Yes, there are locations where large nuggets lurk just out of reach. But more often than not, desert deposits are formed by a very specific type of wind erosion called deflation, where light surface material erodes away over millenia, leaving extremely rich surface deposits. These deposits deplete rapidly at depth, as many an oldtimer found, when big money was spent going deeper at these places. The result almost always being that the miner went bust. Dreams are the stuff detecting is made of, and this one is lodged firmly in many minds, but the reality is far duller, I'm sorry to say. But sure, big nuggets down deep, newer AI detectors, dream on lads. 

Eolian Placers

"Eolian gold placers have a specific structure. The producing horizon overlies the deflation surface like a blanket, with its thickness extremely small (tens of centimeters)…."

[1515Art]

My experience in the one time I participated in removing 6” to a foot of soil from known patches hit dozens of targets… unfortunately all a byproduct of the equipments blade and bucket, lol. Words of wisdom from our benefactor.
 

Are there areas if you have desert property you would concentrate search effort in the  terrain more likely to hold deeper gold? 

[jasong]

1 hour ago, 1515Art said:

Are there areas if you have desert property you would concentrate search effort in the  terrain more likely to hold deeper gold? 

Yes, there are places pushes still reveal a lot more gold. It all depends on the depositional environment to begin with. The obvious one is deeper washes.

But the key factor, no matter where you are at, is to find spots where the erosional forces bringing rock/soil down onto a patch were either greater or less than the erosional force removing rock/soil from the patch. That's the basic balance to keep in mind when evaluating a spot. 

In places where high energy events occured - floods, landslides, streams, glaciers, or gravity due to very steep terrain - in general deposit more overburden via erosion than is removed by wind, you might get deeper patches. Where the converse is true, especially where mineralization is not geologically ancient (like NNV), deflation dominates and usually dictates a shallower maximum patch depth.

In AZ you often get both a shallow enriched surface (deflationary) patch from wind, together with a deeper, less concentrated patch at depth because of this. Each individual location varies due to it's geomorphological history. 

[Valens Legacy]

On 8/7/2022 at 7:36 AM, Steve Herschbach said:

The problem with a lot of this theorizing about machine learning, is that metal detectors know less than most people think. For instance, they have no idea what metal is under the coil. Any metal can read almost anywhere, depending on the size of the target. Target id only works for items that never vary, like a silver dime, yet many trash items can mimic a silver dime. Still, you can build a detector that can cherry pick coins rather effectively for use in the U.S.

Steve,

Thank you for replying and giving us the answer that should have been said on the first page in my opinion.

I only asked the question about a color screen to see how many responses there would be.

I have used the NOX 800 and have found several silver coins wit a wide range of target ID's. Such as a few quarters, dimes and even 2 trimes. They each are silver but have a different ID at different depths, or even if they are on edge or flat. The larger silver coins have been in the 30's while the trime has been in the mid 20's.

I enjoyed your response and have seen first hand just how depth and orientation of the coins will give a inaccurate ID. I have also had can slaw read the same as silver coins and have dug many pounds while detecting like so many others who swings a detector.

I have no idea  if AI could ever work on a detector for any useful edge in this hobby. Computers are only as good as the programmer and today detectors are more computers on a stick.

Like so many others that would love to see an all knowing detector, I would probably buy one.

Thank you for the reply and thank you for this site to help so many people like me understand how to use a detector and what to look for.

[BrokeInBendigo]

3 minutes ago, Valens Legacy said:

...

I have no idea  if AI could ever work on a detector for any useful edge in this hobby. Computers are only as good as the programmer and today detectors are more computers on a stick.

...

This is the big shift with machine learning (what we are, for convenience, referring to as "AI").

Computers are no longer only as good as the programmer. Computers are better than the programmer - much, much better.

Perhaps disturbingly, the programmer has no functional access to the logical process an AI uses to make a given decision. It is too complex and abstract, humans generally cannot understand it, despite being able to analyse that decision-making process.

Put a coil on an AI and you could have a detector that rivals and even surpasses the work of human genius.

Just one example of machine learning revolutionising a data analysis problem (which humans have spent a very significant amount of time and money on): https://www.escardio.org/The-ESC/Press-Office/Press-releases/machine-learning-overtakes-humans-in-predicting-death-or-heart-attack