I hope your right about the A.I.. The problem is I doubt any detector company is investing in it right now. It's still kind of new and training an A.I. to find and distinguish between metals along with determining good vs. bad targets would be a monumental undertaking. I don't think a ChatGPT large language model is going to help you in that endeavor. I see it being years before any detector company jumps on that bandwagon. A.I. will be ubiquitous in our daily lives before that happens. Meaning we will all probably have a personal robot to swing our detectors and dig our targets before we see an A.I. installed in a detector.
I think a lot of people don't understand the potential of AI. I have spent the last 40 years as a computer guy and even helped develop many applications probably being used by you today including A.I. In another discussion, I see comparing tones and VDI. Just as an example, yes, a computer algorithm would see a huge difference between a VDI of say 21 and 22. Tone-wise, there are many that most never hear. Lots of harmonics. AI could easily distinguish the many sub-tones/currents and determine how they factor into the possible ID of the target. It could see a 21.556 vs a 21.525 that you would most likely never hear, and the VDI will never show.
Take the Minelab Manticore 2D screen for example. Each pixel has a number. If that pixel number is sent a 1 instead of the default 0, it turns from white to black. When you run a target over the coil you may see a solid VDI of13 every time but that display doesn't just light up 1 pixel, it gets a "smear" because it is getting many returning frequencies being processed and turning on and off pixels as it processes the signal. Way to much information for the human to analyze, but AI could in realtime see thousands of changes in the frequency in realtime and make a much better informed decision on what it might be.
Most detectors take the returning signals after processing and instead of displaying a VDI of actual processed frequencies, they are compressed and rounded up or down to a simple 2 digit number missing a lot of information the audio could have. Unfortunately, many detectors also copress the audio down to a more simple tone missing a lot of useful information.
That is why I say the D2, and a few others, in my opinion, give the user more audio information than most detectors. You can choose to hear the processed signal as a compressed version if you set it to 2, 3, 4, or 5 tones. This is the normal for most other detectors. It compresses all the audio to a more simplified form. I suppose it is to make the tones more pleasant? On the D2 in Full Tones or Pitch, you get the full spectrum. It gives the audio that zippy tone which I hate, but realize I can use it to make a more informed decision on a target.
AI could do calculations, like seeing a target and then looking at all the surrounding possible influences on that target to make a much more informed decision on what it could be. Who knows yet what AI could possibly do for the EMI issues we face today? I would not be a bit surprised to find that AI could filter out EMI completely and behind the scenes that you would not even need to know about. Saltwater and highly mineralized soil a thing of the past? Yes, very possible.
I think AI's biggest possible contribution today would be a fully automated detector that even someone new to detecting could get the full potential right out of the box.
I see endless possibilities where AI could advance the hobby, but then I know what it can do after a lot of years working with it.