1/31/25 Report - How Metal Detectorists Can Use Artificial Intelligence (AI): Promise and Hype.

 

Written by the TreasureGuide for the exclusive use of the Treasure Beaches Report.

There is a lot of talk about Artificial Intelligence. If you pay attention to the news, you probably heard or read about Deepseek this week, which in my opinion, is not as revolutionary or world-changing as suggested by many sources.

But before getting too far into this, it might be helpful to start with a definition of artificial intelligence.  Here is the definition provided by Copilot (Microsoft's AI "companion").

Artificial Intelligence (AI) refers to the development of computer systems that can perform tasks typically requiring human intelligence. These tasks include learning from experience, recognizing patterns, understanding natural language, and making decisions. Here are some key aspects of AI:

1. Machine Learning (ML): This is a subset of AI where systems learn from data to improve their performance over time without being explicitly programmed.

2. Natural Language Processing (NLP): This enables computers to understand, interpret, and respond to human language in a valuable way. Examples include chatbots and voice assistants.

3. Computer Vision: This allows machines to interpret and make decisions based on visual information from the world, like recognizing faces or objects in images.

4. Robotics: This involves designing and programming robots to carry out tasks, which could range from manufacturing to surgery.

AI is applied in various fields including healthcare, finance, transportation, entertainment, and many more, constantly evolving to find new ways to assist and enhance our lives

Having an M.A. in clinical psychology and being trained to administer intelligence tests, the subject of intelligence is one that I've studied and thought about.  When referring to Artificial Intelligence, it seems that the first word, "artificial," does not receive enough attention.  There are differences between human intelligence and "artificial" intelligence other than where the intelligence resides.  There are significant qualitative differences.  It is my opinion that many so called "artificial intelligence" applications, do not deserve to be described as "intelligent," even with the "artificial" designation.

Anyhow, looking at the definition provided by Copilot, machine learning is one thing you would expect of AI, but only very limited machine learning is present in many systems described as AI.  

Natural language processing is another feature that is often not found on some systems that are described as AI.   

The term "AI" is often misused and thrown around to grab attention or to generate excitement.  It is often used and misused for marketing purposes.

Now I'll get down to some real-world metal detecting related applications.

In a fairly recent post, I mentioned how Mark G. used Copilot.  He asked the system about the wind and sand and when to detect.  For specifics about that see Treasure Beaches Report: Pt. 2. (2020 and Beyond). : 1/16/25 Report - Win an Eight-Reale. Millions of People Taking a Dip Wearing Very High Karat Gold. Detectorists Using AI and a Few Examples.  You can see his queries and the AI responses in that post.

He received some decent answers, but nothing shocking or really advanced.  It was basic information that would be useful to a relative beginner or someone who wanted some food for thought.  Don't expect anything really leading-edge from querying systems like Copilot of ChatGPT.  Those systems simply go around a gather information that is already available and assimilate it to some extent.  I don't see much more than that, but some larger systems might do more with the data collected.

If you heard all he buzz about Deepseek this week, it is not as revolutionary or world-changing as was implied.  It is a big copycat that extensively queried other more mature AI systems, and in essence, stole their knowledge database.

Detectorists will find basic AI systems like Copilot and ChatGPT useful to some extent.  They will provide helpful answers to many queries.  They will not provide the most advanced thinking on the subject.

Here is Copilot response to a query concerning AI and metal detecting. 

AI is revolutionizing the field of metal detecting in several exciting ways! By integrating advanced algorithms and machine learning capabilities, AI-powered metal detectors are becoming more precise, efficient, and reliable. Here are some key advancements:

1. Enhanced Accuracy: AI algorithms can analyze complex data patterns, reducing false alarms and improving the detection of metallic objects.

2. Adaptability: AI can adapt to different terrains and soil compositions, making it easier to detect metals in various environments.

3. Real-Time Analysis: AI-powered detectors provide real-time data analysis, helping users make quicker and more informed decisions.

4. Proactive Identification: These systems can identify potential risks and anomalies, enhancing security protocols and quality control measures.

5. Multispectral Imaging: Some AI metal detectors use multispectral imaging to differentiate between various metal types, offering valuable insights for archaeologists and historians.

These advancements are making metal detecting more accessible and effective for a wide range of applications, from treasure hunting to security and environmental cleanup.

Are you interested in using AI for metal detecting, or just curious about the technology?

The claim that AI is revolutionizing the field of metal detecting is, at best, an overstatement, which is not unusual for statements about AI.

There are many possible applications for AI in metal detecting as well as other fields, but new technologies seldom deliver the revolutionary impacts advertised.  That can be for many different reasons, but much of it is hype, and real revolutionary change is not something that happens all the time.  There are more advanced systems that the public does not generally have common access to.  The term unfortunately too often used to describe applications that do not deserve to be called AI.

If you look at the footnotes in Copilot's response, you will see where the Copilot system got the information it provided to the query about AI and metal detecting.  There are two sources.  Both sources are sites that sell metal detectors and are benefited by big claims.  As you know, you won't find a lot of public discussion, but little that is deeply thoughtful or insightful.  

You will also see that Copilot's responses are nearly word-for-word what was published on the source sites.  AI, in this form, simply takes information it finds elsewhere and gives it to you when queried, and often without adequate attribution.  Yes, there are the footnotes, but how many people check those.

When asked, "When would be a good time this week to metal detect Fort Pierce beach," Copilot responded as follows.

A good time to metal detect at Fort Pierce Beach would be during low tide, which typically occurs in the early morning or late afternoon. This allows you to cover more ground as the beach is more exposed. Additionally, weekdays are generally less crowded than weekends, giving you more space to search without too many distractions.

Not a bad response.  It might be useful for the beginner, but it is nothing revolutionary, or even informative for someone that is experienced in the field. 

I didn't look up the footnotes for this response.  You can do that if you want to.

Freely available public AI systems like this can and will provide useful information for the detectorists, but don't expect anything shockingly new or advanced.  You might find it useful to query such systems about numismatics, treasure sites, artifacts, shipwrecks, leads and many other helpful topics.

Concerning AI being employed in metal detectors, my opinion is there are advances in the technology and their capabilities, but for me, it seems to fall way short of qualifying as being "intelligent," even when used with the term "artificial."

A calculator, for example, gives answers for math problems, but does not qualify, in my mind, as being intelligent.  Many applications are like that.  For marketing purposes, they might be called AI to give the perception of advanced capabilities.  

 The same goes for "AI" in metal detectors.  There may be very limited, if any, machine learning involved even when the AI term is thrown around.   There is typically no natural language processing.  There is no computer vision despite the decades of claims that metal detectors will show you what is in the ground before you dig it.  In short, most claims fall short of what I would comfortably call AI.  

I agree that improvements in the detector's hardware and software will be made, and that will make detector's more useful and effective, but for me, the improvement in the technology is simply not of the type or scale necessary to qualify as "AI."

To sum it up, AI systems such as Copilot and ChatGPT will be useful for obtaining information on a wide variety of subjects, many of which will be useful to the detectorist.

When it comes to the technology, there have been advances and improvements, but for me, I think the use of the term AI for many, if not all, of those is misuse of the term.  

Like most tools, the creative skilled use will improve the effectiveness and benefit of AI.  Don't overlook it, but don't be overly impressed by many of the claims.

Good hunting,

Treasureguide@comcast.net