How Much Does ChatGPT Know—and Not Know—About Coins?

How Much Does ChatGPT Know About Coins and Gold?

We put artificial intelligence to the test with expert questions about numismatics and precious metals

Introduction

ChatGPT is a new artificial intelligence tool that seems to have caught everyone's attention. It is a chatbot powered by a large language model (LLM) that has generated both excitement and skepticism about AI's potential to transform various industries. Everybody seems to think AI is on the cusp of eliminating millions of jobs, but how accurate is this technology when it comes to specialized knowledge?

To cut through the hype, we decided to put ChatGPT to the test in our area of expertise: coins and precious metals. We asked the AI several questions spanning from basic numismatics to complex topics about gold pricing and silver market dynamics. The questions ranged from testing logical coherence to philosophical musings about the monetary system.

The results were fascinating—while ChatGPT provided some impressive correct information, it also delivered several glaringly incorrect answers. For investors who purchase gold or invest in silver, understanding where AI succeeds and fails in precious metals knowledge is particularly relevant.

ChatGPT's Performance Summary

6
Correct Answers
3
Incorrect Answers

Out of 9 total questions testing ChatGPT's knowledge of coins and precious metals

Table of Contents

How ChatGPT Works

ChatGPT was developed by OpenAI, which started as an open-source project in 2015 with prominent seed investors including Elon Musk and Peter Thiel. As of now, OpenAI is no longer as open or transparent about how the software works, though we understand the basic mechanics.

Large Language Model (LLM)

ChatGPT is trained through machine learning by being fed massive amounts of text. It uses neural networks to extrapolate rules about human speech, grammar, and writing patterns.

Pattern Recognition

The AI can recognize patterns, understand relationships between words, memorize facts, and perform text-based tasks ranging from solving math problems to writing poetry.

Industry Applications

This technology is already being used across numerous industries, from writing functional code to potentially disrupting the online search business dominated by Google.

Rapid Evolution

The technology is improving rapidly, with newer versions like ChatGPT-4 showing significant improvements over earlier iterations in handling complex questions and avoiding logical errors.

Testing Methodology

We tested ChatGPT with questions spanning from basic numismatic facts to complex precious metals market dynamics. Some questions were designed to test factual knowledge, while others explored the AI's ability to handle nuanced topics about gold valuation and monetary policy. Our experiment used both earlier and newer versions of the AI to compare improvements over time.

What AI Got Wrong: Critical Failures Exposed

AI chatbots, including ChatGPT, still have certain quirks and bugs that have been well-documented. They have been known to "hallucinate" and provide clearly incorrect logic. When we tested earlier versions of ChatGPT, we encountered several concerning failures in basic numismatic knowledge.

1944 Penny Information

ChatGPT's Error: Claimed that the government melted down copper pennies in 1944 despite saying the copper supply had been replenished. Also incorrectly stated that the U.S. Mint made fewer one-cent coins in 1944.

The Reality: This information contains internal contradictions and factual errors about 1944 penny production and the wartime metal situation.

U.S. Dollar Value Trend

ChatGPT's Error: Earlier versions claimed that the U.S. dollar has generally appreciated in value over time—the exact opposite of what has actually happened.

The Reality: The U.S. dollar has lost approximately 95% of its purchasing power since 1933, making this one of the most fundamental errors possible about monetary history.

Buffalo Nickel Fabrication

ChatGPT's Error: Provided an entirely fabricated explanation for why Buffalo nickels sometimes have no visible date, then doubled down with additional false information when questioned.

The Reality: The AI created completely fictional explanations and even cited real reference books that didn't contain the claimed information. When confronted, it generated additional false information.

The "Hallucination" Problem

Perhaps most concerning was ChatGPT's tendency to fabricate information when it didn't know the answer, particularly about Buffalo nickels. Rather than admitting uncertainty, the AI created plausible-sounding but entirely false explanations. This highlights a critical limitation for anyone considering AI as a source for specialized knowledge about coins or precious metals investing.

These errors are particularly troubling because they weren't simply minor inaccuracies—they were fundamental misunderstandings about basic numismatic facts and economic principles. For investors relying on accurate information about gold market trends or silver valuations, such misinformation could lead to costly mistakes.

What AI Got Right: Impressive Accurate Responses

Despite the notable failures, ChatGPT demonstrated impressive knowledge in several areas, particularly when dealing with well-established historical facts and market dynamics. The newer ChatGPT-4 version showed significant improvements over earlier iterations.

Successful Responses Overview

First Coins in Human History
Correctly identified Lydian staters from 7th century BCE with accurate historical context
Most Valuable Canadian Coin
Accurately identified the 1911 Canadian silver dollar with proper context about rarity
Silver War Nickels
Provided comprehensive details about 1942-1945 nickels including compositions and mintmarks
Historical Use of Gold and Silver
Offered sophisticated analysis of why precious metals were chosen for coinage throughout history
Modern Coin Values vs. Historical
Accurately explained the decline in intrinsic value of modern coins compared to historical precious metal coinage
Platinum vs. Gold Pricing
Demonstrated good understanding of commodity market dynamics affecting precious metal valuations

Historical Knowledge

Question: What were the first coins in human history?

ChatGPT Response: Correctly identified Lydian staters from around 650-600 BCE, mentioning they were made from electrum (gold-silver alloy) and represented the world's first standardized currency.

This response aligned perfectly with scholarly consensus and demonstrated solid understanding of numismatic history.

Market Analysis

Question: Why is platinum cheaper than gold?

ChatGPT Response: Provided nuanced analysis covering supply and demand factors, industrial vs. investment demand differences, and market dynamics affecting both metals.

The AI demonstrated sophisticated understanding of precious metals markets relevant to investors tracking gold price movements.

Economic Principles

Question: Why were coins historically made of gold and silver?

ChatGPT Response: Comprehensive explanation covering durability, portability, divisibility, intrinsic value, and universal acceptance—hitting all the key monetary properties.

This response showed deep understanding of monetary theory and why precious metals remain relevant for modern investors.

Improvement Over Time

Notably, when we asked the same questions that earlier ChatGPT versions had failed, the newer ChatGPT-4 provided accurate answers. The AI corrected its previous errors about dollar value trends and 1944 penny information, suggesting rapid improvement in handling complex economic and numismatic topics. This is encouraging for those seeking AI assistance with gold investment research or silver market analysis.

Analysis: Strengths and Limitations Revealed

Our experiment revealed distinct patterns in ChatGPT's performance that offer insights into both the potential and limitations of AI for specialized knowledge domains like numismatics and precious metals.

Factual Knowledge Strength

ChatGPT excelled at well-documented historical facts, standard numismatic information, and established economic principles. Its knowledge of coin series, dates, and basic market dynamics was generally reliable.

Pattern Recognition

The AI demonstrated ability to connect related concepts, such as linking historical monetary policy to modern precious metals investing, showing sophisticated pattern recognition capabilities.

Leading Question Vulnerability

Earlier versions were susceptible to leading questions that contained false assumptions, such as asking "why" something was rare when it might not actually be rare.

Fabrication Risk

When encountering knowledge gaps, the AI sometimes fabricated plausible-sounding but entirely false information rather than admitting uncertainty.

Knowledge Area AI Performance Reliability Level Investment Relevance
Historical Facts Excellent High Context for precious metals history
Basic Numismatics Good High Coin collecting fundamentals
Market Dynamics Good Medium-High Gold and silver price factors
Economic Principles Excellent High Understanding precious metals role
Specialized Details Variable Low-Medium Specific coin varieties/grades
Current Market Data Limited Low Real-time price information

Critical Limitations

One reason ChatGPT-3.5 failed where ChatGPT-4 succeeded is that the newer version handled leading questions better. The older software seemed to assume that assumptions built into questions must be correct—if asked why a coin was rare, it assumed the coin must be rare. This has important implications for investors seeking AI assistance with gold market analysis or silver price research.

What AI Does Well

  • Synthesizing established knowledge from multiple sources
  • Providing comprehensive overviews of complex topics
  • Explaining economic and monetary principles clearly
  • Connecting historical context to modern applications
  • Remembering conversation context for follow-up questions

Where AI Struggles

  • Specialized technical details and edge cases
  • Admitting uncertainty rather than fabricating answers
  • Handling questions with false assumptions
  • Real-time market data and current pricing
  • Nuanced judgment calls requiring human expertise

Implications for Collectors and Investors

Our ChatGPT experiment offers valuable insights for precious metals investors and coin collectors considering AI as a research tool. While the technology shows promise for certain applications, it also reveals critical limitations that users must understand.

Research Assistant Potential

AI can serve as an effective starting point for research into precious metals history, basic market principles, and general numismatic knowledge, especially for newcomers to gold investing.

Verification Essential

All AI-generated information should be verified through authoritative sources, particularly for investment decisions or specialized collecting choices involving significant financial commitments.

Education Tool Value

ChatGPT excels at explaining complex economic concepts, making it valuable for understanding why precious metals maintain value and their role in modern portfolios.

Human Expertise Irreplaceable

For critical decisions about rare coins, market timing, or specialized investments, human expertise and current market knowledge remain essential.

Best Practices for AI Use

When using AI for precious metals research, treat it as a starting point rather than a final authority. Use it to understand basic concepts and generate research directions, but always verify information through authoritative sources. For current gold prices and silver market data, rely on real-time financial sources rather than AI responses.

For New Investors

AI can help newcomers understand:

  • Basic precious metals investment principles
  • Historical context for gold and silver markets
  • Fundamental economic concepts affecting prices
  • Overview of different investment options

However, specific investment advice and current market analysis should come from human experts and real-time data sources when you're ready to start investing.

For Experienced Collectors

Seasoned numismatists might find AI useful for:

  • Research assistance on unfamiliar series
  • Educational content creation and explanation
  • Cross-referencing general information
  • Exploring connections between different collecting areas

But specialized knowledge, authentication, and valuation should always rely on human expertise and established references.

The Future of AI in Precious Metals

As demonstrated by the improvements from ChatGPT-3.5 to ChatGPT-4, AI technology is rapidly evolving. Future versions may become more reliable for specialized knowledge while maintaining their strengths in synthesis and explanation. However, the fundamental need for verification and human expertise in investment decisions is likely to remain, particularly for significant financial commitments in gold and silver markets.

The technology shows promise for democratizing access to basic precious metals education and making complex economic concepts more accessible. However, our experiment reinforces that successful investing in coins and precious metals still requires human judgment, current market knowledge, and the ability to verify information through authoritative sources.

Conclusion: AI Is a Great Tool, But Isn't Close to Replacing Human Intelligence

Our experiment testing ChatGPT's knowledge of coins and precious metals revealed both impressive capabilities and concerning limitations. While the AI demonstrated sophisticated understanding of established historical facts, economic principles, and general market dynamics, it also made fundamental errors and occasionally fabricated information when facing knowledge gaps.

The improvement from ChatGPT-3.5 to ChatGPT-4 shows this technology is rapidly evolving. The newer version corrected previous errors about dollar value trends and handled leading questions more appropriately, suggesting continued advancement in AI reasoning capabilities. However, critical limitations remain that make human expertise irreplaceable for serious investment decisions.

Expert Perspective

As linguist Noam Chomsky noted, "The human mind is not, like ChatGPT and its ilk, a lumbering statistical engine for pattern matching... the human mind is a surprisingly efficient and even elegant system that operates with small amounts of information; it seeks not to infer brute correlations among data points but to create explanations." This distinction becomes crucial when making investment decisions about gold purchases or silver investments.

For precious metals investors and coin collectors, AI serves best as a research assistant and educational tool rather than a replacement for human expertise. It excels at explaining basic concepts, providing historical context, and synthesizing established knowledge—making it valuable for newcomers learning about gold market fundamentals or silver price dynamics.

However, for critical decisions involving authentication, valuation, market timing, or significant financial commitments, human expertise remains essential. The technology's tendency to fabricate plausible-sounding but incorrect information when facing knowledge gaps makes verification through authoritative sources absolutely critical.

Moving Forward

We shouldn't fear AI technology, but we should remain conscious of how it's applied—an issue known in the AI industry as "alignment." For precious metals investing, this means using AI as a starting point for research and education while relying on human expertise, current market data, and authoritative sources for actual investment decisions. The combination of AI's synthesis capabilities with human judgment and verification creates the most powerful approach to precious metals research and investment.

As AI continues to improve, it will likely become an even more valuable tool for education and research in numismatics and precious metals. However, our experiment confirms that successful investing in these markets still requires the human elements of judgment, creativity, and the ability to navigate uncertainty—qualities that current AI technology cannot replicate, regardless of its impressive pattern-matching capabilities.

Posted In: blog
Login to post comment Login