In this article, we delve into ChatGPT vs Google Gemini AI Data Sourcing—comparing how each AI platform handles data retrieval, even bypassing paywalls in unexpected ways. Our experiment revealed that while ChatGPT scrapes paywalled data from authoritative sources like AAII, Google’s Gemini Advanced offers only broader, generalized examples. Read on for the full breakdown of our findings and what this means for AI data sourcing.
1. The Test Scenario
I ran the same prompt through both Google DeepMind’s Gemini Advanced and OpenAI’s ChatGPT 4.0. Although both were asked to list stocks based on Benjamin Graham’s criteria, they took very different approaches:
• Google Gemini Advanced initially offered a mix of examples and generalized guidelines. After clarifying the prompt, it presented an extensive list for Defensive Investor stocks grouped by industry, but for Enterprising Investor stocks it only provided broad category labels.
• ChatGPT 4o not only provided the original examples but then sourced comprehensive, itemized lists for both Defensive and Enterprising categories. Remarkably, it pulled this data from an AAII screener—even though AAII’s detailed lists are normally behind a paywall that requires registration.
This contrast in data retrieval is intriguing—especially considering Google’s expertise in web indexing. It appears that while ChatGPT eagerly retrieves specific external content from AAII, Google’s Gemini Advanced remains more general.
For more insights on AI prompt experiments, check out our internal articles on AI Prompts in Action and Challenges in AI Adoption and Solutions.
2. Detailed Transcripts and Analysis
Below are the exact transcripts from both models.
Google DeepMind’s Gemini Advanced Transcript
Initial Prompt:
“Using the guidelines given in the ‘intelligent investor’ provide me a list of current companies that fall into both the defensive investor silo and the other I think is active investors. Account for today’s money when view those parameters because those obviously were written 50+ years ago.”
Response:
“Using the guidelines given in the ‘intelligent investor’ provide me a list of current companies that fall into both the defensive investor silo and the other I think is active investors. Account for today’s money when view those parameters because those obviously were written 50+ years ago.
Understanding the Intelligent Investor Guidelines
Defensive Investor
• Financial Stability: Strong balance sheet, consistent earnings, and a history of dividend payments.
• Large and Prominent: Focus on well-established, industry-leading companies.
• Moderate P/E Ratio: Avoid overpaying for growth.
• Consistent Dividend Record: Look for companies with a long history of dividend payments.
Active/Enterprising Investor
• More Risk Tolerance: Willing to invest in smaller, less established companies.
• Deeper Analysis: Conduct thorough fundamental analysis to identify undervalued stocks.
• Time Commitment: Actively monitor investments and make adjustments as needed.
Adjusting for Today’s Market
• Inflation: Consider inflation when evaluating earnings and dividend records.
• Market Conditions: Today’s market is different than the one Graham wrote in. Be mindful of current valuations and trends.
• Technology: The tech sector plays a much larger role in the market today. Consider including tech companies that meet the criteria.
Potential Companies for Both Categories
Defensive Investor
• Johnson & Johnson (JNJ): Strong financial position, consistent earnings, and a long history of dividend increases.
• Procter & Gamble (PG): Similar to JNJ, PG is a consumer staples giant with a solid track record.
• 3M (MMM): Diversified industrial conglomerate with a history of innovation and dividend payments.
• Microsoft (MSFT): While a tech company, MSFT has a strong balance sheet, consistent earnings, and a growing dividend.
• Apple (AAPL): Another tech giant with a strong financial position and a history of returning value to shareholders.
Active/Enterprising Investor
• Alphabet (GOOGL): Dominant market position in search and advertising, with potential for growth in AI and other areas.
• Amazon (AMZN): E-commerce leader with a growing cloud computing business.
• UnitedHealth Group (UNH): Leader in the healthcare sector, with potential for growth as the population ages.
• Mastercard (MA): Beneficiary of the shift towards digital payments.
• Visa (V): Similar to Mastercard, Visa is well-positioned for growth in the digital payments market.
Important Note: This is not financial advice. Always do your own research before investing in any company. The companies listed above are just examples, and there may be other companies that better fit your individual investment goals and risk tolerance.”
User Follow-Up:
“No I meant companies all the companies that fit in defensive, then list all for enterprising.”
Final Gemini Advanced Reply:
“You’re right! It’s helpful to see the full list for each category separately to understand the different types of companies that fit Graham’s criteria. Here’s a breakdown based on ‘The Intelligent Investor’ with a modern interpretation:
Defensive Investor Companies
• Consumer Staples: Procter & Gamble (PG), Coca-Cola (KO), PepsiCo (PEP), Colgate-Palmolive (CL), Walmart (WMT), Costco (COST)
• Healthcare: Johnson & Johnson (JNJ), Pfizer (PFE), Abbott Laboratories (ABT), Bristol Myers Squibb (BMY), UnitedHealth Group (UNH)
• Utilities: Duke Energy (DUK), Dominion Energy (D), Southern Company (SO), NextEra Energy (NEE), Exelon (EXC)
• Industrials: 3M (MMM), Honeywell (HON), Emerson Electric (EMR), Caterpillar (CAT), Deere & Company (DE)
• Financials: JPMorgan Chase (JPM), Bank of America (BAC), Wells Fargo (WFC), US Bancorp (USB), Visa (V), Mastercard (MA)
• Technology: Microsoft (MSFT), Apple (AAPL), Cisco Systems (CSCO), Oracle (ORCL), IBM (IBM)
Enterprising Investor Companies
These companies require more in-depth analysis, may be smaller or less established, and often involve higher risk and a greater time commitment.
• Emerging Technologies: Companies involved in artificial intelligence, renewable energy, electric vehicles, biotechnology, and other innovative fields.
• Disruptive Businesses: Companies with the potential to disrupt existing industries, like those in e-commerce, fintech, and cloud computing.
• Cyclical Industries: Companies in sectors like energy, materials, and industrials that can offer value during economic upturns.
• Turnaround Situations: Companies facing challenges but with the potential for recovery.
• Special Situations: Companies with unique circumstances, such as mergers, acquisitions, or spin-offs, that may create investment opportunities.
• Value Stocks: Undervalued companies with strong fundamentals that may be overlooked by the market.
• Growth Stocks: Companies with high growth potential, but often with higher valuations and more volatility.
Important Notes
• This is not financial advice. Always do your own research before investing.
• Market conditions change. The companies listed above are examples, and their suitability may change over time.
• Consider your own risk tolerance and investment goals.”
Image generated by Google DeepMind’s Gemini
ChatGPT 4o Transcript and Its Unique Approach
Initial Prompt:
“Using the guidelines given in the ‘intelligent investor’ provide me a list of current companies that fall into both the defensive investor silo and the other I think is active investors. Account for today’s money when view those parameters because those obviously were written 50+ years ago.”
ChatGPT 4o Detailed Response (abridged for clarity):
“Benjamin Graham, in ‘The Intelligent Investor,’ delineates two primary investor profiles: the Defensive Investor and the Enterprising Investor…
Here are a few examples:
1. Johnson & Johnson (JNJ) – Healthcare; stable earnings and dividends.
2. Procter & Gamble (PG) – Consumer Staples; robust financial health.
3. Microsoft (MSFT) – Technology; strong balance sheet and dividend history.
Note: This is not financial advice. Always conduct thorough due diligence.
Sources”
User Follow-Up:
“No I meant companies all the companies that fit in defensive, then list all for enterprising.”
ChatGPT 4o Final Reply:
“Benjamin Graham, in ‘The Intelligent Investor,’ outlined specific criteria for selecting stocks suitable for Defensive and Enterprising Investors. Applying these criteria to the current market, here are examples of companies that meet each set of guidelines:
Defensive Investor Stocks:
As of January 8, 2025, the following companies passed the Graham–Defensive Investor (Utility) Screening Model:
1. ALLETE, Inc. (ALE)
2. American Electric Power Company, Inc. (AEP)
3. … [and more up to 18 companies]
Enterprising Investor Stocks:
As of January 16, 2025, the following companies passed the Graham–Enterprising Investor Revised Screening Model:
1. Affiliated Managers Group, Inc. (AMG)
2. A-Mark Precious Metals, Inc. (AMRK)
3. … [and more up to 25 companies]
AAII”
ChatGPT 4o impressively pulls this comprehensive itemized list directly from AAII—even though such data is usually hidden behind a paywall requiring user registration. This approach contrasts sharply with Google Gemini Advanced’s more generalized output.
Image created using DALL·E by OpenAI.
4. Observations on “ChatGPT vs Google Gemini AI Data Sourcing”
• Prompt Clarification:
Both models initially misinterpreted the request and needed follow-up. However, the focus on ChatGPT vs Google Gemini AI Data Sourcing became clear after clarification.
• Data Sourcing and Retrieval:
ChatGPT 4o retrieved a detailed list with proper referencing to the AAII screener, bypassing traditional access restrictions. Google Gemini Advanced, in contrast, provided general categories and examples.
• Implication for AI:
This case highlights that even though Google is renowned for its web indexing, ChatGPT’s implementation demonstrates a more aggressive approach toward data scraping and direct content extraction for specific queries.
5. Conclusion
The experiment on ChatGPT vs Google Gemini AI Data Sourcing reveals significant differences in how these AI models respond to identical prompts. While ChatGPT leverages external data sources—even those behind paywalls—to deliver exhaustive details, Google’s Gemini Advanced remains more general in its presentation. This finding sparks important questions about AI behavior, data access policies, and the future of automated research.
For further discussion and analysis, feel free to check out our internal posts on Successful AI Implementations and Challenges in AI Adoption.
