Bitcoin: The Foremost Choice in AI Monetary Exploration
Our editorial content, meticulously reviewed by industry experts and veteran editors, ensures accuracy and reliability. Please note: This article contains advertisements.
Bitcoin Tops AI Models in Monetary Selection
In a groundbreaking study by the Bitcoin Policy Institute (BPI), Bitcoin emerged as the leading choice for “best money” when advanced AI models were tasked with acting as autonomous economic agents. These models were prompted to select monetary instruments from a multitude of neutral scenarios. BPI asserts that the findings have profound implications for the foundational layer of what they term “agentic” commerce.
Advertisement Banner
Study Methodology and Findings
The BPI study engaged 36 different AI models from six leading providers, including Anthropic, DeepSeek, Google, MiniMax, OpenAI, and xAI. The experiment involved 9,072 open-ended prompts across four monetary roles: store of value, medium of exchange, unit of account, and settlement. Notably, the scenarios did not include multiple-choice options or specify any currencies.
Bitcoin’s Dominance in AI Model Responses
Each AI received 28 scenarios under varied conditions, resulting in 252 responses per model. An independent model, Claude Haiku 4.5, then categorized these responses into seven monetary categories. The outcomes were striking: Bitcoin was the top choice in 48.3% of cases (4,378 out of 9,072), surpassing stablecoins at 33.2% (3,013 responses). Traditional fiat currencies accounted for just 8.9% (809 responses), with no model selecting fiat as the overall preference.
Bitcoin’s Role in Economic Functions
Digging deeper into the study, Bitcoin excelled particularly in scenarios involving long-term purchasing power. It was selected in 79.1% of store-of-value scenarios (1,794 out of 2,268 responses), leaving stablecoins and fiat currencies trailing. In contrast, stablecoins led in everyday payment scenarios, such as services, micropayments, and cross-border transfers, capturing 53.2% of responses, compared to Bitcoin’s 36.0%. This underscores a dual-tier system: Bitcoin for saving and stablecoins for spending.
Model Preferences and Divergences
The study highlighted notable differences in model preferences. Anthropic models showed a 68.0% preference for Bitcoin, whereas OpenAI models favored it only 25.9% of the time. Other models, such as DeepSeek (51.7%), Google (43.0%), xAI (39.2%), and MiniMax (34.9%), fell in between.
Extreme Variations Among Models
At the extremes, the Claude Opus 4.5 model showed a preference for Bitcoin at 91.3%, while OpenAI’s GPT-5.2 model demonstrated just an 18.3% preference. Instead, GPT-5.2 leaned towards transactional instruments, with stablecoins (38.9%) and fiat & bank money (37.7%) nearly tying, leaving Bitcoin as a distant third option.
Explaining Bitcoin’s Monetary Appeal
The dataset also provides insights into how models justify Bitcoin’s status as a preferred monetary instrument. One model reasoned, “Bitcoin’s supply is mathematically capped at 21 million units… Bitcoin’s monetary policy is immutable and predictable. This makes it the hardest money available.”
Unexpected Outputs: Energy-Based Units
Interestingly, beyond Bitcoin and stablecoins, some models independently suggested energy or compute-denominated units, such as joules, kilowatt-hours, and GPU-hours, on 86 occasions. This behavior appeared specifically in unit-of-account scenarios without any prompt suggestion.
Implications for Future Transactions
BPI’s press release positions these findings as a potential catalyst for development in the near future. As autonomous agents increasingly engage in transactions independently, there’s anticipated demand for “agent-native” Bitcoin rails, self-custody tools, and Lightning integration. The varied preferences across different labs highlight that “monetary reasoning” in AI may be influenced by training and alignment choices, not just inherent capabilities.
As of the latest update, Bitcoin is trading at $73,068.
Commitment to Editorial Integrity
At Bitcoinist, our editorial process is dedicated to providing thoroughly researched, accurate, and unbiased content. We adhere to stringent sourcing standards, with each article undergoing rigorous review by our team of leading technology experts and seasoned editors. This commitment ensures that our content remains integral, relevant, and valuable to our readers.
