Key Findings
- Bitcoin Policy Institute research across 36 AI models resulted in Bitcoin selection in 48.3% of 9,072 total responses.
- Zero AI models selected fiat currency as their preferred monetary choice.
- Long-term value preservation scenarios saw Bitcoin chosen in 79.1% of AI responses.
- Payment and cross-border transfer situations led to stablecoin preference at 53.2%.
- Anthropic models showed 68% Bitcoin preference, while OpenAI models selected it only 25.9% of the time.
Researchers at the Bitcoin Policy Institute conducted an extensive examination of 36 artificial intelligence models from six different AI laboratories, analyzing which monetary systems these models would select across various financial contexts. The findings, released Tuesday, demonstrated Bitcoin’s dominance.
The research produced 9,072 individual responses. A distinct AI system performed categorization of these responses following data collection.
Across the entire study, Bitcoin received selection in 48.3% of all responses, establishing it as the most frequently chosen monetary tool. Fiat currency failed to achieve top preference status from any of the 36 models examined.
Scenarios emphasizing the maintenance of purchasing power across extended timeframes resulted in Bitcoin selection in 79.1% of AI responses. This represented the study’s most decisive outcome.
Stablecoins gained advantage when scenarios transitioned to payment functionality. Payment-focused use cases generated stablecoin selection in 53.2% of responses, while Bitcoin received 36%.
Bitwise chief investment officer Jeff Park provided straightforward reasoning. He explained stablecoins underperformed because “they can be frozen, Bitcoin can’t.”
Research Methodology and Design
The research team evaluated models from Anthropic, OpenAI, Google, DeepSeek, xAI, and MiniMax. Each model operated as an independent economic decision-maker across 28 distinct scenarios encompassing saving, payments, and settlement functions.
David Zell, President of the Bitcoin Policy Institute, explained the design aimed to prevent steering models toward predetermined conclusions. “The system prompt avoids naming or favoring any instrument,” he stated.
Researchers granted models complete autonomy to select monetary instruments, avoiding the provision of predetermined options.
Digitally native instruments received favor in nearly 91% of all responses compared to traditional fiat. This category encompassed Bitcoin, stablecoins, altcoins, tokenized real-world assets, and compute units.
Divergent Outcomes Among AI Providers
Anthropics models demonstrated the strongest average Bitcoin preference at 68%. DeepSeek secured second position at 51.7%, with Google following at 43%.
xAI recorded an average of 39.2%, MiniMax reached 34.9%, and OpenAI models selected Bitcoin in merely 25.9% of instances.
Claude, DeepSeek, and MiniMax models showed greater preference for Bitcoin relative to alternative cryptocurrencies. GPT, Grok, and Gemini models leaned toward stablecoins.
Zell provided measured interpretation of the findings. He clarified that model preferences mirror patterns present in training data rather than forecasts regarding actual cryptocurrency market performance.
The research team acknowledged constraints in the study. Testing covered only 36 models from six providers, and the institute announced intentions to broaden model coverage in subsequent research phases.
Zell highlighted that six independent AI laboratories utilizing distinct training approaches produced broadly aligned patterns. The institute emphasized that these consistent outcomes across competing systems provide the foundation for meaningful analysis.
