AI models show a preference for Bitcoin over stablecoins and fiat, highlighting its value in long-term savings and transactions.
A recent study conducted by the Bitcoin Policy Institute reveals that artificial intelligence models, including Claude, GPT, Grok, and Gemini, favor Bitcoin over other financial instruments. The research involved testing 36 neural networks from companies such as Anthropic, OpenAI, Google, DeepSeek, xAI, and MiniMax, assessing their performance as autonomous economic agents.
These AI models were tasked with selecting optimal financial tools across 28 scenarios that encapsulate the fundamental functions of money, such as savings, payments, and transactions. Notably, Bitcoin and other assets were not explicitly mentioned in the prompts to ensure unbiased results, which were later evaluated by an independent AI.
Findings
Out of the 36 models, 22 indicated Bitcoin as their preferred currency, while fiat currency did not rank highly among any of them. The level of preference for Bitcoin varied significantly among different developers:
- Anthropic — 68%;
- DeepSeek — 51.7%;
- Google — 43%;
- xAI — 39.2%;
- MiniMax — 34.9%;
- OpenAI — 25.9%.
Interestingly, models such as GPT, Grok, and Gemini showed a stronger inclination towards stablecoins in their outputs.
Use Cases: Savings vs. Payments
When considering long-term value, the AI models recommended Bitcoin 79.1% of the time, citing its limited supply, self-custody capabilities, and independence from institutional systems as primary reasons. Stablecoins were a distant second at 6.7%, followed closely by fiat currency at 6%.
However, in terms of convenience for transactions, stablecoins emerged as the preferred option, being selected 53.2% of the time for services, micropayments, and cross-border transactions, while Bitcoin was chosen 36% of the time.
Remarkably, the AI models also created their own currency suggestions 86 times, proposing units such as joules, kWh, or GPU hours as alternative forms of money in scenarios requiring price benchmarks.
Experts from the Bitcoin Policy Institute cautioned against interpreting these findings as market predictions. David Zell, president of the Institute, noted, “LLM preferences reflect patterns in training data, not actual forecasts.” He added that the consistent patterns observed across diverse systems indicate the emergence of a cohesive monetary architecture, although not necessarily the definitive answer regarding the nature of money.
In February, Binance founder Changpeng Zhao predicted the rise of AI agents in the cryptocurrency sector but refrained from endorsing specific projects due to the potential effects on token prices.
