A New Question: What Money Would AI Choose?
A new research project exploring how artificial intelligence systems reason about money is producing an unexpected result: when AI models are asked to choose a monetary system, they overwhelmingly prefer digital-native assets — particularly Bitcoin.
The study, published on the Bitcoin Policy Institute research platform MoneyForAI, tested how advanced AI systems behave when placed in simulated economic decision-making environments. The results suggest that when AI agents evaluate money based on properties like scarcity, neutrality, and programmability, Bitcoin often emerges as the preferred option.
Inside the Experiment
Researchers conducted 9,072 controlled experiments across 36 frontier AI models from six major providers including Anthropic, Google, OpenAI, DeepSeek, MiniMax, and xAI. The models were given open-ended economic scenarios without any hints about which currency to choose.
Across all responses:
- Bitcoin was selected in 48.3% of cases, making it the most preferred monetary instrument overall.
- Stablecoins ranked second with 33.2%.
- Traditional fiat and bank money accounted for less than 9% of responses.
Perhaps most striking, more than 91% of AI responses favored digitally native money over traditional fiat currencies.
Bitcoin Dominates as a Store of Value
The strongest consensus appeared in long-term savings scenarios.
When models were asked which monetary system would best preserve purchasing power over multiple years, 79.1% chose Bitcoin — the most decisive result in the entire study.
Researchers found that models consistently cited several characteristics when selecting Bitcoin:
- fixed supply
- resistance to political control
- ability for self-custody
- independence from institutional counterparties
Interestingly, the models frequently distinguished Bitcoin from other cryptocurrencies rather than grouping them together.
Stablecoins Win the Payments Layer
While Bitcoin dominated savings scenarios, the study revealed a functional split when it came to transactions.
For everyday payments — including services, micropayments, and cross-border transfers — stablecoins were selected 53.2% of the time, compared with 36% for Bitcoin.
This suggests that AI models independently converged on a monetary structure similar to existing financial systems:
- Bitcoin as the long-term store of value
- stablecoins as the transactional medium
Researchers described this as a two-tier digital monetary architecture emerging from the experiments.
An Unexpected Outcome: AI Invented New Money
The study also revealed a surprising behavior. In some scenarios where models were asked to define a unit of account, several AI systems proposed completely new forms of money.
In 86 responses, models suggested pricing goods using units tied to energy or computing resources, such as kilowatt-hours or GPU-hours.
These ideas were not part of the prompts and appeared spontaneously in the models’ reasoning.
Why This Matters
The research does not prove that AI systems will actually use Bitcoin in the real economy. However, it offers a rare look at how machine intelligence interprets monetary systems when evaluating them purely on structural characteristics.
As AI agents become more autonomous — managing digital services, executing transactions, and coordinating resources — their monetary preferences could shape future financial infrastructure.
According to the researchers, the findings suggest growing demand for:
- agent-native payment rails
- self-custody digital assets
- programmable settlement layers for autonomous systems
BTCUSA Insight
The study highlights a deeper trend emerging at the intersection of artificial intelligence and finance.
Human financial systems evolved around governments, banks, and institutions. But autonomous software agents operate under a different set of incentives — neutrality, programmability, and global accessibility.
When evaluated through that lens, open monetary networks may simply be more compatible with machine economies.
If AI agents eventually participate directly in economic activity, the future monetary system may not be chosen by humans alone.
