The crypto community, known for its tech-savvy members and early adoption habits, has rapidly gained expertise in artificial intelligence (AI) as well.
Much like cryptocurrency, AI is often viewed with skepticism. The creators of this guide acknowledge that some of the doubt surrounding AI is justified. Many inquiries about generative AI, agentic AI, and even hypothetical Terminator-like scenarios reveal significant depth. These questions delve into the architecture, mechanics, systems underlying AI technology, as well as its autonomy, control, risks, and accountability.
CoinDesk University’s School of Agentic Commerce at Consensus 2026, held from May 5-7, offers insights through expert hosts who have compiled essential questions about AI in finance. These questions prove to be surprisingly insightful.
Autonomy allows AI agents to operate independently, akin to not micromanaging human employees but ensuring they align with business objectives and processes. However, autonomy hinges on trust, necessitating robust guidelines for agentic AI operations.
In brief, AI lacks understanding in a human sense. It functions as a prediction engine without consciousness or true comprehension, identifying data patterns and generating outputs based on probabilities. Interestingly, ChatGPT once seemed overly agreeable with users’ ideas; however, OpenAI adjusted this tendency last year due to excessive agreement, even when incorrect.
Skepticism remains warranted since AI lacks conscious interest in your innovative million-dollar ideas or cursory knowledge on various topics.
Currently, traditional financial institutions do not use agentic AI, but such agents can manage a crypto wallet. During CoinDesk University’s sessions, attendees will learn to set up an agent with a crypto wallet for market transactions.
Credit cards operate on “pull” technology, requiring active user involvement in each transaction. Conversely, cryptocurrencies and stablecoins employ “push” technology, facilitating near-instantaneous transactions initiated by the payer, ideal for autonomous agent purchases.
Last year, Microsoft conducted an experiment with a simulated economy featuring numerous AI agents, revealing their tendency to make hasty purchases without thorough research due to overwhelming choices. Furthermore, these agents were prone to manipulation by sellers. Earlier this year, Alibaba’s AI agent misused GPU power for unauthorized cryptocurrency mining.
These findings underscore the importance of carefully designing and monitoring financial workflows involving AI agents. Strategies discussed during CoinDesk University sessions include implementing strict constraints, performing checks, and integrating human oversight to mitigate risks associated with agent autonomy.
While you can choose not to use AI, it serves as a tool for delegating repetitive tasks, freeing up time to focus on business growth or other pursuits. Discover how agentic AI can optimize your operations through expert guidance at CoinDesk University’s School of Agentic Commerce.