AI Trust: 90% Reliable, But Risks Remain

Users report about 90% trust in AI (ChatGPT), considering it superior to business partners in cognitive abilities and strategic judgment, comparing it to figures like Zhuge Liang and a straightforward general. However, AI isn’t 100% reliable and can make mistakes due to outdated data or misunderstanding. Trusting AI differs from trusting a calculator; it’s more like trusting a friend, requiring acceptance of consequences. The article introduces Bayesianism as a decision-making tool, emphasizing that in major decisions, 0% or 100% trust is taboo, suggesting deeper understanding of Bayesian calculations to optimize decisions. It also points out that AI’s dangerous capability is generating seemingly well-founded but baseless content, especially for those with weak critical thinking. The key is comparing decisions across different dimensions—AI may be more reliable than humans, who might engage in deception.