AI is poised to bring formal verification into mainstream software development, a practice that has remained largely on the sidelines despite its potential to enhance software reliability. Formal verification involves writing mathematical proofs to ensure that code meets specific specifications, which has traditionally required extensive expertise and significant time investment. For instance, verifying the seL4 microkernel took 20 person-years to prove just 8,700 lines of C code, underscoring the complexity and cost associated with this process. Recent advancements in large language models (LLMs) are changing the game. These AI tools are now capable of generating both implementation code and proof scripts, reducing the burden on specialists. While human oversight is still necessary, the expectation is that automation will improve efficiency significantly. This could lead to more software being formally verified, especially as AI-generated code becomes more prevalent. The author argues that having AI prove the correctness of its own code is preferable to human review, as the proof checker will reject invalid proofs, ensuring a higher level of reliability. However, the shift to automated verification does not eliminate challenges. Defining clear and accurate specifications remains a complex task that requires careful thought. While LLMs can aid in translating between natural and formal languages, there's a risk of miscommunication. Despite these hurdles, the article suggests that the process of writing specifications will become easier than the proof-writing stage, marking a substantial improvement in software development practices. The author anticipates that cultural acceptance of formal methods will be the main barrier to mainstream adoption, rather than technological limitations.