Stanford researchers used AI to complete a formal software verification process in under half a day for under $500.
The development highlights shifting approaches to mathematical code correctness, alongside new tools like Harmonic's Aristotle and startup funding for universal language verification.
How they test software with mathematics
Formal verification is based on mathematical proofs and is considered difficult and time-consuming. Large companies and organizations, such as Boeing, Amazon and Apple, already use it for critical pieces of their software.
Companies are developing new tools
&Truth raised $70 million to develop tools that can verify mathematical programs written in any programming language.
Harmonic announced its “Aristotle” tool, which verifies software written in C and hardware designs written in Verilog. Meanwhile, Anthropic and Theorem announced a partnership to use artificial intelligence to detect bugs in open source software. They also announced a plugin for Claude.
Verification does not replace testing
Critics of the method argue that mathematical verification is no substitute for exhaustive software testing. Will Wilson of Antithesis says that both ways of checking are needed.
↳
READ ALSO
HACKS.GR / RELATED
AI Unlocked
Mintive raises $11 million to audit AI programs
↗
The technology is still in the prototype stage. It is not clear how businesses will use it.
G
HACKS.GR / GOOGLE SEARCH
Hacks.gr on Google
Set Hacks.gr as a preferred source.
Preferred source
↗
end of article
Top comments (0)