Mathematicians warn AI of threats to mathematics.

Message

This article has been viewed 868 times

Mathematicians warn AI of threats to mathematics.

Mathematicians have joined forces to publish the International Mathematics Alliance-supported manifesto, Leiden Declaration, warning AI to destroy mathematics by producing seemingly reasonable but unreliable or even erroneous evidence, weakening attribution, changing incentives, and giving technology companies too much influence on research priorities. Hundreds of people have signed the Declaration, which warns that the development of AI threatens the intrinsic value of mathematical research. The declaration begins by noting that it is very difficult to distinguish between the certificates produced by AI and the correct mathematical certificates, placing increasing pressure on the reviewer to generate the low-cost but costly validation of the AI papers, and that errors can be magnified if the follow-up study is based on a false premise. Second, AI ' s training is based on existing mathematical thesis, but it is often not correctly quoted when it is published, and copyright infringement issues are prevalent in AI model training. Thirdly, AI's incentives run counter to the values of mathematics. The declaration urged mathematicians to see AI as a tool, not a substitute for human responsibility. Mathematicians should make the use of AI public and take responsibility for the correctness of their work. The Declaration also warns that mathematics can be used in war, oppression, mass surveillance and the undermining of democracy, so that mathematicians should carefully weigh the ethical issues of cooperation with the scientific and technological industries.

https://leidendeclaration.ai/

https://science.slashdot.org/story/26/06/02/199206/mathmaticians-warn-of-ai-threats-to-policy-as-industry-encroaches

https://leidendeclaration.ai/

https://science.slashdot.org/story/26/06/02/199206/mathmaticians-warn-of-ai-threats-to-policy-as-industry-encroaches