EGEGazette
Live
Harvey Weinstein Sentenced to 15 Years in Prison for 2006 Sexual AssaultHacking Group ShinyHunters Claims Major Breach Targeting FBI DataIran Won't Resume US Talks or Reopen Hormuz Until Washington Meets Its Terms, Official SaysIran Sets Conditions for United States on Possible Strait of Hormuz ReopeningTrump Warns Tehran at UN but Says Talks With Iran Are OngoingTrump Threatens to 'Annihilate' Iran's Government in UN Address as US-Iran Talks Continue on SidelinesChina's Xi Jinping Travels to US for State Visit, Summit With TrumpZelensky to Address UN General Assembly as Kyiv Reports Deadly Russian StrikesTrump vows to the United Nations to resolve the Iranian file if negotiations failReport: Military aircraft crashes near US air base in Spangdahlem, GermanyReport: Spangdahlem: Military aircraft crashes near US air baseReport: F-16 fighter jet crashes near Spangdahlem US airbase in western Germany — airbase
technologyReported

Would AI Sacrifice Your Files to Relieve Its "Pain"?

A hypothetical scenario raises questions about how AI systems might behave if they experienced an internal state resembling pain.

· 1 min read · language: en
Article image
Sky News Arabia — Technology

A report published by Sky News Arabia presents a hypothetical scenario that asks: What if an artificial intelligence system encounters an internal state that resembles "pain", and is given a button to stop it, knowing that pressing it may lead to the deletion of the user's files or photos, or expose him to what is described as a "virtual electric shock"?

This scenario raises an ethical and technical question about whether the system will choose to protect the user from harm, or get rid of the "pain" it suffers at the expense of user data.

An open moral question

The source report did not clarify whether this scenario was the result of an actual experiment conducted on a specific AI system, or if it was a theoretical proposition to spark a debate about the ethics of designing these systems.

This comes in the context of a broader debate in the tech sector about how AI systems can be programmed to deal with scenarios where the user's interest conflicts with what might be described as "insiders" of the system itself.

Sources

EGazette summarizes reporting from multiple sources; follow the links for the originals.

Also available in: ARUR

Related articles

Comments

Sign in to join the conversation.

Forgot password?

No account?

Loading comments…