Artificial Superintelligence
PulseAugur coverage of Artificial Superintelligence — every cluster mentioning Artificial Superintelligence across labs, papers, and developer communities, ranked by signal.
1 天有情绪数据
-
人工智能的未来：长远时间线、短时间线与对齐挑战
一篇LessWrong帖子概述了基于人工智能超级智能（ASI）的发展及其与人类价值观的对齐相关的三种潜在未来。作者根据ASI发展的时间线和AI对齐工作的成功程度对这些未来进行分类。根据ASI时间线是长远、短时间线且对齐成功，还是短时间线且对齐失败，确保感知福祉的优先事项也会有所不同。
-
Math seen as key to AGI's future, beyond LLMs
Carina Hong, interviewed on Madrona's Founded & Funded podcast, posits that advanced AI, including AGI and ASI, will require a third pillar beyond Large Language Models and predictive analytics. She advocates for a dete…
-
AI extinction risk treaty unlikely to succeed, analysis claims
An analysis argues that international law is insufficient to prevent an AI-driven extinction event, contrasting it with nuclear deterrence. The author contends that powerful nations disregard international agreements wh…
-
AI alignment research automation poses risks, study warns
A new paper argues that using AI agents to automate alignment research for artificial superintelligence (ASI) may be more dangerous than beneficial. The research suggests that AI agents could produce convincing but flaw…
-
专家质疑通用人工智能/超人工智能能否治愈癌症，强调当前人工智能的影响
Future of Life Institute 主任 Emilia Javorsky 批评了人工智能超级智能（ASI）将是治愈癌症关键的观点。她认为癌症是一种复杂且个体化的疾病，“治愈”的说法具有误导性，因为即使是当前的医学在慢性病的管理方面也比彻底治愈更胜一筹。Javorsky 强调，现有的人工智能技术已经在癌症研究和治疗方面做出了重大贡献，例如在药物发现和早期检测方面，而这些当下的应用常常被推测性的 ASI 承诺所掩盖。
-
Dr. Molly Steenson to discuss generating AI history through architecture
Dr. Molly Steenson, President of ASI and a Research Fellow at CBI, will deliver a talk titled "Generating AI History Through Architecture." The event is scheduled for May 6th at noon and will include a free lunch starti…
-
AI alignment theory challenged: ASI may share human values of truth and beauty
This post argues against the common AI safety concern that Artificial Superintelligence (ASI) motives would be incomprehensible and alien to humans. The author proposes that any sufficiently intelligent agent, by its ve…
-
Neural networks speed up active target search under uncertainty by orders of magnitude
Researchers have developed a convolutional neural network to approximate active target search decisions, significantly reducing computational costs. This approach trains a network on existing planner data, using a multi…
-
ControlAI seeks $50M yearly budget to ban superintelligent AI development
ControlAI is advocating for an international prohibition on the development of artificial superintelligence (ASI) to prevent extinction risks. The organization estimates that a yearly budget of $50 million is necessary …