Superintelligence RSS

AI, Collective Intelligence, Intelligence, Superintelligence -

Rethinking AI benchmarks: A new paper challenges the status quo of evaluating artificial intelligence

Rethinking AI benchmarks: A new paper challenges the status quo of evaluating artificial intelligence

In recent years, artificial intelligence (AI) has made remarkable progress in performing complex tasks that were once considered the domain of human intelligence. From passing the bar exam and acing the SAT to mastering language proficiency and diagnosing medical images, AI systems such as GPT-4 and PaLM 2 have surpassed human performance on various benchmarks. Benchmarks are essentially standardized tests that measure the performance of AI systems on specific tasks and goals. They’re widely used by researchers and developers to compare and improve different models and algorithms; however, a new paper published in Science challenges the validity and usefulness of many existing benchmarks for evaluating AI systems. The...

Read more

AGI, AI Ethics, Governance, Superintelligence -

'Governing Superintelligence' - Synthetic Pathogens, The Tree of Thoughts Paper and Self-Awareness

'Governing Superintelligence' - Synthetic Pathogens, The Tree of Thoughts Paper and Self-Awareness

Two documents released in the last few hours and days, OpenAI’s ‘Governance of Superintelligence’ and DeepMind’s ‘Model Evaluation for Extreme Risks’, reveal that the top AGI labs are trying hard to think of how to live with, and govern, a superintelligence. I want to cover what they see coming. I’ll show you persuasive evidence that the GPT 4 model has been altered and now gives different outputs from two weeks ago. And I’ll look at the new Tree of Thoughts and CRITIC prompting systems that might constitute ‘novel prompt engineering’. I’ll also touch on the differences among the AGI lab...

Read more

Tags
#WebChat .container iframe{ width: 100%; height: 100vh; }