Search AI Products and News

The Arc Prize Foundation, responsible for maintaining and managing the competition, last week revised its cost estimate for OpenAI's upcoming o3 inference AI model with a staggering adjustment—from an initial estimate of $3,000 per ARC-AGI task to $30,000. This price correction reveals that the operational costs of today's most complex AI models may be ten times higher than previously anticipated. While OpenAI has yet to announce an official pricing strategy for o3, or even officially release the model, the Arc Prize...

Amazon Web Services (AWS) has announced the release of Ocelot, a new quantum computing chip that significantly reduces the cost of error correction by approximately 90%. This breakthrough promises to accelerate advancements in the field of quantum computing.

Microsoft's trajectory in the field of artificial intelligence is undergoing a significant transformation. Once ambitiously investing in self-developed large models, it now shifts its focus to collaboration with OpenAI and has launched a new 'Correction' feature aimed at addressing the 'hallucination' issues in AI-generated content. From self-development to leveraging partnerships: the evolution of Microsoft's AI strategy. Initially, Microsoft invested a substantial amount of resources in the field of generative AI (GenAI), hoping to create AI models comparable to or even surpassing OpenAI's GPT series. However, during the lengthy development process...

The Google DeepMind research team has recently achieved a significant breakthrough with the development of an innovative technology called SCoRe (Self-Correction through Reinforcement Learning). This technology aims to address the long-standing challenge of large language models (LLMs) struggling to self-correct, enabling them to identify and fix errors without relying on multiple models or external checks. The core of SCoRe's technology lies in its two-phase approach. The first phase optimizes the model.

In today's rapidly advancing world of artificial intelligence, while generative AI has brought many conveniences, the false information it produces has also become a significant problem. To address this challenge, tech giant Microsoft has recently launched a new tool called "Correction", aimed at rectifying misinformation in AI-generated content. As part of the Microsoft Azure AI Content Safety API, "Correction" is currently in preview. This tool can automatically flag texts that may contain errors, such as incorrectly quoted company earnings.

Quantum computing startup Riverlane has completed a $75 million Series C funding round to advance quantum error correction technology. Quantum computing relies on fragile qubits that are susceptible to external influences, making error correction crucial for ensuring qubit accuracy. The demand for error correction technology has surged over the past year due to technological breakthroughs and improved qubit quality. Riverlane aims to achieve one ....

The World Privacy Forum report indicates that over one-third of AI governance tools have 'defect corrections' that affect fairness and explainability assessments. Some tools and technologies developed by companies like Microsoft, IBM, and Google have also been found to have issues, raising concerns. The lack of quality assurance and evaluation requirements is the main problem, as some tools may have issues in different contexts and 'out-of-label usage'. Experts suggest that through collaboration between international organizations and standard-setting bodies, substantial improvements in AI governance tools could be achieved by 2024.

Google DeepMind's latest research finds that large language models have difficulty autonomously correcting reasoning errors. The study found that models often err when trying to correct their initial responses solely based on their own judgment. The researchers believe that in currently sensitive safety scenarios, self-correction capabilities are essential. Achieving consensus responses through voting can ensure consistency but is not equivalent to self-correction. Further research is needed to improve the self-correction abilities of language models.