TLRD of AI News – August 17, 2023

Bots Outperforming Humans in CAPTCHA Tests

Researchers have developed AI models that are capable of solving CAPTCHA challenges with a higher success rate than humans. The study, conducted by a team from Lancaster University, Northwest University in China, and Peking University, revealed that the AI models were able to solve text-based CAPTCHAs with a 95.12% success rate, compared to humans’ 87.21%. The researchers used a Generative Adversarial Network (GAN) to train the models, and their success raises concerns about the future effectiveness of CAPTCHAs as a security measure. Read more

New Standards for AI in Newsrooms

The European Broadcasting Union (EBU) has released a set of guidelines for the use of AI in newsrooms. The guidelines aim to ensure that AI technologies are used responsibly and ethically in journalism. They cover areas such as transparency, accountability, data protection, and human oversight. The EBU believes that these guidelines will help news organizations maintain trust and credibility while leveraging AI to enhance their operations. The guidelines were developed in collaboration with various stakeholders, including journalists, engineers, and legal experts. Read more

Open-Source Platform for 2D Simulation of Quantum Systems

A team of researchers from the University of Bristol has developed an open-source Python-based platform for the 2D simulation of quantum systems. Named “Quipper-D”, the platform allows users to simulate quantum circuits and algorithms, providing a valuable tool for researchers and educators in the field of quantum computing. The platform is designed to be user-friendly and can be used to explore various quantum phenomena, including entanglement and quantum error correction. Quipper-D is expected to contribute to the growing field of quantum computing by making simulations more accessible to a wider audience. Read more

Buildbox Introduces StoryGames.AI for Game Narratives

Buildbox has launched StoryGames.AI, a new tool that combines generative AI with user-friendly tools to enable creators to develop visual narrative games. The platform allows users to create visual novel games by typing a storyline, and it will automatically generate the first 10 chapters. StoryGames.AI integrates with the Buildbox ecosystem, enabling creators to build games for various platforms, including Apple App Store, Google Play, and Steam. The platform’s ease of use and accessibility are expected to attract aspiring game developers and offer limitless creative possibilities. Read more

OpenAI Acquires Startup Global Illumination, Inc.

OpenAI has announced the acquisition of Global Illumination, Inc., a New York City-based startup founded by former Facebook and Instagram workers. Global Illumination has been leveraging AI to build creative tools, infrastructure, and digital experiences. The acquisition is seen as an excellent “acqui-hire” for OpenAI, and the talent from Global Illumination will work on core products including ChatGPT. The move indicates that OpenAI may be pursuing more multimedia features for ChatGPT or expanding its Dall-E 2 image generation service. Terms of the deal were not disclosed. Read more

Opera’s iOS Web Browser Gains an AI Companion with Aria

Opera’s web browser app for iOS now includes Aria, an AI assistant built in collaboration with OpenAI. Aria is integrated directly into the web browser and is free for all users. The AI solution provides intelligent insights, ideas, and responsive voice commands. Aria is based on Opera’s “Composer” infrastructure and connects with OpenAI’s GPT technology. The addition of Aria to Opera’s iOS app follows its successful launch on Opera for desktop and Android, where it has reached over 1 million users. Read more

Arthur Releases Open Source Tool to Find the Best LLM for a Job

Machine learning monitoring startup Arthur has released Arthur Bench, an open-source tool designed to help users find the best Language Model (LLM) for a specific set of data. Arthur Bench allows users to test and measure the performance of different LLMs against specific prompts, enabling them to make informed decisions on the best model for their particular application. The tool is available as an open-source project, with a SaaS version planned for those with larger test requirements. Arthur Bench follows the release of Arthur Shield, an LLM firewall designed to detect hallucinations in models and protect against toxic information and private data leaks. Read more

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top
Share via
Copy link