ChatGPT: A New Era of Crowd Work
OpenAI’s language model, ChatGPT, is reshaping the landscape of crowd work. The AI model is trained using a method called Reinforcement Learning from Human Feedback (RLHF), which involves collecting comparison data where two or more model responses are ranked by quality. This method has led to a significant improvement in the model’s performance. However, it has also raised concerns about the working conditions of the human raters involved in the process. OpenAI has addressed these concerns by ensuring raters’ anonymity, providing them with a support system, and maintaining an ongoing relationship with them for iterative feedback. Read more: Wired
GPT Detectors Biased Against Non-Native English Speakers
A recent study has revealed that GPT detectors, designed to identify text generated by AI, are biased against non-native English speakers. The detectors often misclassify non-native English text as AI-generated, leading to potential discrimination. The study suggests that the training data used for these detectors, which is predominantly in English and from specific regions, is the root cause of this bias. This highlights the need for more diverse and inclusive training data in AI models. Read more: TechXplore
AI Copyright Case: ChatGPT in Court
In a landmark case, OpenAI’s ChatGPT is being taken to court over a copyright dispute. The case revolves around the question of whether AI-generated text can infringe on copyright laws. The plaintiff argues that ChatGPT used their copyrighted material to generate similar content, thereby violating their rights. The outcome of this case could have significant implications for the future of AI and copyright law. Read more: TechXplore
Google’s AI Health Chatbot: A Medical Revolution
Google has launched an AI health chatbot that can provide medical advice. The chatbot, which is still in the experimental phase, uses machine learning to understand and respond to health-related queries. It can provide information on a wide range of health topics, from common illnesses to mental health. However, Google emphasizes that the chatbot is not a substitute for professional medical advice. Read more: TechXplore
Code Interpreter for All: ChatGPT Plus Users Turn Data Analysts
VentureBeat has announced the introduction of a code interpreter to all ChatGPT Plus users. This new feature allows anyone to become a data analyst, regardless of their coding skills. The code interpreter can understand and execute Python code, making it easier for users to interact with data directly within the chat. This is a significant step towards democratizing data analysis. Read more: VentureBeat
Claude 2: AI Model for Longer, Safer Responses
Anthropic has unveiled Claude 2, an AI model that produces longer and safer responses. The updated model can generate responses of up to 4,000 tokens, significantly improving performance on metrics like coding, math, and logic problems. Claude 2 is designed to be helpful, harmless, and honest, and is being used by thousands of businesses, including productivity companies like Slack and Notion. Read more: VentureBeat
Elon Musk’s xAI: Understanding the Universe
Elon Musk has announced the formation of xAI, an organization aimed at using AI to understand the true nature of the universe. The team, led by Musk and veterans of DeepMind, OpenAI, Google Research, Microsoft Research, Tesla, and the University of Toronto, plans to build what Musk refers to as “TruthGPT,” a maximum-truth-seeking AI. The details of xAI’s mission and methods will be revealed in a Twitter Spaces event. Read more: TechCrunch
Google’s Bard: Now in EU and Supporting 40 Languages
Google’s AI-powered chatbot, Bard, has finally launched in the European Union and now supports more than 40 languages. The launch comes after a delay due to data privacy concerns raised by the Irish Data Protection Commission. Bard’s expansion includes new features such as a tone adjustment toggle and a text-to-speech AI feature. Despite early challenges, Google claims that Bard is improving, particularly in areas like math and programming. Read more: TechCrunch
Stable Doodle: Turning Sketches into Artwork
Stability AI has released Stable Doodle, a new tool that turns sketches into artwork. The service uses the latest Stable Diffusion model to analyze the outline of a sketch and generate a visually pleasing artistic rendition of it. Stable Doodle is available through ClipDrop, a platform Stability acquired recently. The tool is designed for both professionals and novices, allowing anyone with basic drawing skills to generate high-quality original images in seconds. Read more: TechCrunch
