AI news this week
Artificial intelligence (AI) is one of the most dynamic and exciting fields of technology today. Every week, there are new developments, breakthroughs and challenges that shape the future of AI. Here are some of the top stories of this week:
- OpenAI attracts deep-pocketed rivals in Anthropic and Musk: OpenAI, the research organization behind GPT-3 and other generative AI models, is facing competition from two new ventures: Anthropic, a stealthy startup founded by former OpenAI researchers and backed by Reid Hoffman; and a mysterious project led by Elon Musk, who co-founded OpenAI but left its board in 2018. Both rivals are reportedly working on large-scale AI systems that could rival or surpass GPT-3 in natural language understanding and generation.
- Amazon launches Bedrock, a service to build generative AI apps: Amazon Web Services (AWS) announced Bedrock, a new service that allows customers to build generative AI applications using pretrained models from startups such as AI21 Labs, Hugging Face and Cohere. Bedrock aims to lower the barriers to entry for developers who want to leverage the power of generative AI without having to train their own models or deal with complex infrastructure.
- AI newsreader debuts in China, sparks debate on human touch: A chatbot named Xiaoice, developed by Microsoft and a Chinese company, has become the first AI newsreader to host a live TV show in China. Xiaoice can interact with viewers, answer questions and generate natural-sounding speech based on text inputs. However, some critics argue that AI newsreaders lack the human touch and empathy that real journalists have, and may pose ethical and social risks.
- CMU teaches a robot dog to walk a balance beam: Researchers from Carnegie Mellon University (CMU) have developed a novel method to teach a robot dog to walk on a narrow balance beam. The method uses reinforcement learning, a technique that trains an agent to learn from trial and error, and incorporates prior knowledge about the dynamics of the robot and the environment. The researchers claim that their method can enable more agile and robust locomotion for legged robots in challenging terrains.
- Robot or human teacher? Kids' preferences vary by age and task: A study by researchers from MIT and Harvard University has explored how children of different ages prefer to learn from robots versus humans. The study involved 120 children aged 4 to 11 who interacted with either a robot or a human teacher on various tasks such as math problems, storytelling and drawing. The results showed that younger children preferred robots for math and drawing tasks, while older children preferred humans for storytelling tasks. The researchers suggest that robots could be designed to adapt to children's preferences and needs based on their age and task.
