Information

1AI noted that the National Internet Information Office released 2024 Generative Artificial Intelligence Services has filed information. To promote the innovative development and standardized application of generative AI services, in 2024, the net information department, together with relevant departments, continued to carry out the filing of generative AI services in accordance with the requirements of the Interim Measures for the Administration of Generative Artificial Intelligence Services. As of December 31, 2024, a total of 302 generative AI services had completed the filing at the State Internet Information Office, of which 238 were newly filed in 2024; for those generative AI services that are provided through API interfaces or their ......

Recently, Hong Kong startup Weitu AI announced the successful completion of its angel round of financing with a valuation of up to $100 million. Investors include hundreds of millions of monthly activity of global Internet technology companies and famous angel investors. Weitu AI has just been founded, and more than half of its team members graduated from prestigious North American universities and have experience working in large overseas factories. The team brings together research experts who have long been active in the multimodal field of artificial intelligence, including the technical leader who led the team to develop the first batch of Chinese large language models in China early last year. The company said that the research and development of multimodal large models requires both basic large model research and development capabilities and long-term ......

January 14, 2011 - A study from New York University has revealed the potential risks of large-scale language models (LLMs) in medical information training. The study shows that even training data containing as little as 0.001% of incorrect information can cause the model to output inaccurate medical answers. Data "poisoning" is a relatively simple concept: LLM is usually trained on large amounts of text, mostly from the Internet. By injecting specific information into the training data, it is possible for the model to treat this information as fact when generating answers. This approach doesn't even require direct access to the LLM itself, but rather just the contents of .......

According to a report by Copyleaks, 60% of OpenAI's GPT-3.5 model outputs were plagiarized.Copyleaks used a proprietary scoring methodology that takes into account identical text, minor modifications, and rewrites, and assigns a "similarity score" to each output. ". GPT-3.5 is an advanced natural language processing model from OpenAI, but the originality of its output has been questioned. According to the latest findings, 45.71 TP3T of GPT-3.5's output has the same text, 27.41 TP3T has been slightly modified, ......

ByteDance recently launched Trae, an AI integrated development environment (IDE) for Chinese developers, which aims to compete with internationally recognized tools such as Cursor and Windsurf, especially in terms of the experience of Chinese developers, and hopes to fill the shortcomings of existing tools. In the daily programming, many developers often need to switch between English and Chinese, and although this mixed working mode is commonplace, it is inconvenient for many Chinese developers. Existing IDEs such as Cursor, Windsurf, and ...... have been used for this purpose.

Due to the aging of Japan's population, the elderly have become one of the main groups of victims of telecommunication fraud, and there are often criminals posing as relatives of the elderly, and on the pretext of trouble and inconvenience, they direct the elderly to use ATMs to complete money transfers over the phone. According to CCTV news, Japanese police have installed AI recognition systems in some ATMs to warn potential victims. Simply put, when the AI recognizes that a user is using an ATM while talking on the phone, it will automatically play a warning video in which a police officer, played by singer Keita Orange, points a magnified finger at the user and says: ......

January 24, technology media NeoWin published a blog post yesterday (January 23), reporting that Microsoft plans to deploy Copilot Chat for Outlook and Copilot Chat for Teams online starting in February 2025, which will bring a new Copilot Chat experience and provide easier access. Citing Microsoft's updated Microsoft 365 roadmap, 1AI attached the relevant information as follows: late January 2025 ......

The latest release of ScreenAI from Google Research marks another important advancement in language and speech-controlled computer interfaces. This AI model not only understands user interfaces and infographics, but also sets new performance benchmarks in a number of tasks such as answering infographic-based questions, summarizing content, and navigating user interfaces. The core innovation of ScreenAI is the textual representation of screenshots. The model is able to recognize the type and location of UI elements, an approach that uses synthetic training data generated by Google LLM PaLM2-S, enabling it to answer questions about screen information, screen ......