SCRGoogle has spent around $2.7 billion to bring AI expert Noam Shazeer back to the company. Shazeer, co-author of a groundbreaking AI research paper, left Google in 2021 to start his own company, Character.AI. When that ran into trouble, Google offered the large sum – officially for a technology license. In reality, however, the main …
In their new book “AI Snake Oil,” two Princeton researchers warn against exaggerated expectations of artificial intelligence. Arvind Narayanan and Sayash Kapoor criticize companies, scientists, and journalists for making misleading claims about AI. The authors emphasize the need for better education about AI to demystify the technology. They divide AI into two subcategories: predictive AI, …
Researchers are working on new AI systems that can check their own mathematical answers. This is reported by Cade Metz in the New York Times. The technology is meant to prevent chatbots like ChatGPT from providing incorrect information. Instead, the new systems can generate mathematical evidence to verify their answers. Silicon Valley startup Harmonic, for …
Cloudflare is introducing new tools to protect websites from AI bots and control their scraping. The company is offering monitoring and selective blocking of AI data scraping bots to all customers, including its 33 million free users. CEO Matthew Prince says these measures are designed to give website owners more control over how and when …
A new open-source AI model called Molmo could help advance the development of AI agents. Developed by the Allen Institute for AI (Ai2), the model can interpret images and communicate via a chat interface. According to Wired’s Will Knight, this enables AI agents to perform tasks such as web browsing or document creation. In some …
Researchers at ETH Zurich have completely cracked Google’s reCAPTCHAv2 system. Using advanced machine learning methods, they solved 100% of the CAPTCHAs designed to distinguish humans from bots. Although the process still requires human intervention, fully automated CAPTCHA circumvention may soon become a reality. In response, companies like Google are developing more complex systems. However, this …
OpenAI is providing insight into its training data for the first time as part of a lawsuit, Winston Cho writes for The Hollywood Reporter. Several authors, including Sarah Silverman, accuse OpenAI of using their works to train ChatGPT without permission. Representatives of the plaintiff authors are now being allowed to review the material in a …
Google today announced major updates to its Gemini AI platform. The Gemini-1.5-Pro-002 and Gemini-1.5-Flash-002 models have been improved and now offer higher quality for mathematical tasks, long contexts, and visual applications. At the same time, Google is reducing the price of Gemini 1.5 Pro by more than 50% while significantly increasing the limits. A major …
Duolingo is introducing new AI-powered features to improve language learning. The company unveiled the “Video Call” and “Adventures” features at its annual Duocon 2024 conference. As CEO Luis von Ahn explains, these innovations are designed to create immersive learning opportunities. “Video Call” allows Duolingo Max subscribers to have realistic conversations with the AI character Lily. …
Microsoft has introduced new security features for AI offerings. The company wants to address growing concerns about AI security, privacy, and reliability. As Michael Nuñez reports for VentureBeat, the “Trustworthy AI” initiative includes confidential inference for Azure OpenAI Services, improved GPU security, and better tools for evaluating AI results. A key feature is the new …