Here are this week's top trending stories:
1. **AI Meets Quantum: A Significant Leap for AI Understanding**
AI researchers have long aspired to create systems that truly understand the world, moving beyond mere predictions. A team of quantum computing scientists has made notable progress by developing a new framework that equips machines to learn in a human-like manner. Their paper introduces the Compositional Quantum Framework, which allows AI systems to automatically structure and learn concepts from data using both classical and quantum computing methods. This innovation enables machines to visually comprehend and recognize images, grasping the significance of various objects by learning attributes such as shape, color, size, and position. By integrating insights from quantum computing and cognitive science, the researchers have advanced the quest for AI that can visualize and interpret actions.
2. **Air Canada Held Accountable for Chatbot Miscommunication**
In a recent ruling by a Canadian small claims tribunal, Air Canada was ordered to reimburse a traveler due to misleading information provided by its chatbot. After purchasing tickets for a bereavement trip, Jake Moffatt relied on the chatbot's guidance regarding the airline's bereavement policies. However, when he attempted to apply for a discounted fare after his return, he was informed that retroactive requests were not permitted. Instead, the airline offered a $200 coupon for future travel. Unhappy with this outcome, Moffatt sought justice from the Civil Resolution Tribunal, which concluded that Air Canada was liable for the misinformation. Ultimately, the airline was instructed to pay Moffatt a total of $812.02, covering his refund, interest, and tribunal fees.
3. **Hugging Face Launches Open Source AI Assistant Builder**
Hugging Face, known for its popular code repository, has unveiled a free and open-source AI assistant builder. Located within Hugging Face Chat, this new tool enables developers to create bespoke AI assistants utilizing open-source language models. This initiative offers a similar functionality to OpenAI's GPT builder available to ChatGPT Plus subscribers. According to technical lead Philipp Schmid, the assistants can leverage any public open large language model (LLM) such as Llama2 or Mixtral. However, Schmid cautioned users against entering confidential information, as all AI assistants are public by default. Currently, the function calling feature introduced by OpenAI for GPT is not available, but users can create and host their private AI assistants.
4. **Google Introduces Open Source Language Models to Compete with Rivals**
Google has launched a new family of open-source language models named Gemma, positioning them as a competitor to offerings from Meta, Mistral, and other open-source entities. Based on the Gemini architecture—Google's most advanced multimodal model—the Gemma models focus exclusively on text and code. The company released two sizes: a 2-billion parameter model and a 7-billion parameter model. Early assessments indicate that Gemma models excel in areas such as dialogue, reasoning, mathematics, and coding, potentially outperforming Meta’s Llama 2 and Mistral models of similar sizes.
5. **CISOs' Concerns with Generative AI Explored**
CrowdStrike Chief Scientist Sven Krasser has shared insights on the security challenges and opportunities associated with generative AI. His analysis delves into the predominant concerns that today's security leaders face when contemplating the integration of generative AI into their organizational environments. Whether utilized as a tool for employee use or embedded within product components, understanding these concerns is crucial for a secure and effective implementation.
For a comprehensive overview of the latest developments in AI, stay tuned for future updates.
Ranking
