Sign Up | Advertise | Podcast | AI University | | | Welcome, AI enthusiasts. | OpenAI just armed ChatGPT with new real-time web access — and the future of search just got a lot more conversational. | Is this the move that reshapes how we interact with information forever, or will Google's search empire still reign supreme? Let's get into it… | | In today's AI rundown: | OpenAI launches ChatGPT search Sam Altman's Reddit AMA, Dev Day updates Find and connect with company executives Tiny AI model masters humanoid control 6 new AI tools & 4 new AI jobs More AI & tech news
| Read time: 4 minutes |
|
| | | | OPENAI | |  | Image source: OpenAI |
| The Rundown: OpenAI just announced the integration of new web search capabilities directly into ChatGPT, transforming the chatbot into a real-time information hub that puts the platform in direct competition with Google and traditional search engines. | The details: | The new search feature provides instant answers about topics such as news, sports, stocks, and weather, with direct source attribution and links. ChatGPT will automatically trigger searches when needed, though users can manually activate web searches using a globe icon. The system runs on a specialized version of GPT-4o, fine-tuned for web info — with outputs including sourcing and clickable links to original content. Major publishers, including AP, Reuters, Axel Springer, and others, have signed licensing deals with OpenAI to provide content directly to the platform. Access starts with Plus and Team users today, expanding to Enterprise/Edu users within weeks and free users in the coming months.
| Why it matters: After months of rumors and speculation, OpenAI's search transformation is finally here. The move could signal a major shift in how we access info online, blending AI's conversational abilities with traditional search functionality, which could reshape both the search engine landscape and user habits. |
|
| | TOGETHER WITH SYNCLY | | | The Rundown: Syncly (YC W23) turns daily customer feedback into actionable product and operations insights, giving you full visibility into your customers' journey and proactively addressing issues before customers churn. | With Syncly, you can: | Centralize all feedback from various sources, integrating smoothly with your CX tools Track customer sentiments dynamically to act before issues escalate Automate VoC reports with powerful dashboards, eliminating manual work Ask Syncly AI critical business questions via AI Search to get concise feedback
| Try Syncly for free and transform customer feedback into success. |
|
| | OPENAI | |  | Image source: OpenAI |
| The Rundown: OpenAI hosted its first international Dev Day in London this week, showcasing new o1 and real-time voice API capabilities — alongside a revealing Reddit AMA with CEO Sam Altman and other members of the leadership team. | The details: | Demos included the o1-mini model creating a functioning drone control app, utilizing the Realtime API to order pies for the large group, and more. OpenAI slashed prices for its Realtime API, discounting text inputs by 50% and audio by 80%, along with five new voices for speech-to-speech capabilities. During the Reddit AMA, Altman admitted that OpenAI "can't ship as many things in parallel" as hoped due to model complexity and computing constraints. He also noted that the team prioritizes 'shipping o1 and its successors,' with OpenAI Chief Product Officer Kevin Weil saying the full o1 release will be 'soon.' Altman revealed that 'nothing called GPT-5' will arrive this year, but 'very good releases' are still planned for late 2024.
| Why it matters: OpenAI is shipping fast, but the candid admissions show that even the AI leader faces infrastructure limitations when it comes to executing each ambitious roadmap item. Even if GPT-5 isn't coming in 2024, there is enough smoke to guess that Altman and Co. still have something big up their sleeves for December. |
|
| | AI TRAINING | | | The Rundown: Perplexity's new Reasoning Mode allows you to find executive information and craft personalized connection requests that get accepted. | Step-by-step: | Access Perplexity's new Reasoning Mode with free Pro Search credits. Use multi-layered prompts to find executive profiles and LinkedIn URLs. Gather verified recent achievements and initiatives. Use ChatGPT to craft a personalized connection request.
| Pro tip: We have written detailed prompts that you can copy and paste from here. Additionally, we are having a workshop today that teaches you how to use Perplexity for enhanced research and collaboration. |
|
| | PRESENTED BY WEIGHTS & BIASES | | | The Rundown: RAG++ is a free new course designed to equip engineers with the skills needed to take their RAG proof-of-concept apps to production — helping you optimize performance, cut costs, and enhance accuracy. | In this course, you'll learn: | How to use query enhancement, hybrid retrieval, and agentic RAG Systematic techniques for RAG evaluation Best practices to minimize hallucinations and ensure consistent outputs Hands-on experience with course notebooks and Cohere API credits
| Enroll now for free and enter a raffle for W&B merch—exclusive to Rundown readers. |
|
| | NVIDIA | |  | Image source: NVIDIA |
| The Rundown: Nvidia just published new research showcasing HOVER, a small 1.5M parameter neural network that can control whole-body robotic movement effectively across various modes and input methods. | The details: | Despite being thousands of times smaller than typical AI models, the model achieves superior performance compared to specialized controllers. Nvidia trained the system in its 'Isaac simulator,' which compresses a year of robot training into just 50 minutes on a single GPU. The system works seamlessly with diverse input methods, including VR headsets, motion capture, exoskeletons, and joysticks. HOVER also transfers directly from simulation to real robots without requiring additional fine-tuning.
| Why it matters: While the AI world races toward larger models, HOVER shows that smaller systems can still outperform. The ability to work across multiple control modes with minimal compute requirements could help democratize advanced robotics control — making building sophisticated humanoid systems more accessible and practical. |
|
| | | | | 🗣️ ChatGPT Advanced Voice - Now available on desktop apps for both Mac and Windows ⚡️ GitHub Spark AI - Create and manage full-stack micro-apps with AI prompts ⚙️ Laminar - All-in-one platform for engineering AI projects 💻 Browser AI Kit - Run AI tools completely free and directly in your browser 💳 Trove - Real-time AI intelligence for financial transactions
| | |
|
| | | | Anthropic announced that Claude is now available via desktop apps on Apple and Windows and is also rolling out new dictation capabilities for mobile and iPad. | Google Maps integrated Gemini into the platform for new personalized recommendations, AI-powered navigation features, and expanded Immersive View capabilities. | Meta's FAIR team revealed three major robotics advances with open-source tactile sensing systems, including a human-like artificial fingertip and a unified platform for robotic touch integration. | D-ID unveiled Personal Avatars, a new hyper-realistic AI avatar suite for marketers — featuring digital humans capable of real-time interaction generated from just one minute of source footage. | Microsoft is pushing the Copilot Plus 'Recall' feature launch to December, marking the third delay this year as the company overhauls security and opt-in controls for the AI-powered screenshot system. | Google also launched a new experimental tool called 'Learn About,' which allows for AI-powered exploration of a range of topics via a conversational interactive interface. | ElevenLabs debuted a new 'X to Voice' feature allowing users to design a unique AI voice based on their X social media account. |
|
| | | | That's it for today!Before you go we'd love to know what you thought of today's newsletter to help us improve The Rundown experience for you. | | See you soon, | Rowan, Joey, Zach, and Alvaro—aka The Rundown Team | |
|
|
|
Tidak ada komentar:
Posting Komentar