News from the AI & ML world
DeeperML - #users
|
Michal Langmajer@Fello AI
//
OpenAI has launched its latest AI model, GPT-4.5, described as the company's most advanced language model to date. This new model features substantial enhancements over its predecessors, particularly in advanced reasoning, problem-solving, and contextual understanding. GPT-4.5 is designed to offer a more natural and engaging conversational experience, with improvements including superior capabilities in handling complex reasoning tasks, enhanced creativity, and the ability to manage intricate logic problems while maintaining nuanced conversations with improved contextual recall.
However, the launch of GPT-4.5 is facing challenges due to a shortage of GPUs, according to OpenAI CEO Sam Altman. This limitation is restricting access to the priciest tiers of ChatGPT Pro subscribers and developers initially. Altman stated that OpenAI has "run out of GPUs" due to growing demand, leading to a staggered rollout. The company plans to add tens of thousands of GPUs next week and expand access to Plus, Team, Enterprise, and Edu users in the following weeks. Recommended read:
References :
Dr. Hura@Digital Information World
//
OpenAI has released exciting updates for ChatGPT's Advanced Voice Mode, aimed at creating more natural and engaging user interactions. The primary focus of these updates is to reduce interruptions during conversations, a common issue where the AI would interject during pauses, hindering the flow of natural dialogue. This improvement allows users to take short breaths or think without the AI prematurely responding.
The Advanced Voice Mode is now available to all ChatGPT users with paid plans. Those with the free version of the chatbot will get access to the latest Advanced Voice Mode that enables users to pause without getting interrupted or when they want to speak to the AI assistant. The system requirements include Android app version 1.2024.206 or later, and for iOS, app version 1.2024.206 or later with iOS 16.4 or later. In addition to minimizing interruptions, the update introduces a more personable tone to ChatGPT's voice interactions. The AI is designed to be more specific, direct, creative, and engaging in its replies, making conversations feel less robotic and more human-like. These changes come amid competition from other companies launching similar AI voice assistants, such as Sesame's new tool, Maya and Miles. Recommended read:
References :
Matthias Bastian@THE DECODER
//
Google is enhancing its Gemini AI assistant with the ability to access users' Google Search history to deliver more personalized and relevant responses. This opt-in feature allows Gemini to analyze a user's search patterns and incorporate that information into its responses. The update is powered by the experimental Gemini 2.0 Flash Thinking model, which the company launched in late 2024.
This new capability, known as personalization, requires explicit user permission. Google is emphasizing transparency by allowing users to turn the feature on or off at any time, and Gemini will clearly indicate which data sources inform its personalized answers. To test the new feature Google suggests users ask about vacation spots, YouTube content ideas, or potential new hobbies. The system then draws on individual search histories to make tailored suggestions. Recommended read:
References :
Matthias Bastian@THE DECODER
//
Baidu has released its advanced AI models, ERNIE 4.5 and ERNIE X1, making them freely available to users through the ERNIE Bot platform. This move is a direct challenge to AI giants like OpenAI, Google, and DeepSeek, aiming to provide broader access to cutting-edge AI technology. By offering these models for free, Baidu seeks to accelerate user engagement and gather real-world data to refine their AI capabilities, potentially shifting the balance in the AI landscape by making sophisticated AI tools a new standard rather than a luxury.
ERNIE 4.5 is a multimodal foundation model capable of integrating and understanding text, images, audio, and video. It enhances language understanding, reasoning, generation, and memory, and can even interpret internet memes and satirical cartoons. ERNIE X1, on the other hand, is designed for reasoning-intensive tasks, excelling in logical inference, problem-solving, and structured decision-making. Baidu claims ERNIE 4.5 can match the performance of models like DeepSeek R1 at half the deployment cost, while ERNIE X1 has demonstrated strong capabilities in areas like Chinese knowledge Q&A and complex calculations. Recommended read:
References :
Ryan Daws@AI News
//
Anthropic has announced that its AI assistant Claude can now search the web. This enhancement allows Claude to provide users with more up-to-date and relevant responses by expanding its knowledge base beyond its initial training data. It may seem like a minor feature update, but it's not. It is available to paid Claude 3.7 Sonnet users by toggling on "web search" in their profile settings.
This integration emphasizes transparency, as Claude provides direct citations when incorporating information from the web, enabling users to easily fact-check sources. Claude aims to streamline the information-gathering process by processing and delivering relevant sources in a conversational format. Anthropic believes this update will unlock new use cases for Claude across various industries, including sales, finance, research, and shopping. Recommended read:
References :
Chris McKay@Maginative
//
OpenAI has recently unveiled new audio models based on GPT-4o, significantly enhancing its text-to-speech and speech-to-text capabilities. These new tools are intended to give AI agents a voice, enabling a range of applications, with demonstrations including the ability for an AI to read emails in character. The announcement includes the introduction of new transcription models, specifically gpt-4o-transcribe and gpt-4o-mini-transcribe, which are designed to outperform the existing Whisper model.
The text-to-speech and speech-to-text tools are based on GPT-4o. While these models show promise, some experts have noted potential vulnerabilities. Like other large language model (LLM)-driven multi-modal models, they appear susceptible to prompt-injection-adjacent issues, stemming from the mixing of instructions and data within the same token stream. OpenAI hinted it may take a similar path with video. Recommended read:
References :
@www.eweek.com
//
OpenAI's ChatGPT has achieved a major milestone, now boasting over 400 million weekly active users. This growth reflects the platform's increasing adoption by both individual consumers and enterprise clients. The number of paying enterprise users has doubled since September, reaching 2 million. This surge is attributed to users recommending ChatGPT to their workplaces, with companies recognizing its potential as a time-saving tool.
OpenAI is also expanding the availability of its "Operator" AI agent, granting access to ChatGPT Pro subscribers in countries like Australia, Brazil, Canada, India, Japan, South Korea, and the United Kingdom. This AI agent can independently operate a web browser and perform tasks for users. Additionally, OpenAI has taken action against malicious activities by removing accounts suspected of being used for surveillance and opinion-influence operations, particularly those originating from China and North Korea. These measures underscore OpenAI's commitment to broadening access while safeguarding against misuse. Recommended read:
References :
@laptopmag.com
//
Apple is reportedly developing future Apple Watch models with integrated cameras to enhance its artificial intelligence capabilities. According to reports, Apple aims to add small cameras to both the standard Apple Watch and the Ultra version, potentially positioning them on the front screen or where the side button is located. The goal is to provide the watch with visual perception of its environment, making it smarter and more useful for users.
Adding cameras to the Apple Watch would enable new features, similar to object and text scanning currently available on iPhones. For example, the watch could identify objects, translate foreign text, or provide information about food packaging for allergy-conscious individuals. The camera system could also have applications in health monitoring, such as examining skin tone to determine sleep quality or measuring heart rate through wrist readings. Apple has been exploring various camera designs for the Apple Watch for years, with patents showcasing different approaches. These include hiding a camera in the watch band for health tracking, a pop-up camera for occasional use, and even integrating a camera into the Digital Crown or using a flip-up screen. Despite these advancements, Apple is shuffling its executive team, replacing the head of AI with the VP of Vision Pro, suggesting a strategic shift in its AI efforts, with internal sources labeling delays with new AI rollouts as "ugly." Recommended read:
References :
Vasu Jakkal@Microsoft Security Blog
//
Microsoft and Google are enhancing their AI security measures and capabilities to address the evolving landscape of artificial intelligence. Microsoft is prioritizing secure AI integration, focusing on data protection and ensuring AI systems are transparent and compliant. Microsoft Secure, an online event, showcased AI innovations for the security lifecycle, providing tools for smarter, faster, and stronger security. The event highlighted how to secure data used by AI, AI apps, and AI cloud workloads, as well as how to protect AI investments from cyberthreats through data security and compliance tools.
NASA is leveraging Microsoft AI capabilities through its new Earth Copilot to democratize access to complex Earth Science data. This empowers scientists to discover patterns and gain insights from the vast amount of data collected by NASA's satellites, which can inform policy decisions and support various industries. Meanwhile, Google is developing Gemini to enable it to take actions within apps, potentially transforming AI assistants into more useful tools. Google is also releasing the GEMMA-3 models to improve the capabilities of the Gemini project. Recommended read:
References :
S.Dyema Zandria@The Tech Basic
//
Google is enhancing its Gemini AI with a new feature that allows users to create AI podcasts from research materials. This new capability, called Audio Overviews, converts research and study materials into engaging, podcast-style discussions featuring AI hosts. This aims to make learning and information consumption more accessible and enjoyable, particularly for educational purposes.
The Audio Overviews feature leverages Gemini's Deep Research capabilities. Users can input a topic, have Gemini generate a detailed report, and then convert that report into a conversational podcast with AI hosts. These hosts discuss the information in an approachable manner, similar to two friends exploring a topic. This tool is available to both free and paid Gemini Advanced users. Recommended read:
References :
@felloai.com
//
Apple is facing mounting criticism over its AI implementation, with reports of its "Apple Intelligence" features causing significant issues. The AI powered summaries are distorting headlines, fabricating news, misreporting deaths and incorrectly stating events. There have been several instances of these inaccuracies with a man being falsely reported to have died, a darts player being announced as a winner before a competition, and a tennis star being inaccurately outed as gay. These errors, have sparked concerns about trust and reliability in the tech company's AI, leading some to suggest that Apple is facing an AI trust crisis.
Apple's AI push is also being scrutinized for how it handles messages. The AI is rewriting scam messages to appear legitimate and marking them as priority messages, further complicating the issue. This is compounded by notification summaries that misinterpret data, resulting in inaccurate information being delivered to users. Apple’s attempts to mitigate the problems with a UI tweak to clarify when summaries are AI-generated are seen as insufficient, with critics arguing that such measures will not stop the spread of misinformation, and are not doing enough to address the core problems of its flawed AI. Recommended read:
References :
@www.gadgets360.com
//
OpenAI’s ChatGPT, API, and Sora services experienced a major outage, causing high error rates and inaccessibility for users globally. This disruption affected various functionalities, including text generation, API integrations, and the Sora text-to-video platform. The root cause was identified as an issue with an upstream provider, and OpenAI worked to restore services. This outage highlights the challenges and dependencies in AI infrastructure.
Recommended read:
References :
Evelyn Blake@The Tech Basic
//
References:
The Tech Basic
, gHacks Technology News
,
Google has begun rolling out real-time interaction features to its AI assistant, Gemini, enabling live video and screen sharing. These enhancements, powered by Project Astra, allow users to engage more intuitively with their devices, marking a significant advancement in AI-assisted technology. These features are available to Google One AI Premium subscribers.
The new live video feature allows users to utilize their smartphone cameras to engage in real-time visual interactions with Gemini, enabling the AI to answer questions about what it observes. Gemini can analyze a user’s phone screen or camera feed in real-time and instantly answer questions. The screen-sharing feature enables the AI to analyze and provide insights on the displayed content, useful for navigating complex applications or troubleshooting issues. Google plans to expand access to more users soon. Recommended read:
References :
Evelyn Blake@The Tech Basic
//
References:
The Tech Basic
, gHacks Technology News
,
Google has started rolling out new AI tools for Gemini, allowing the assistant to analyze your phone screen or camera feed in real time. These features are powered by Project Astra and are available to Google One AI Premium subscribers. The update transforms Gemini into a visual helper, enabling users to point their camera at an object and receive descriptions or suggestions from the AI.
These features are part of Google's Project Astra initiative, which aims to enhance AI's ability to understand and interact with the real world in real-time. Gemini can now analyze your screen in real-time through a "Share screen with Live" button and analyze your phone's camera. Early adopters have tested the screen-reading tool, and Google plans to expand access to more users soon. With Gemini's live video and screen sharing functionalities, Google is positioning itself ahead in the competitive landscape of AI assistants. Recommended read:
References :
@blogs.microsoft.com
//
References:
IEEE Spectrum
, IEEE Spectrum
,
Anthropic, Google DeepMind, and OpenAI are at the forefront of developing AI agents with the ability to interact with computers in a human-like manner. These agents are designed to perform a range of tasks, including web searches, form completion, and button clicks, enabling them to order groceries, request rides, or book flights. The models employ chain-of-thought reasoning to decompose complex instructions into manageable steps, requesting user input when necessary and seeking confirmation before executing final actions.
To address safety concerns such as prompt injection attacks, developers are implementing restrictions, such as preventing the agents from logging into websites or entering payment information. Anthropic was the first to unveil this functionality in October, with its Claude chatbot now capable of "using computers the way humans do." Google DeepMind is developing Mariner, built on top of Google’s Gemini 2 language model and OpenAI launched its computer-use agent (CUA), called Operator. Recommended read:
References :
Merin Susan@Analytics India Magazine
//
OpenAI is facing internal and external scrutiny regarding the ethical implications of its AI technologies. Employees have voiced concerns about a potential military deal with the startup Anduril, fearing damage to OpenAI's reputation due to its association with a weapons manufacturer. One employee noted that the company seemed to be downplaying the implications of working with a weapons manufacturer. Another employee said that they were concerned the deal would hurt OpenAI’s reputation.
OpenAI's technologies, specifically ChatGPT, have also come under scrutiny regarding their potential impact on mental health. Research indicates that specific types of ChatGPT usage, particularly "personal conversations" involving emotional expression, may be linked to increased loneliness among users. A study found that users who were more prone to emotional attachment were more likely to report increased loneliness in response to frequent personal conversations with the chatbot. Interestingly, the research also highlights that most people use ChatGPT for practical purposes rather than seeking emotional support. Recommended read:
References :
@singularityhub.com
//
OpenAI models, including the recently released GPT-4o, are facing scrutiny due to their vulnerability to "jailbreaks." Researchers have demonstrated that targeted attacks can bypass the safety measures implemented in these models, raising concerns about their potential misuse. These jailbreaks involve manipulating the models through techniques like "fine-tuning," where models are retrained to produce responses with malicious intent, effectively creating an "evil twin" capable of harmful tasks. This highlights the ongoing need for further development and robust safety measures within AI systems.
The discovery of these vulnerabilities poses significant risks for applications relying on the safe behavior of OpenAI's models. The concern is that, as AI capabilities advance, the potential for harm may outpace the ability to prevent it. This risk is particularly urgent as open-weight models, once released, cannot be recalled, underscoring the need to collectively define an acceptable risk threshold and take action before that threshold is crossed. A bad actor could disable safeguards and create the “evil twin” of a model: equally capable, but with no ethical or legal bounds. Recommended read:
References :
@www.reuters.com
//
References:
Engineering at Meta
, www.eweek.com
,
Meta is expanding its artificial intelligence research into the realm of humanoid robotics, aiming to develop AI-driven software and sensors. This initiative focuses on creating intelligent machines that can interact with the physical world, potentially powering consumer robots. The company's efforts are concentrated on "embodied AI," which combines intelligence with real-world interactions, enabling robots to move, sense, and make decisions in three-dimensional environments.
Meta is not initially planning to release its own branded robots. Instead, the company is concentrating on developing AI-powered software and sensor technology that can be utilized by other robotics manufacturers. This strategy positions Meta alongside tech giants like Tesla, Apple, and Google, all of which are also investing in the robotics sector. Meta is also prioritizing user data protection by using source code analysis to detect and prevent unauthorized data scraping across its platforms, including Facebook, Instagram, and Reality Labs. Recommended read:
References :
@www.ghacks.net
//
Recent security analyses have revealed that the iOS version of DeepSeek, a widely-used AI chatbot developed by a Chinese company, transmits user data unencrypted to servers controlled by ByteDance. This practice exposes users to potential data interception and raises significant privacy concerns. The unencrypted data includes sensitive information such as organization identifiers, software development kit versions, operating system versions, and user-selected languages. Apple's App Transport Security (ATS), designed to enforce secure data transmission, has been globally disabled in the DeepSeek app, further compromising user data security.
Security experts from NowSecure recommend that organizations remove the DeepSeek iOS app from managed and personal devices to mitigate privacy and security risks, noting that the Android version of the app exhibits even less secure behavior. Several U.S. lawmakers are advocating for a ban on the DeepSeek app on government devices, citing concerns over potential data sharing with the Chinese government. This mirrors previous actions against other Chinese-developed apps due to national security considerations. New York State has already banned government employees from using the DeepSeek AI app amid these concerns. Recommended read:
References :
Matthias Bastian@THE DECODER
//
Google is launching a new experimental feature called 'Daily Listen' that uses AI to create personalized podcast-style summaries within its Discover feed. This feature curates roughly five-minute audio episodes based on a user's search history and news preferences. The system pulls together information on topics and stories the user follows, delivering it within the Google app on Android and iOS devices. Daily Listen is currently available in the US for users who have opted into Search Labs experiments, with a limited number of spots being available. The personalized podcast appears below the search bar, with a date and "Made for you" label.
Once enabled, the AI-generated episode will appear the following day in the Discover feed, providing a brief audio overview of the user’s favorite topics. The podcast includes a full-screen player with audio playback controls, a text transcript, and related stories corresponding to each section. The technology is similar to Google’s NotebookLM audio overviews, but while NotebookLM is based on a user prompt, Daily Listen examines topics and stories the user already follows, with a focus on delivering helpful, although fallible, information. Recommended read:
References :
Chris McKay@Maginative
//
References:
AI & Machine Learning
, Maginative
,
Google is enhancing its NotebookLM tool with interactive mind maps, a feature designed to help users visualize and navigate complex information from uploaded sources. These mind maps present document topics as branching diagrams, allowing users to explore connections and ask questions about specific areas by clicking on nodes. This visual approach aims to transform how users interact with their content, moving beyond linear reading to a more intuitive exploration of interconnected concepts.
LlamaIndex, a framework for building knowledge-driven AI agents, has also been integrated with Google Cloud's Gen AI Toolbox for Databases. This integration empowers developers to construct sophisticated AI agents with customizable workflows. LlamaIndex offers pre-built agent architectures for common use cases, along with tools to tailor the behavior of AI agents to specific requirements, which will benefit those using Gen AI Toolbox for Databases. Recommended read:
References :
@timesofindia.indiatimes.com
//
References:
www.artificialintelligence-new
, www.eweek.com
,
Recent developments highlight both the expanding influence and the regulatory hurdles faced by the AI company DeepSeek. In South Korea, the government has halted downloads of DeepSeek's applications, citing concerns over data privacy. This action has removed the company's apps from both the Apple and Google mobile app marketplaces, though their website remains accessible.
Simultaneously, DeepSeek's AI technology is rapidly integrating into China's transportation sector, extending from electric vehicles (EVs) to e-scooters. Major automakers, including BYD, Geely, and Chery Automobile, are incorporating DeepSeek's AI into their vehicles, offering features like preliminary self-driving capabilities. E-scooter brands like Segway-Ninebot and Niu Technologies are also integrating DeepSeek for enhanced features such as AI-powered content creation, data analytics, and driver assistance systems, reflecting what some industry observers are calling "DeepSeek fever" due to its cost-effective AI integration. Perplexity has released "1776," a modified version of DeepSeek-R1. This model addresses the original version's limitations by mitigating censorship on sensitive topics, particularly those related to Chinese history and geopolitics. The modifications were made using post-training techniques to ensure more open and contextually accurate responses, making the modified model available on Perplexity's Sonar AI platform and GitHub. Recommended read:
References :
@shellypalmer.com
//
OpenAI has recently launched a new beta feature in ChatGPT called 'Tasks'. This functionality allows users to schedule future actions and reminders, expanding ChatGPT's capabilities beyond basic conversations. Users can now set up both one-time and recurring tasks. These tasks could include daily summaries of news or sports scores, scheduled searches, or even conditional alerts based on specific criteria. The implementation uses the iCal VEVENT format for scheduling and is integrated through a tool in the system prompt.
The Tasks feature can be accessed through the "4o with scheduled tasks" model option on the ChatGPT website. Tasks can be managed through chat threads or a dedicated Tasks section, which simplifies modification and cancellation. Notifications are delivered across web, desktop, and mobile platforms upon task completion. Although it's still in beta, this feature aims to make ChatGPT a more versatile AI assistant, capable of automating daily tasks and workflows. Some users have reported early beta bugs with push notifications and date formats. Recommended read:
References :
@www.verdict.co.uk
//
OpenAI is shifting its strategy by integrating its o3 technology, rather than releasing it as a standalone AI model. CEO Sam Altman announced this change, stating that GPT-5 will be a comprehensive system incorporating o3, aiming to simplify OpenAI's product offerings. This decision follows the testing of advanced reasoning models, o3 and o3 mini, which were designed to tackle more complex tasks.
Altman emphasized the desire to make AI "just work" for users, acknowledging the complexity of the current model selection process. He expressed dissatisfaction with the 'model picker' feature and aims to return to "magic unified intelligence". The company plans to unify its AI models, eliminating the need for users to manually select which GPT model to use. This integration strategy also includes the upcoming release of GPT-4.5, which Altman describes as their last non-chain-of-thought model. A key goal is to create AI systems capable of using all available tools and adapting their reasoning time based on the task at hand. While GPT-5 will be accessible on the free tier of ChatGPT with standard intelligence, paid subscriptions will offer a higher level of intelligence incorporating voice, search, and deep research capabilities. Recommended read:
References :
|
BenchmarksBlogsResearch Tools |