Politics

Elon Musk at CPAC

• Elon Musk spoke at the Conservative Political Action Conference (CPAC) on Thursday, interviewed by Newsmax presenter Rob Schmitt. Musk even pronounced 'I am become meme.'

Artificial Intelligence

Advances in Reasoning Models

DeepSeek's advanced reasoning capabilities are driving enterprises to adopt Together AI's optimized infrastructure platform. Additionally, research indicates that a 1B small language model can outperform a 405B large language model in reasoning tasks with the appropriate test-time scaling strategy, highlighting the potential of smaller models with optimized techniques.

AI Technology and Revenue Generation

AI-Driven Revenue Growth at Taboola

Boaz Shor's AI systems have generated billions globally, contributing significantly to Taboola's revenue with $80 million annually. His innovations include natural language processing, URL classification, and user behavior modeling. Companies with AI patent portfolios experience 20% higher valuation growth.

AI Advancement

Magma's Multistep Action Capabilities

Magma, an AI agent developed by Microsoft, is designed to execute multistep actions in both the real and digital realms.

Artificial Intelligence

Microsoft's AI Vision

Microsoft CEO Satya Nadella believes that the success of AI should be gauged by its impact on global economic growth, aiming for a 10% inflation-adjusted increase, rather than merely solving abstract problems. He encourages the tech industry to focus on practical applications and profitability, moving away from the hype surrounding artificial general intelligence (AGI).

Apple's Visual Intelligence Expansion

Apple's AI-based Visual Intelligence skill, initially on iPhone 16 models, is expected to extend to the iPhone 15 Pro with the iOS 18.4 update. This feature enables users to identify and learn about objects through the camera using the Action button or a Control Center button, helping identify animals, plants, landmarks, and businesses, providing users with requested details and search options.

AI Platform Updates

ChatGPT User Growth and Future Models

ChatGPT has reached over 400 million weekly active users, doubling its user count since August 2024. OpenAI plans to simplify its offerings and will soon make its GPT-4.5 and GPT-5 models available. The availability of GPT-5 to free users aims to further expand ChatGPT's user base.

Google's AI-Enhanced "Ask Photos" Feature

Google's AI-enhanced "Ask Photos" feature, available through Google Labs, enables users to search photos using conversational prompts. While intended to improve upon classic search, it sometimes struggles to differentiate conversational terms and may require keyword input similar to the classic search.

Apple's Visual Intelligence Expansion

Apple's Visual Intelligence, currently on iPhone 16 models, is expanding to iPhone 15 Pro and 16e via a future software update, likely iOS 18.4, expected in early April. This AI-based feature identifies objects through the camera, providing relevant information. Activation will be via the Action button or Control Center button for iPhone 15 Pro and 16e users. Visual Intelligence allows users to run a Google search or ask ChatGPT-specific questions about identified objects.

Legal and Business Changes

Clearview AI Leadership Change and Financial Status

Hoan Ton-That, CEO of Clearview AI, has resigned, remaining on the board. Early investor Hal Lambert and co-founder Richard Schwartz will take over as co-CEOs. Clearview AI, which sells facial recognition database access to law enforcement, has over $100 million in GDPR fines from European data protection agencies and claims to be in its "strongest position ever" financially, with record growth and revenue in 2024.

AI & Machine Learning

OpenAI Legal Challenges and Future Models

OpenAI faces a lawsuit in New York federal court regarding the alleged misuse of news articles, including those from The Intercept, to train its ChatGPT chatbot.

Rumors suggest that OpenAI might release an updated AI model, GPT-4.5 (Orion), potentially next week. GPT-5 may arrive in May, offering improved fact-checking and possibly unlimited chat access. Microsoft is expected to showcase the new model at its annual developer conference, Microsoft Build, in May.

AI for Protein Interaction Prediction

Researchers have developed FragFold, a new AI method published in PNAS, that uses AI models, including AlphaFold, to predict protein fragments capable of binding to and inhibiting full-length proteins in E. coli. This tool could lead to genetically encodable inhibitors against any protein. Experimental results confirmed that over half of FragFold's predictions for binding or inhibition were accurate, even without prior structural data.

AI Development and Applications

OpenAI's Growth and Future Plans

OpenAI's ChatGPT has reached 400 million weekly users and doubled enterprise adoption to over 2 million users, indicating a significant shift towards AI adoption. The company is preparing for the launch of GPT-5, amidst increasing competition from DeepSeek and xAI.

AI in Medical Education

NYU Langone has developed an LLM research companion and medical advisor, pioneering what it terms AI-driven “precision medical education.”

AI for Learning American Sign Language

NVIDIA has launched an interactive website called Signs that uses AI to train users in American Sign Language (ASL). The site utilizes a device's camera to ensure correct hand and finger movements, offering tutorials for beginners and various levels of complexity. Developed in collaboration with the American Society for Deaf Children and Hello Monday/DEPT, users can replicate signs demonstrated by a 3D avatar, receiving feedback and contributing their own videos to expand the site's vocabulary.

AI Funding and Infrastructure

OpenInfer Secures Funding for Edge AI Inference

OpenInfer, founded by Behnam Bastani and Reza Nourai, has raised $8 million to redefine AI inference for edge applications. They aim to leverage their experience from Meta’s Reality Labs and Roblox in building and scaling AI systems.

AI Security Landscape

Enterprise AI Agent Vulnerabilities

Enterprises need to protect against the unique vulnerabilities of AI agents by treating them as their own identities.

The Rise of AI-Powered Cyberattacks and Cyber Insurance

Cybersecurity insurance is becoming increasingly critical as AI transforms and simplifies hackers' methodologies, leading to a reckoning in coverage.

Cisco's AI Defense

Cisco AI Defense is designed to address the security challenges created by the exponential growth of AI in enterprises.

AI-First Security Strategy

Enterprises must embrace an AI-first strategy that unifies endpoint, identity, and network security within a zero-trust framework to stay ahead of endpoint attacks.

AI Agents and Applications

Apple's Siri Upgrade

Apple's upgraded Siri, powered by Apple Intelligence, is not yet fully available despite the release of iOS 18 updates. The new Siri is expected to be more contextually aware and capable of performing actions on behalf of the user. The full form of the AI-powered Siri may arrive in later iOS 18 updates.

Rabbit's Android AI Agent

Humane discontinued its Ai Pin after less than a year, while Rabbit released a preview of its new Android AI agent capable of controlling an Android device based on user prompts. The company's wearable AI agent, Rabbit R1, is not involved in the performance of any of the tasks being demoed.

Google's Gemini App Promotion

Google is encouraging users to download its Gemini app by removing the AI tool from its iOS search app. Users are prompted to download the standalone app from the App Store to continue using Gemini.

AI Platforms and Services

AI Observability

• Arize AI offers an AI observability platform to evaluate and monitor AI products, from machine learning to generative AI. Their platform uses a "council of judges" approach, evaluating AI with different AI models and human input. Arize works with enterprises like Uber and Tripadvisor. The company has experienced rapid growth due to increased AI accessibility.

AI-Narrated Audiobooks

• Spotify now accepts audiobooks narrated using ElevenLabs' AI voice technology, potentially increasing the number of AI-narrated audiobooks on the platform. Authors can upload these audiobooks through Findaway Voices, with titles labeled as AI-narrated. ElevenLabs supports narration in 29 languages. This may spark debate within the publishing community due to concerns about the quality of AI recordings.

AI for Scientific Discovery

• Google has developed an "AI Co-scientist" version of its Gemini 2.0 large language model. This AI is designed to generate novel scientific hypotheses faster than human researchers, accelerating discovery. In a test case, Co-scientist generated a hypothesis about bacteria evolution in two days, simultaneously with human scientists at Imperial College London, whose work took a decade.

AI Model Updates and Development

Upcoming OpenAI Models

• Microsoft is preparing server capacity for OpenAI's upcoming GPT-4.5 (codenamed Orion) and GPT-5 models. GPT-4.5 is expected to launch as early as next week, while GPT-5 is anticipated in late May. GPT-5 will integrate a lot of OpenAI's technology and include more significant changes.

Humanoid Robot Control

• Figure CEO Brett Adcock unveiled Helix, a "generalist" Vision-Language-Action (VLA) model for humanoid robots. Helix combines visual data and language prompts to control robots in real-time, enabling them to perform household tasks by processing natural language voice prompts and visually assessing their environment. The company is showcasing the VLM with its 02 humanoid robot.

Artificial Intelligence

The Rise of Generative AI

The emergence of Generative AI tools such as ChatGPT in early 2023 has revolutionized various products and workflows. This was enabled by innovations in AI, especially transformer models, which eliminated the need for specific pre-training. Larger models and vast datasets have allowed AI to handle a wider range of tasks, boosting its adoption across industries.

OpenAI's Reorganization

OpenAI is planning to reorganize its for-profit arm into a public benefit corporation (PBC) in Delaware. PBCs balance the interests of shareholders with those of other stakeholders like employees, customers, and the public.

AI and Machine Learning Platforms

Nvidia's Sign Language AI Platform

Nvidia has introduced a new AI platform, Signs, designed to teach American Sign Language (ASL) and bridge communication gaps. The platform is creating a validated dataset for sign language learners and developers of ASL-based AI applications, addressing the fact that ASL is the third most prevalent language in the U.S.

Voltron Data & Accenture Partnership for GPU-Powered Analytics

Voltron Data is partnering with Accenture to revolutionize enterprise data processing using their GPU-powered analytics engine, Theseus. This collaboration aims to deliver up to 100x performance gains for AI-driven companies grappling with massive data challenges.

AI Application Development & Automation

Rise of Citizen Developers in GenAI

Experts predict that citizen developers will be responsible for creating 30% of GenAI-infused automation applications, capitalizing on their domain expertise. These developers may require training on the safe provisioning of AI models and copilot platforms. A hybrid approach is recommended, as citizen developers might not be fully prepared for bare-metal GenAI. Visual representation for tasks like UI design and easy customization through declarative models are considered vital. Considerations for design trade-offs, best practices, and compliance within enterprise environments are also important.

Khoj: Your AI Second Brain

[] Khoj is an AI "second brain" designed to augment the capabilities of large language models (LLMs). It allows users to build custom agents, schedule automations, and conduct in-depth research. Khoj can transform online or locally hosted LLMs into personal, autonomous AI assistants while ensuring privacy and security through its open-source and self-hostable nature.

AI in Recruitment

Mercor's AI-Powered Recruitment Platform

Mercor, an AI recruiting startup founded by Thiel Fellows, has secured $100 million in a Series B funding round led by Felicis, valuing the company at $2 billion. Mercor uses AI to streamline hiring processes by automating resume screening, candidate matching, AI-powered interviews, and payroll management. The platform promotes bias reduction and improved candidate selection, with companies like OpenAI already utilizing its tools. Mercor is expanding its talent pool to meet the increasing demand for professionals beyond software engineering.

Ethical Concerns in AI

Controversy Surrounding AI Voice Recreation

Netflix is facing criticism for using AI to recreate the voice of murdered travel vlogger Gabby Petito in a new documentary. While filmmakers defended their decision by stating they had the Petito family's permission and wanted to use Gabby's own words by animating the deceased, many find the use of AI voice recreation disrespectful and unsettling, sparking a debate over the ethical implications of using AI in sensitive contexts.

AI Market Growth & Expansion

OpenAI's User Growth

OpenAI has reported substantial growth, with 400 million weekly active users, a significant increase from 300 million in December 2024. Their enterprise plans are also expanding, now serving 2 million paying enterprise users, doubling the number since September 2024. Additionally, developer API traffic has doubled over the past six months. This announcement coincided with the release of competing AI technology by China’s DeepSeek, highlighting OpenAI's continued business success.

Spotify and ElevenLabs Partnership

Spotify is expanding its AI-narrated audiobook offerings by partnering with ElevenLabs, which will allow authors to use ElevenLabs' AI voice software to record audiobooks in 32 languages. This partnership is expected to significantly increase the presence of AI-recorded audiobooks on Spotify. The company believes that digital voices can make audiobook production more cost-effective for smaller authors and help revive older titles, with all AI-narrated titles clearly marked for listeners. Spotify emphasizes its continued support for human narration but sees potential in digital voices for market expansion.

Quantum Computing

Microsoft's New Chip Boosts Quantum Computing Stocks

Quantum computing stocks experienced a surge in premarket trading following Microsoft's unveiling of a new chip. This development suggests that the technology is closer to realization than Nvidia's AI chip leader previously indicated.

Artificial Intelligence

xAI's Grok 3 Now Freely Available on X

xAI has made its most advanced large language model, Grok 3, available for free on X for a limited period, even without a subscription. According to Elon Musk, the company launched its Grok 3 model family a couple of days ago, and some of the models in the family are capable of "human-like reasoning".

Artificial Intelligence

Ethical Concerns in Generative AI

The article highlights that no generative AI tool from major players is more ethical than any other. Ethical concerns are divided into model development, particularly data sourcing, and environmental impact. The data used to train these models is acquired through questionable and secretive decisions. Furthermore, the environmental impact of generative AI is significant, requiring far more energy to create and run compared to non-generative alternatives.

Autonomous Vehicles

Waymo's Expansion and the Promise of Robotaxis

Waymo, along with other autonomous vehicle (AV) companies, is expanding, raising questions about a potential transportation revolution. While fully self-driving cars are not yet available for purchase, current models offer lower-level autonomous features that require human oversight. Waymo promises safer transportation through programmed adherence to traffic laws and is operating in multiple cities with plans for further expansion.

The Risks of Current Autonomous Systems

A Tesla driver using the FSD (Supervised) mode caused a fatal accident while distracted, highlighting the risks associated with relying on current autonomous systems. Experts emphasize the driver's responsibility to take over when necessary, pointing out the clash between the technology's promise and real-world challenges and potential dangers. Skepticism about robotaxis arises from viral videos showing them making errors.

Digital Abuse and Content Removal

The Plight of Breeze Liu

Breeze Liu experienced a distressing online ordeal after an explicit video of her, filmed without her consent when she was a minor, was posted on PornHub and subsequently spread across the internet. This led to severe emotional distress and suicidal thoughts. Despite her efforts to remove the content, Microsoft delayed removing approximately 150 explicit images of Liu stored on its Azure cloud services, even after repeated pleas.

The Need for Effective Systems

Liu's experience highlights the urgent need for more effective systems to navigate and control digital abuse, especially concerning the removal of intimate images from the web when the age of the individual in the imagery is disputed. This prolonged process underscores the difficulties faced by victims of intimate image abuse.

Technology

Lenovo's Strong Quarterly Performance

Lenovo, a Chinese tech giant, announced significant gains in its quarterly revenue and net profit, exceeding market expectations. The company anticipates further benefits from integrating DeepSeek's artificial intelligence model into its product line.