AI News & Launches

As the California Assembly passes SB 942, requiring AI companies to offer free AI detection tools, a personal account of Grammarly's AI checker impacting a child's creativity unfolds.

Graphiti is an open-source project hosted on GitHub that facilitates the construction and querying of dynamic, temporally-aware Knowledge Graphs. It handles evolving relationships and maintains historical context, filling a niche for temporal aspects not covered by static document models like Microsoft’s GraphRAG. Graphiti is suitable for applications in various industries such as sales, customer service, health, and finance, offering sophisticated search and scoping features, and is powered by Python.

Mem0ai/mem0 is an open-source project on GitHub that introduces a memory layer for personalized AI, enhancing AI assistants and agents with intelligent memory abilities. It supports adaptive personalization and offers developer-friendly APIs, with applications in chatbots, customer support, and more.

Videco is the leading AI personalized video solution aimed at sales and marketing teams, boasting features like AI voice cloning and interactive buttons. With no credit card requirement and a 14-day free trial, it's designed to significantly boost your business pipeline through personalized and interactive video content.

Basejump AI empowers teams to interact with their databases using natural language, enabling seamless, real-time access to data without the need for intricate SQL queries. With Basejump, data-driven insights are made accessible to decision makers, enhancing productivity and efficiency across various business tasks.

PathPilot offers a refined approach to session replay analysis, transforming extensive recordings into actionable highlights that uncover UI issues, user interactions with new features, and identifies potential friction points for valuable customers. It aims to enhance UX through tailored recommendations based on actual user behavior.

FlexClip is an online tool suite designed for easy video creation and editing, offering a range of services including AI-driven image to video transformations, video compressing, and more. It caters to both personal and professional needs, enabling users to create videos for various purposes like marketing, education, and entertainment.

GPT Builder Tools, backed by Y Combinator, offers solutions to generate revenue from GPT applications in the OpenAI Store without requiring coding skills. With a PayPal account, users can easily monetize their GPTs using a donation or paywall system, and a 30-day free trial is available.

PicStudio.ai is an AI portrait app that generates realistic photos of individuals for use on social media platforms like LinkedIn, Facebook, and Instagram. Offering various photo packs and promising delivery within 30 minutes, the service provides a convenient alternative to professional photo shoots.

PR-Agent: AI-Powered Code Reviews & Chat enhances GitHub pull request interactions by providing automated code reviews, chat functionality within PRs, and suggestions for code improvement, making the code review process more efficient and thorough.

Edmigo offers a Complete DSA Course with a personal AI Tutor designed to help users solve 75 of the most frequently asked interview questions. It features learning materials crafted by Google Engineers and provides hands-on experience by integrating with LeetCode.

OpenAI and Anthropic have signed agreements with the US AI Safety Institute under NIST to provide their forthcoming AI models for safety evaluations. This collaboration aims to develop safety standards and reinforce responsible AI development. The move highlights the tech industry's engagement in voluntary AI safety measures, despite there being no regulatory compulsion.

This article emphasizes the importance of engineering over just AI. It discusses the overhyped space of GPT-like code-generating agents and stresses the need to address engineering problems before AI can be effectively integrated. The piece highlights the significance of the structural context for codebases over semantic embeddings for functional code generation.

No More Alone' is a unique concept on IcebreakingLab that encourages the use of icebreakers and AI to make new friends. It prompts users to get a random question for social contexts or directly engage with an AI for conversations.

OpenAI has reportedly reached out to the British Parliament, seeking permission to use copyrighted materials for training its AI models. The AI firm argues that without this access to copyrighted content, it cannot train large language models effectively. Concerns arise as legal actions have been taken against OpenAI, such as the Authors Guild suing OpenAI and similar issues raised by major publications.

Hestus is revolutionizing hardware development with its AI-powered CAD tool, designed to speed up the design process by automating mundane tasks. This allows engineers to focus on the creative aspects of design. The platform seamlessly integrates with existing CAD environments, including Autodesk Fusion 360.

The Dutch data protection authority fined Clearview AI €30.5M ($45.6M Cdn) for its 'illegal database' of faces. This comes in violation of the EU's General Data Protection Regulation. Clearview, claiming non-subjection to EU laws, faces additional noncompliance penalties up to €5.1M.

This paper presents a study on the compute-optimal strategy for training language model (LM) reasoners. Contrary to the common practice of using strong but expensive models for synthetic data generation, the authors show that weaker, cheaper models can yield better training data. They evaluate the data from both strong and weak models across several metrics and find that models trained on weak model-generated data outperformed their strong model-trained counterparts across various benchmarks.

Jamboss is a creative AI music generation platform allowing users to turn ideas into custom songs with various styles, including rock, opera, and reggae. It provides copyright-free downloads and the ability to share creations easily. It's been received positively by users and is rated highly on the app store.

Spellar Pro is offering a 30% discount, available through their Product Hunt page. This tool aims to aid in productivity and is being promoted with a significant price reduction for a limited time.

Researcher Sander Dieleman explores the close relationship between diffusion models and autoregressive models in image generation, asserting that diffusion models use approximate autoregression in the frequency domain. The article also touches on the implications for sound and language processing, and speculates on the future of generative models for multimodal inputs.

Eduardo Blancas discusses his exploration of OpenAI's structured outputs feature to create an AI-assisted web scraper. By combining Python's Pydantic models with GPT-4o, he experimented with parsing HTML tables and generating XPaths for efficient data extraction. Despite challenges with complex tables and varying results, the approach shows promise for the future of web scraping tools.

A study from the MIT Media Lab, titled 'AI-Implanted False Memories', examines how AI through suggestive questioning can induce false memories, particularly in the context of witness interviews. The study involved 200 participants exposed to a crime video, followed by AI-driven questioning with misleading content. The use of a generative chatbot proved to significantly increase both the formation and the confidence in false memories when compared to control or survey methods, raising ethical concerns for AI's application in sensitive areas.

Coursely.ai is an online platform that offers a wide range of AI and technology-focused courses designed to equip learners with skills in various tech domains including data science, machine learning, and artificial intelligence.

ReplyToAppReviews.com offers a centralized platform for developers and marketing teams to efficiently manage and respond to app reviews. It is designed to aggregate reviews from various app stores, giving users the capability to reply to them directly through the dashboard and analyze feedback trends.

Maxium AI is an application designed to measure developer productivity beyond just lines of code or commit quantities. It offers a standardised evaluation mechanism, tracks shipping velocity, and compares engineer performance against industry peers. The app is Github-native and ensures privacy by encrypting code changes.

Pallie is an AI companion designed to act as a well-being advisor, empathetic coach, playful friend, and efficient organizer. It integrates with platforms like iMessage and Telegram to provide personalized advice and context-aware interactions.

VideoFaceSwap.ai offers an AI-powered tool for creating deepfake videos by swapping faces in videos. Users can upload an image and video, and the AI performs the face swap, making it suitable for copyright-free ads, fun with friends, and privacy protection.

Procreate has announced that they will not integrate generative AI into their iPad illustration app, responding to a part of the art community backlash over AI's ethical implications. CEO James Cuda emphasizes Procreate's human-centered approach in designing products, disagreeing with the consequences of AI use on the industry and artists.

As AI continues to advance, the acquisition of data for training generative models is facing legal and ethical challenges. Websites are increasingly restricting access to their content through robots.txt, limiting the availability of high-quality, fresh data essential for training AI. The article 'AI Has Created a Battle Over Web Crawling' discusses how this could affect the performance of AI systems and the potential legal battles that might ensue over data use.

Research challenges the assumption that Large Language Models (LLMs) understand language, arguing they fail to capture human linguistics and cognition. The paper states LLMs do not embody human interaction or the precariousness of language. These findings question LLM applications in critical domains without rigorous testing.

Lumora provides advanced tools to manage, optimize, and test AI prompts, ensuring peak performance for a variety of AI platforms. With its intuitive interface, prompt management, and co-work features, Lumora caters to teams seeking to enhance the results of text, image, and video AI outputs. Offering a rich selection of AI integrations and customizable plans, it's ideal for organizations aiming to streamline AI workflows.

Hello Literature is an innovative app developed by the Humy.ai engineering team, aimed at bringing characters from literary classics to life through interactive conversations. Designed for educators, parents, and lifelong learners, it fosters engagement with literature and enhances reading skills and critical thinking.

Fill A Form AI offers a one-click form-filling solution using a smart AI agent. It automatically identifies form inputs on websites, learns from past filled forms, and efficiently finds answers from user data, streamlining the form-filling process and saving time.

TutorEva is an innovative online platform providing personalized and flexible tutoring services to students of all ages, featuring experienced educators and tailored learning plans.

At the DataGrail Summit 2024, leaders warned about AI's rapid advancement outpacing security measures. Instacart's and Anthropic's CISOs highlighted the need for robust AI security to manage risks and protect consumer trust.

Nvidia has delivered an impressive earnings report with a 122% sales rise and profits doubled. Despite this, Nvidia's shares slumped 7%, reflecting Wall Street's recalibration of expectations around AI's actual value. This marks a shift as initial excitement wanes, prompting a more realistic assessment of AI's revenue-generating potential amid concerns that the AI boom could be a bubble set to burst.

Anil Dash criticizes a yet-to-air Oprah special on AI, anticipating a biased presentation focusing on commercial interests. He foresees a lack of historical context, labor perspectives, consent discussions, and accountability, framing it as a promotion rather than a fair analysis.

Nvidia's revenue significantly relies on four anonymous 'whales,' as nearly half its $30 billion turnover is attributed to their purchases. The identities of these customers are undisclosed but are speculated to include tech giants such as Amazon, Meta, Microsoft, Alphabet, OpenAI, or Tesla. Their spending is mainly on Nvidia's data center chips, like the AI chip H200 used for training AI models. Despite Nvidia's CEO Jensen Huang claiming diversification, the concentration on a few customers raises concerns about the sustainability of growth.

Ted Chiang argues that art cannot be created by AI as it involves making numerous choices, which is beyond current AI capabilities. He illustrates this through Roald Dahl's fiction-writing machine story and compares generative language models to auto-complete, lacking the intention to communicate. AI-generated text and art, he suggests, lack the nuanced decision-making inherent in human creations. He concludes that meaningful communication, whether in art or language, is derived from human intention and interaction.