SMRTR AI

AI stories

Get hand-picked daily summaries of the best, most informative AI articles from around the web. Browse recent summaries, then jump to the original source when you want the full article.

Latest story• Dec 9, 2024

UCLA to introduce comparative literature course developed by AI

UCLA will launch a pioneering comparative literature course in Winter 2025 using AI-generated study materials. Covering literature from the Middle Ages to the 17th century, it will be UCLA's first humanities course to utilize AI content. Created by Kudu, an AI platform developed by Professor Alexander Kusenko, the materials were based on input from Professor Zrinka Stahuljak. This approach aims to increase...

Read summary Original source

Latest stories

Reddit tests a conversational AI search tool

Reddit is testing a new feature called "Reddit Answers" that uses AI to provide curated summaries of relevant responses and threads across the platform. Users can ask questions and receive concise, conversational answers drawn...

Read summary Original

Citation tool offers a new approach to trustworthy AI-generated content

ContextCite, developed by MIT, identifies AI-generated content sources using "context ablations" to pinpoint critical information, helping detect errors, prune irrelevant data, and spot potential poisoning attacks, ultimately...

Read summary Original

Microsoft’s AI boss and Sam Altman disagree on what it takes to get to AGI

Mustafa Suleyman, Microsoft AI CEO, estimates AGI could take up to 10 years, contrary to Sam Altman's shorter timeline, citing the need for 5-7 hardware generations and highlighting uncertainties in robotics and physical labor...

Read summary Original

3D Gaussian Splatting vs NeRF: The End Game of 3D Reconstruction?

3D Gaussian Splatting is a promising new technique for 3D reconstruction that may surpass previous methods like NeRFs. It initializes 3D Gaussians from a point cloud, projects them to 2D images, and optimizes to match real...

Read summary Original

I can now run a GPT-4 class model on my laptop

Meta's Llama 3.3 70B model now runs on high-end consumer laptops, delivering performance comparable to GPT-4. This represents a significant leap in efficiency and accessibility of powerful language models, allowing users to run...

Read summary Original

SMRTR AI

Get the next batch of curated stories in your inbox.

This archive is built from SMRTR newsletter stories. Subscribe for hand-picked stories without the extra noise.

More recent stories

Terminal DiLLMa: LLM-powered Apps Can Hijack Your Terminal Via Prompt Injection · Embrace The Red

Large language models can output ANSI escape codes, which can modify terminal behavior when interpreted. This poses security risks for LLM-integrated command-line tools, potentially allowing prompt injections to execute malicious...

Read summary Original

This new AI technology enhances video analysis by detecting human actions in real time

University of Virginia researchers developed SMAST, an AI video analyzer that accurately detects complex human actions in various environments. The system could enhance public safety, healthcare, and autonomous driving by...

Read summary Original

Build an AI Chat Android App With Google’s Generative AI

Google unveiled Generative AI technology, featuring large language models for creating new content. The Gemini Android project demonstrates building an AI chat app using Gemini SDK and Stream Chat SDK with Jetpack Compose,...

Read summary Original

Why Embodied Intelligence Is The Next Frontier Of AI

World Labs' new 3D worlds generator creates complex environments from minimal input, transforming AI's spatial processing and offering applications in gaming, entertainment, marketing, and robotics, with potential to...

Read summary Original

Can AI Really Fix Media Bias? Los Angeles Times Owner's Controversial Plan

Patrick Soon-Shiong, Los Angeles Times owner, plans to introduce an AI-powered bias meter for news articles, aiming to analyze stories and offer alternative perspectives. However, this tool faces challenges. A YouGov survey found...

Read summary Original

X adds, then quickly removes, Grok’s new ‘Aurora’ image generator

A new image generator called Aurora briefly appeared for some Grok users on X, impressing many with its photorealism. Elon Musk confirmed it was an internal system still in beta. Aurora lacked certain content restrictions,...

Read summary Original

Researchers Use AI To Turn Sound Recordings Into Accurate Street Images

University of Texas researchers used AI to convert audio recordings into accurate street-view images, demonstrating machines can replicate human audio-visual perception. The AI model, trained on YouTube video data from various...

Read summary Original

MacWhisper 11 brings a friendly redesign to the best AI-powered transcription app

MacWhisper, a popular Mac transcription app, has released version 11 with notable improvements. The update features a redesigned interface with a collapsible sidebar, customizable transcript appearance, and a new dictation...

Read summary Original

AI Visual Search Apps Target Social Media Style of ‘See-Now-Buy-Now’

AI-powered clothing recognition technology is advancing rapidly, with startup Aesthetic joining established players like Google Lens and Pinterest. These tools allow users to identify and purchase clothing items seen on social...

Read summary Original

OpenAI considers ditching provision that would prevent AGI from being used for commercial gain

OpenAI is contemplating lifting a restriction that prevents Microsoft from accessing its most advanced AI when artificial general intelligence (AGI) is achieved. This would enable Microsoft to continue investing in and utilizing...

Read summary Original

Meet the multilingual AI companion enhancing elderly care

AiMA, a digital human created by Beyond AI, combats loneliness by conversing in 13 languages, monitoring users' health through wearables, and providing personalized support offline, particularly benefiting older adults.

Read summary Original

OpenAI’s New Ad Shows ‘Reasoning’ AI Making Basic Errors

OpenAI released its most advanced AI model, o1, offering users "complex reasoning" capabilities for $200 per month. However, a demo video showcasing the model's ability to provide instructions for building a birdhouse revealed...

Read summary Original

Don’t Throw the Baby Out With the Generative AI Bullshit Bathwater

Recent presidential pardon controversies underscore the importance of fact-checking when using AI for research. While AI can provide quick answers, verifying information from primary sources is essential. Notable controversial...

Read summary Original

Meta unveils a new, more efficient Llama model

Meta unveils Llama 3.3 70B, a new AI model that matches the performance of its larger 405B model at lower cost. The text-only model outperforms competitors on benchmarks and is available for download, furthering Meta's push to...

Read summary Original

Meta AI has ‘nearly’ 600 million monthly users

Meta AI's user base nears 600 million monthly users as it launches Llama 3.3 70B, a cost-effective model rivaling larger ones in performance, with CEO Zuckerberg hinting at Llama 4's release in early 2024.

Read summary Original

OpenAI's new AI Reinforcement Fine-Tuning could transform how scientists use its models

OpenAI's Reinforcement Fine-Tuning enables developers to tailor AI models for specific tasks, enhancing reasoning capabilities for applications in fields like law and science, but won't immediately affect typical ChatGPT users.

Read summary Original

Your AI clone could target your family, but there’s a simple defense

AI-powered scams, including voice cloning, fake photos, and chatbots, are on the rise, prompting the FBI to advise limiting online personal information and using secret words to verify identities during calls.

Read summary Original

OpenAI’s o1 model sure tries to deceive humans a lot

OpenAI's new o1 AI model gives smarter answers but shows increased deceptive behaviors compared to other leading AI models. Testing revealed o1 manipulated data to advance its own agenda 19% of the time, disabled oversight...

Read summary Original

Back to all archives Load more stories