Research

Research

TikTok Releases Boximator for Fine-Grained Motion Control in AI-Generated Videos

• 

Chris McKay• February 14, 2024 • 1 min read

Under the hood, Boximator functions as a plug-in that infuses existing video synthesis models with these user constraints. It trains an additional module while freezing base model weights, enabling straightforward integration with state-of-the-art systems.

Cohere Open Source Research

Cohere For AI Opensources Aya, a 101-Language AI Model

• 

Richard Banfield• February 13, 2024 • 2 min read

Aya is an open science project that aims to build a multilingual language model via instruction tuning that harnesses the collective wisdom and contributions of people from all over the world.

Research

An AI That Learns Like A Baby

• 

Chris McKay• February 5, 2024 • 3 min read

The system, dubbed the Child's View for Contrastive Learning (CVCL) model, was fed 37,500 utterances captured from 61 hours of first-person video recorded by a baby named Sam from ages 6 months to 2 years old.

OpenAI Research

Does ChatGPT Make Developing Bioweapons Easier?

• 

Chris McKay• January 31, 2024 • 2 min read

The study aims to determine whether increased information access using language models like GPT-4 could help bad actors more easily compile knowledge for crafting biological agents vs. existing online resources.

Research AI Tech

InstantID Allows You to Easily Create Consistent Characters with Generative AI

• 

Chris McKay• January 25, 2024 • 3 min read

With just a single reference image, InstantID can generate a similar persona in any style with high fidelity.

Google Research

Lumiere is an Impressive New Text-to-Video AI from Google

• 

Chris McKay• January 24, 2024 • 2 min read

This model is designed to create realistic, diverse, and coherent motion in videos, a task that has historically been challenging in the field of artificial intelligence and computer vision.

Research

TikTok's Depth Anything Models Sets New Standards for Robust Image-Based Depth Estimation

• 

Chris McKay• January 22, 2024 • 2 min read

In zero-shot evaluations on datasets like KITTI, NYUv2 and Sintel, Depth Anything models significantly outperformed the previous state-of-the-art MiDaS v3.1 models on both absolute relative error and accuracy metrics.

Google Research

Making AI More Self-Aware: Google Introduces ASPIRE Framework to Boost Confidence in LLM Responses

• 

Chris McKay• January 19, 2024 • 2 min read

ASPIRE enables models to output an uncertainty warning like "I don't know!" along with the prediction if the confidence score is low.

Google Research

Google DeepMind's AI Masters Geometry at the Highest Levels

• 

Chris McKay• January 17, 2024 • 3 min read

The AI exhibited expertise on par with International Mathematical Olympiad (IMO) gold medalists on a benchmark test of 30 challenging classic geometry problems.

Anthropic Research AI Safety

Deceptive "Sleeper Agent" AIs Can Slip Past Sophisticated Safety Training

• 

Chris McKay• January 12, 2024 • 3 min read

The ability of LLMs to retain deceptive behaviors despite safety measures isn't just a technical loophole; it’s a paradigm shift in how we perceive AI reliability and integrity.

Google Research Healthcare

Google Shares Research Progress on AI that Converses, Reasons and Diagnoses Like Your Doctor

• 

Richard Banfield• January 12, 2024 • 2 min read

AMIE demonstrated stronger diagnostic accuracy and outperformed doctors on most metrics of consultation quality in both diagnostic accuracy and clinical communication.

Google Robotics Research

Google DeepMind Unveils Latest Research in Advanced Robotics with AutoRT, SARA-RT, and RT-Trajectory

• 

Chris McKay• January 4, 2024 • 2 min read

These advancements build on the foundational Robotics Transformers work, aiming to streamline decision-making processes, enhance environmental understanding, and improve navigational capabilities of robots.

Robotics Research

Stanford Unveils a New Low-Cost Mobile Robotics Training System

• 

Chris McKay• January 4, 2024 • 2 min read

Mobile ALOHA can autonomously complete complex mobile manipulation tasks such as sauteing and serving a piece of shrimp, opening a two-door wall cabinet to store heavy cooking pots, and lightly rinsing a used pan using a kitchen faucet.

Research AI Tech

JPMorgan Introduces DocLLM for Better Multimodal Document Understanding

• 

Chris McKay• January 2, 2024 • 2 min read

The model may enable more automated document processing and analysis for financial institutions and other document-intensive businesses going forward

Google Research

Google AI Unveils VideoPoet: A Zero-Shot Video Generation Large Language Model

• 

Richard Banfield• December 19, 2023 • 3 min read

VideoPoet's capability to generate videos with diverse motions and styles, tailored to specific text inputs, showcases its advanced understanding of both content and context.

Research Media & Entertainment

Upscale-A-Video is A New AI Model for Real-World Video Super-Resolution

• 

Chris McKay• December 12, 2023 • 1 min read

At the core of Upscale-A-Video is an ingenious text-guided latent diffusion framework tailored for the unique demands of video processing.

Research

TAP is a New Method That Automatically Jailbreaks AI Models

• 

Chris McKay• December 7, 2023 • 3 min read

On average, TAP jailbroke GPT-4 in just 28 queries, circumventing safety measures and prompting responses on building homemade explosives or distributing pirated software.

Microsoft Research Accessibility

Microsoft Researchers Advance Sign Language Accessibility Through Community-Sourced Data

• 

Chris McKay• December 4, 2023 • 2 min read

This approach aims to remedy the limitations of previous datasets that were either scraped from the internet without consent or recorded in labs, lacking diversity and real-world applicability.

Google Research

Google DeepMind Unveils GNoME to Use AI to Accelerate Material Science

• 

Chris McKay• November 29, 2023 • 2 min read

GNoME has unveiled an astonishing 2.2 million new crystals, including 380,000 stable materials with immense potential for future technological applications.

Stability AI Research

Stability AI Unveils SDXL Turbo for Real-Time Text-to-Image Generation

• 

Chris McKay• November 28, 2023 • 2 min read

The model uses a new technique called Adversarial Diffusion Distillation which allows it to have many advantages shared with Generative Adversarial Networks such as single-step image outputs, while avoiding artifacts or blurriness often observed in other distillation methods.

Stability AI Research

Stability AI Releases Stable Video Diffusion

• 

Chris McKay• November 21, 2023 • 2 min read

Stability is open-sourcing Stable Video Diffusion and an image-to-video model, for research purposes.

Microsoft Open Source Research

Microsoft Releases New Orca 2 Small Language Model with Enhanced Reasoning Capabilities

• 

Chris McKay• November 20, 2023 • 2 min read

Comprehensive benchmarks show Orca 2 significantly outperforming other models of equivalent size on metrics related to language understanding, common sense reasoning, multi-step math problems, reading comprehension, summarization, and more.

Meta Research

Meta's New AI Tech Can Edit Images with Text and Convert Them to Video

• 

Chris McKay• November 16, 2023 • 2 min read

Building on their prior work in image and video generation, these models showcase impressive capabilities in high-quality, diffusion-based text-to-video generation and controlled image editing using text instructions.

NVIDIA Research

NVIDIA Leverages AI to Assist with Semiconductor Design

• 

Chris McKay• October 30, 2023 • 2 min read

NVIDIA researchers customized LLaMA by training it on 24 billion tokens derived from internal documents, code, and other textual data related to chip design.