AI roundup #7 👩‍💻

AI roundup #7 👩‍💻

🧠 Google DeepMind: merging AI research teams


On April 20th, blog posts from Google & DeepMind confirmed the merge of the two big AI teams, Brain [from Google's research team] and DeepMind [Alphabet's independent company].

Earlier today we announced some changes that will accelerate our progress in AI and help us develop more capable AI systems more safely and responsibly.

Some state that this is the byproduct of Google in 'panic' owing to rapid developments & emerging competition in the AI niche. DeepMind and Google Brain have been forced to pause the age old rivalry and work on a new product, stated to be a ChatGPT competitor called "Gemini".

Sources: The Information, Ars Technica

🦾 Athena: Microsoft's very own AI chips project

Microsoft has reportedly been secretly developing its own AI chips since 2019, internally referred to as "Athena",  aiming to avoid reliance on Nvidia's costly chips for large language models such as GPT-4.

Although there is no clarity if the chips may be available for Azure customers yet, they have significant potential to reduce costs as Microsoft rolls out AI-powered features in various apps.

Microsoft also has a roadmap for multiple future generations of the chips. Other tech giants, including Amazon, Google, and Meta, also have their own in-house chips for AI.

Sources: The Verge, The Information

✨ Synthesis AI: text to 3D tech

Credits: Barsee

Synthesis AI, a San Francisco-based startup, has created a groundbreaking text-to-3D technology that generates highly detailed and lifelike 3D digital humans from text prompts.

Building on years of work with synthetic data, Synthesis AI’s 3D  beta release introduces photoreal digital humans for use in gaming, visual effects, and synthetic data generation.

This technology is said to be the first of its kind and will support advanced AI applications by providing perfectly labeled synthetic data for machine learning models. Although still in its early stages, it has the potential to revolutionize industries such as gaming, VR/AR, filmmaking, advertising, training simulations, and more.

Sources: Venture Beat, Unite.AI

👂 Hear better with Amazon's Dialogue Boost

Amazon's Prime Video has introduced Dialogue Boost, a new accessibility feature that allows users to increase the volume of dialogue relative to background music and effects.

The feature is available on select Amazon Originals and can be accessed on all devices that support the Prime Video experience. Dialogue Boost was built with the needs of hard-of-hearing customers in mind but can be used by anyone to suit their personal listening preferences.

Sources: Amazon, Gizmodo

🪢 StableLM: Stability AI's latest launch


On 19th April, Stability AI has launched StableLM, an open-source language model suite consisting of 3-7 billion parameter models, with larger models to follow. The models can generate text and code and have high performance in conversational and coding tasks.

StableLM builds on earlier open-sourced models and is transparent, accessible, and supportive.

The fine-tuned models are for research use only, and the suite is available on GitHub.

Stability AI is also working on crowd-sourced programs and open-source datasets.

Sources: Venture Beat, Stability AI

Also read:
AI Roundup #6: Last Week in AI