Good morning, human brains. Welcome back to your daily munch of AI news.

Here’s what’s on the menu today:

• Stability Unveils StableStudio 🎨 

  Stability AI released an open-source text-to-video tool.

• Meta’s New Speech-To-Text AI 🗣

  Meta launched an AI model that performs speech-to-text (and vice versa) for 1100+ languages.

• Anthropic Raised $450M 💸

  Google, Salesforce, Zoom invest millions into a ChatGPT Competitor.

P.S. Bot Eat Brain passed 10,000 subscribers today. Thank you for reading and sharing this newsletter with your friends!

Sponsor Bot Eat Brain | New here? Subscribe!

APPETIZER

Stability AI unveils StableStudio 🎨

Stability AI, the company best known for its open-source image generation tool Stable Diffusion, just announced a new entrant to its growing portfolio of open-source projects.

Two weeks ago, they announced StableStudio, an open-source counterpart to its existing text-to-image tool, DreamStudio.

What's the big deal?

One of the most sophisticated image-generation AI tools is now fully open-source on GitHub. That means any developer can now contribute code to the project or copy it and make their own unique version.

Stability also revealed plans to incorporate a ChatGPT-style interface.

Take: The world is building powerful new AI tools that few understand. Many of these tools are being built privately and are closed-source. We like when we can look under the hood. Kudos to Stability.

Countertake: Open-sourcing AI is kind of playing with fire. If AI gets out of control it would be very difficult to stop it. How many people do we want to have access to the big red button?

BUZZWORD OF THE DAY

WITH COMMANDBAR

Custom ChatGPT For Any Website

HelpHub is an AI chat + search for any website or web app.

Just add source content (via URL, copy/paste, or sync with a CMS). HelpHub’s chatbot trains on your website’s content and will answer questions based solely on that content.

The embeddable widget offers two super easy ways to help your customers: (1) a chatbot interface for users to ask questions directly, and (2) a search interface that allows them to search through and navigate to your resources for more details.

MAIN COURSE

Meta’s massive new speech-to-text language model

Meta’s MMS (Massively Multilingual Speech) model is capable of speech-to-text and text-to-speech for 1107 languages, and identification for nearly 4000.

Hasta la vista baby…

Why do we care?

The new tech promises to broaden access to information and technology across the globe. There are over 7,000 known languages worldwide, and many of them are at risk of completely disappearing within our lifetimes. Giving machines the ability to process spoken language unlocks the internet to those who were unable to do so before — whether that’s because they communicate in a spoken-only language, or because of a lack of translations.

How does it do all that?

Audio from religious texts. Because of their global popularity, texts like the Christian Bible were a good source for capturing such a wide spread of languages.

Like other language models, MMS employs self-supervised learning (where the model learns from unlabeled data without human intervention).

MMS models show substantial improvement over OpenAI’s Whisper, both in word error rates and language coverage.

For more details on the MMS project, visit their blog post HERE.

Our Take: There are still challenges with language coverage and accuracy. And it’s trained on religious texts — how does it perform in more secular settings? All that said, Meta’s making strides in language preservation and inclusivity — so, amen to that.

A LITTLE SOMETHING EXTRA

Claude vs ChatGPT 🔥

Anthropic AI, a startup focused on developing ethical AI, recently raised $450 million in Series C funding, with investments from tech giants Google, Salesforce, and Zoom.

What's the big deal?

Anthropic’s AI assistant, Claude, aims to be a safe and beneficial tool for consumers and businesses.

What's going down:

1/ Ex-employees of OpenAI started Anthropic as a rival company. They're using the new cash to enhance their product offerings to compete with OpenAI.

2/ Anthropic claims that Claude is more transparent about its capabilities and limitations, can handle complex conversations, and follow precise instructions.

3/ Major companies like Notion, Quora, and DuckDuckGo have tested Claude in closed alpha and reported positive feedback.

4/ There's controversy surrounding Anthropic's biggest investor, Alameda Research, due to its involvement in the Sam Bankman-Fried scandal/FTX bankruptcy case. The impact on Anthropic remains unclear.

Our take: Increasingly large cash injections — yup, expect the trend to continue. OpenAI has a commanding lead, but all the big tech players are in the AI arena now. Grab some popcorn.

MEMES FOR DESSERT

YOUR DAILY MUNCH

Flamingo: Google DeepMind’s AI language model makes descriptions for YouTube Shorts

GrimesAI-1: Pop singer & producer Grimes created a tool that allows other producers to use her voice, using generative AI

MIT Researchers use AI to match similar materials in images

UCSD: AI rejuvenates gene activation research and uncovers rare DNA sequences

Startup News

Neeva announces it’s shutting down

Adobe to integrate Firefly into Photoshop

Research

A-Z’s of LLMOps: Unraveling the tasks that drive Large Language Model Success

Drag Your GAN: Interactive Point-based manipulation on the generative image manifold

Microsoft’s CTO open-sources his social media post generator powered by ChatGPT.

Tools

ChatGPT for IOS

DeepGram Speech-to-text API

Speechify AI Voiceover Generator

HelpHub: ChatGPT-ify your site [Sponsored]

TWEET OF THE DAY

NVIDIA’s stock jumps after financial forecasts show AI’s greening effect on the tech sector.

AI ART-SHOW

Until next time 🤖😋🧠