Bot Eat Brain
Posts
Google I/O 2023

Google I/O 2023

PLUS: Meta ImageBind

Hilbert Publications & Anthony Castrio
May 11, 2023

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

Here’s what’s on the menu today:

Meta’s Multimodal AI ♾️
Meta announced a multi-modal AI with six(!) input types.
Google I/O 2023 🏅
Google showcased a suite of new AI products and features.
Bard goes public 📢
Google removed the waitlist for Bard + added new features.

Sponsor Bot Eat Brain | New here? Subscribe!

APPETIZER

Meta’s multimodal model 🔉

Meta just announced ImageBind, an open-source AI model that links text, sounds, images, temperature, and movement readings into a single powerful tool.

It’s the first mainstream model to combine so many different data types together and ImageBind promises to convert seamlessly from one “modality” to another.

Imagine this: You could ask ImageBind to compose a song based on a picture of your Mother for Mother’s Day, or have it generate images based on a song you wrote about her yourself.

Meta plans to add other streams of sensory input soon, including touch, speech, smell, and brain fMRI signals, features which would be a huge leap toward generating the immersive “Metaverse” Zuckerberg promised to build when he renamed the company.

BUZZWORD OF THE DAY

Modality

The way something happens or is experienced.

You can think of each of your senses as a seperate modality: sight, touch, taste, smell, etc.

Each gives you a unique and different way to experience the world. Your brain combines these different inputs into a single cohesive experience.

A multi-modal ai-model is a model which is simlary able to process information from multiple different sources, synthesize those inputs, and generate a cohesive result.

FROM OUR PARTNERS

A community for bootstrapped startup founders 🌐

Indie Worldwide is a social group and virtual incubator for bootstrapped startups. Get access to an actively moderated Slack group, members-only events, and 1-1 intros to other founders making similar money.

What you'll find at Indie Worldwide:

Passionate founders and indie hackers to bounce ideas off of.
Business partners, clients, and beta testers.
Motivation and support from people going through the same journey.
A global network impossible to create if you only meet people in your own city.

MAIN COURSE

Everything at Google I/O 🤖

Google I/O 2023 was last night, and AI was at the forefront of their giant suite of announcements. Buckle up, here’s everything you need to know:

1/ Google Duet

Google’s AI integration tools in all Google Workplace apps - this is Google’s version of Microsoft Copilot. It’ll help you draft docs, spreadsheets, and emails. It can generate images and complex projects from scratch.

Duet in action

2/ Gmail, “Help me write this…”

Type in a prompt, hit create, and a full email draft appears. You won’t need to search through email chains or really with customer support again - Gmail's AI can generate emails that pull context from other conversations.

3/ Search 2.0

The front page of Google will now scrape the web and give nuanced answers to conversational questions.

SEO is about to be transformed

4/ Immersive View in Maps

With Immersive View’s AR for routes in Google Maps, you can see your whole trip in advance + watch projections for traffic and weather. Rolling out this summer to 15 cities.

5/ Google Bard’s Rehaul

Bard’s allegedly now more precise, dramatically better at code generation, and seamlessly integrates with the Google toolkit. It’s also publicly available in 180+ countries without a waitlist now.

Bard in action: Research + Citations + Mapping

6/ Google + Adobe

Google’s teaming up with Adobe’s Firefly team to help Bard generate images.

Bard generating unicorns and cakes.

7/ Magic Editor in Photos

Google Photos can now reposition your elements in photos - even when they aren’t there completely - by automatically recreating parts of images.

8/ PalM 2

Google announced a more polished version of its in-house GPT competitor, PalM 2. It’ll generate text in most languages, code, and reason. But it thrives with specialized knowledge, like with cybersecurity and medical research tests.

Watch the full keynote here, a 16-minute recap, or Google’s I/O blog.