AI Valley
Posts
🐶 ChatGPT gets a humanoid body

🐶 ChatGPT gets a humanoid body

Google just announced SIMA AI

Barsee
14 Mar

Together with

_{Sign up}_|_{Follow us on X}_|_Sponsor

Happy Saturday, AI family!

🧵 Here’s what I got today:

🤖 ChatGPT gets a humanoid body
😋 Google just announced SIMA AI
🧑‍💻 The world's first AI software engineer
🔍️ + new AI Tools, Resources, and other news

Ready, Set, Go: Today’s read should be about a 3.5-minute read.

TOP STORY

🤖 ChatGPT gets a humanoid body

Two weeks ago, OpenAI signed a collaboration agreement with Figure AI to "develop next-generation AI models" for the robots as it believes they can "help in everyday life."

Just yesterday, the Figure 01 founder shared this interesting update.

With OpenAI, Figure 01 can now have full conversations with people.

It can:

describe its visual experience
plan future actions
reflect on its memory
explain its reasoning verbally
use common sense for decisions, like knowing where dishes go.

How it works:

The robot uses its cameras/microphones to gather images/speech, which are then analyzed by OpenAI's AI model, which understands both visuals and text.

Google AI

😋 Google just announced SIMA AI

SIMA is the world's first AI agent that can perform tasks across diverse video game worlds like a human.

How it works:

It learns from pictures and videos to know how to act in 3D worlds without needing special access to the game's code.

Users just show it what's on the screen and tell it what to do using prompts.

SIMA can then play the game by using a keyboard and mouse, just like a person would.

To train and test SIMA, Google has worked with eight game studios and used nine different video games:

No Man’s Sky
Teardown" by Tuxedo Labs

Together with Octaml

🔮 Explore the evolution of AIs with a time capsule

In the fast-paced world of AI, keeping up with the latest LLM releases is crucial for builders to deliver cutting-edge experiences and minimize costs. With OctoAI’s new LLM Time Capsule, you can:

Witness the exponential growth of text-gen models, from 14,000 to 61,000 in just 8 months
Discover the rise of Mistral- and Mixtral-based fine tunes as alternatives to GPT-3.5-Turbo
View the massive impact of model releases on the industry with a single toggle

Explore 60,955 text AI models here

TRENDING AI

🧑‍💻 The world's first AI software engineer

Cognition AI just introduced Devin, the world's first AI software engineer.

It is capable of coding, and creating websites and software through a single prompt, designed to work alongside human engineers.

What it does:

Devin can easily train and fine-tune its own AI models (see demo)
Taking up side hustles from freelancers from Upwork (see demo)
Devin can learn how to use new technologies (see demo)
Devin can build and deploy apps end-to-end (see demo)

A new benchmark shows:

Source

It's far better than ChatGPT 4, GPAT 3, and Claude combined in terms of coding/software engineering.

AUDIO AI

🔉 Pika launched sound effects

Pika Labs has added the ability to create sound effects from a text prompt for its generative artificial intelligence videos.

Now you can make bacon sizzle, lions roar, or add footsteps to the video of someone walking down the street.

Pika’s use cases:

Pika introduced lip-syncing just under two weeks ago, making it better for both creators and businesses.
It's one of the first platforms where users can create complete AI videos, including sound effects, voiceovers, and visuals, all in one place.

Listen to the demo here