Midjourney v5.1

PLUS: US Govt 🤝 AI safety

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

Here’s what’s on the menu today:

  • The White House talks AI 🎒 

    The president met with big tech CEOs to discuss AI & launched a 9-figure fund to contribute. [jump to]

  • Midjourney v5.1 is here 🖌️ 

    Midjourney came out with a new update to v5, and its a major jump in both quality and performance. [jump to]

  • Sam Altman at MIT 🏫 

    Sam Altman was interviewed about the future of GPT-4 by Lex Friedman and an MIT prof yesterday. [jump to]

APPETIZER

US Govt 🤝 AI safety

US President Joe Biden recently met with the CEOs of big tech - including Microsoft, Google and OpenAI - to talk AI safety.

Google & OpenAI CEOs walking into the White House

The two-hour meeting focused mostly on the short-term risks of AI. Mainly privacy violations, employment discrimination, and the potential for scams and misinformation campaigns.

Sam Altman claimed that all companies at the table were on the same page on regulations.

we must support responsible innovation that serves the public good... this means companies have a fundamental responsibility to make sure their products are safe before they are deployed.

White House Representative

Shortly before the meet, the White House also announced a $140 million investment to build “trustworthy” AI. This will enable the National Science Foundation to found seven new institutes to research AI safety.

The Office of Management and Budget is going release policy guidance on the use of AI by the US federal government soon.

What this means: the US understands it’s going to be the cradle of AI experiments, and the government is now beginning to take its potential for mass danger seriously.

BUZZWORD OF THE DAY

AnthropicAI

Eleven OpenAI employees left two years ago to found Anthropic, a competitor building what they call steerable, interpretable and reliable A.I. systems with humans at the center of it. Now valued at $4B+.

FROM OUR PARTNERS

Build & Deploy AI Models Quickly

Cerebrium makes deploying ML Models seamless. Deploy & Fine-Tune your model without worrying about infrastructure.

1/ Serverless GPU Deployment - <1s cold start

2/ All Major ML Frameworks Supported (Pytorch, Onnx, XGBoost, Etc).

3/ Don’t have your own model? Access 15+ pre-build models.

4/ Fine-tune FlantT5, GPT-Neo, Stable Diffusion, and more.

MAIN COURSE

The new Midjourney 🖌️ 

Midjourney just dropped v5.1. Here’s what’s new:

Midjourney v5 & v5.1 - underwater portrait of a young lady

/1 improved coherence, higher accuracy to text prompts, reduced unwanted borders or text artifacts, and enhanced image sharpness.

/2 it’s more opinionated and will give you images similar to v5 but with smaller prompts and with much higher quality.

/3 its got a raw mode for when you want more conservative results (as opposed to more creative ones).

#notarealpicture

4/ The lighting now looks more even and natural + most images with humans have a polished, editorial look that actually resembles an actual photographer.

5/ The new version uses many more colors to create captivating visuals and dynamic compositions, resulting in more vibrant and eye-catching images.

v5.1’s greatest flaw is… legs

Midjourney is going to gather and analyze data from users of v5.1 to improve their modeling systems even more. While there’s no API or chat interface just yet, they plan on making V5.1 and AI mods default modes in their Discord channels by the end of the week.

To play with v5.1 - go to the Midjourney discord, enter /settings, and select ‘MJ version 5.1’. For raw mode, click ‘RAW Mode’ after selecting V5.1.

Levelsio, builder making $82k selling Stable Diffusion products

Where this stands: miles ahead of every other AI image generator, and we wouldn’t say its close. Stable Diffusion and DALL-E have a whole lot of catching up to do here.

MEMES FOR DESSERT

YOUR DAILY MUNCH

Think Pieces

Breakthrough potential of AI: Sam Altman | MIT 2023 (21m).

Why the best way to avoid a down round is to build an AI startup.

We need to regulate AI. Here’s How.

Startup News

Acapela lets anyone back up their own voice for free in minutes.

Antimetal is putting AI to work to root out cloud cost inefficiencies.

Research

FlexiBERT: are current AI transformer architectures too homogeneous and rigid?

Tools

Parsio: automate the data extraction from emails and files.

WGPT: the Wordpress plugin to monetize chatbots.

Breeze: create professional product pictures.

PageGPT: create landing pages with AI.

TWEET OF THE DAY

Tag us on Twitter @BotEatBrain for a chance to be featured here tomorrow.

AI ART-SHOW

Until next time 🤖😋🧠

What'd you think of today's newsletter?

Login or Subscribe to participate in polls.