- Superhuman AI
- Posts
- Generating new materials with AI
Read time: under 4 minutes
Welcome back, Superhuman
You don’t know what you’ve got ‘til it’s gone. Meta learned that lesson the hard way after a biotech team it axed last year showed off a new AI platform that can create new proteins from scratch.
Today’s Insights
Level up your AI skills with The AI Academy
Generating new materials with AI
Tutorial: How to generate a diagram in Claude 3.5
AI’s impact on the VC industry and the economy
5 new AI tools to boost your productivity
AI-Generated Images: Calla Lily Colors
THE AI ACADEMY
Learn how to use AI at work and boost your productivity
AI is supposed to make your life easier — but the overwhelming number of ChatGPT and AI courses out there make it impossible to know where to start.
After getting thousands of requests for legit AI courses, we’re finally launching the Superhuman AI Academy to help you cut through the hype and learn the AI skills that actually matter.
Here’s what you’ll get:
The Intro to AI & ChatGPT Mastery course
Access to 100+ courses and tutorials
Our top 100 AI tools and 1,000+ prompts
A network of professionals that are using AI around the world
And much more
Our courses and tutorials have been developed by a team that consists of AI engineers, entrepreneurs who have built 8-figure AI startups, and some of the leading AI content creators in the world.
The AI Academy launches in 7 days. We’re giving the first 100 people who join our waitlist below 50% off your first month.
NEXT IN AI
Former Meta developers chart their own path with a model that creates new proteins
Source: EvolutionaryScale
While some AI companies focus on the big picture — like how many tokens their models can handle at once — one new startup is zooming in…way in. Instead of generating new words or images, EvolutionaryScale, based in New York and San Francisco, is working on an AI model that creates blueprints for proteins that aren’t found in nature.
Deja Vu: If the idea sounds familiar, it might be because founder Alexander Rives and much of his team were working on something similar at Meta – that is, until their department was disbanded last year. They didn’t miss a beat and used the knowledge they’d gained to start something new. Their startup just raised $142 million from backers like former GitHub CEO Nat Friedman as well as corporate clients, including Amazon and NVentures, a wing of Nvidia.
How does it work? EvolutionaryScale’s new model, ESM3, was trained on almost 2.8 billion proteins — allegedly the largest-ever data set for a biology-focused AI system. With that knowledge, it can spin up brand-new proteins and predict how humans might react to them.
Why proteins?
Proteins are like the worker bees of the body
They play such a crucial role that biologists often refer to them as the building blocks of life
They can repair our injuries, facilitate chemical reactions, and regulate our blood flow
So if we can engineer new ones, we can put them to work to help with specific tasks, whether fighting off cancer or preventing the onset of Alzheimer’s
Mind Blown: As a proof-of-concept, EvolutionaryScale already created a new green fluorescent protein that’s similar to the substances sea creatures use to glow in the depths of the ocean. As Fortune points out, it would have taken millions of years for this material to have developed evolutionarily.
It’s only the beginning: Several other startups are working on similar AI-powered biotechnologies. Eli Lilly just announced its plans to work with OpenAI on new medicines that could be used to treat drug-resistant bacteria. And you might recall that DeepMind recently released AlphaFold3, which can predict the shape of proteins, DNA, and RNA. Meanwhile, French startup Alcove is working on models that generate recipes for how to concoct inorganic materials like rare earth elements.
PRESENTED BY SPINACH
Hire Your First AI Team Member - The AI Project Manager
Spinach AI always joins meetings on time, takes super accurate meeting notes, captures action items, and creates tickets and PRDs based on your conversations.
Backed by Y Combinator, Zoom and Atlassian
Used by thousands of companies ranging from small startups to Fortune 500
Supports 100 languages
Sounds too good to be true? Start your 14-day trial here
AI AT WORK
How to visualize anything with diagrams using Claude 3.5 Sonnet
Go to Claude’s website and log in.
Make sure you select Claude 3.5 Sonnet as your model.
Write in your prompt in the prompt box:
Create a simple visualization showing me supply chain management process
Refine your prompt if needed. Wait for a few seconds and you are done.
PROMPT OF THE DAY
Summer Patio
Prompt: Create a patio layout that seamlessly integrates indoor and outdoor living spaces for a cohesive feel.
Recommend outdoor furniture arrangements that maximize seating capacity without overcrowding the space.
Suggest creative lighting solutions to enhance the ambiance and usability of your patio during evening hours.
You can adapt the prompt to your specific needs.
Source: Archgyan
PRESENTED BY BRILLIANT
Falling Behind? Learn AI In Minutes A Day
If you want to get ahead quickly, you should try Brilliant – the app that helps you master AI basics in just minutes a day. LLMs, quantum computers, data science… Lessons are bite-sized, interactive, and designed for all levels (even yours)!
Join 10M lifelong learners and save 20% on an annual subscription.
AI & ECONOMICS
AI jolts VC funding back to life, but impact remains to be seen
Chart Source: Bloomberg
VC funding is back — thanks in large part to AI. Nearly half of the $55.6 billion in VC funding raised last quarter went to AI-related firms, according to PitchBook. Excitement around AI boosted overall funding to a two-year high, putting an end to several quarters of declines. For investors, the chance to ride the AI wave was enough to ease worries over high interest rates and other economic concerns.
On the other hand, there were fewer overall transactions last quarter than at any point since 2020, and a big portion of the funding went to only a few top players. Elon Musk’s xAI alone walked away with $6 billion – more than a fifth of the total given out to AI companies. That means it was likely just as hard, if not harder, for smaller startups to break through.
Why it’s important: When there’s little funding to go around, startups have to make big cuts to keep afloat. An influx of cash, on the other hand, means they can keep growing and can take more risks. Startup funding also helps set the tone for the overall economy, and AI is now an integral part of that equation. Some experts think the industry could have a bigger impact on our financial system than even the internet.
Regulators are taking notice: During one recent panel, Fed Chair Jerome Powell said his agency is investigating how generative AI will impact inflation, productivity, and employment trends. “There’s a sense of something big coming here,” Powell said. But ultimately, Powell thinks “it’s too early to say” how exactly AI will shape the economy. An Economist analysis suggests he’s onto something: Even though AI is already playing a bigger part in our everyday lives, it could take nearly a decade for big tech companies to actually generate earnings from those investments.
PRODUCTIVITY
5 AI Tools to Supercharge Your Productivity
✅ BrainyAI: Chat with multiple models — including GPT-4o, Claude, and Llama — and compare AI responses in one chat.
✅ CreAItion: Effortlessly create stunning objects in a visual workflow powered by advanced AI.
✅ Braincast: Turn links and PDFs into Duolingo-like quizzes in seconds for better, faster learning.
✅ Resistbot: A chatbot that turns your texts into faxes, postal mail, or emails to your representatives in minutes.
✅ Wanderboat: Chat with an AI travel assistant to generate a personalized itinerary complete with attractions, cuisine, and events.
PS: Want more? Check out our Top 100 AI Tools.
* indicates a promoted tool, if any
AI & TECH NEWS
Everything else you need to know today
Source: WhatsApp
Me, Myself, and AI: WhatsApp is developing a feature that will let users create their own AI-generated avatars, complete with customizable backgrounds.
Unintended Audience: OpenAI updated its new ChatGPT app for Macs after a developer noticed that the program was storing unencrypted conversations in plain text — a potential security concern.
Sneak Peak: Tesla showed off of its second-generation Optimus robot at the World AI Conference in Shanghai last week. The Elon Musk-led firm was one of the few US-based companies to appear at the event.
😄 One Fun Thing: ElevenLabs is releasing a new set of voices for its Reader app that are based on real-life Hollywood icons, including Liza Minnelli, James Dean, and Burt Reynolds. You can even have Judy Garland read “The Wizard of Oz” aloud with the new AI-powered feature.
🧠 Brain Food: Dutch researchers used AI to accurately recreate images from macaque monkeys’ thoughts. Similar studies have been conducted in the past, but according to one researcher, “these are the closest, most accurate reconstructions” yet. They could pave the way for brain implants that can help restore vision for people with visual impairments.
AI-GENERATED IMAGES
Callalily Colors
Source: @chriswei. on Midjourney
Prompt: portrait of Chinese Callalily flower, in the style of dark purple, innovative page design, quiet moments captured in paint, all-over composition, elegant calligraphy, soft and dreamy atmosphere, white background
--style raw --ar 3:4 --stylize 150Acquire new customers and drive revenue by partnering with us
Superhuman is the world’s biggest AI newsletter for businesses and professionals with 600,000+ readers working at the world’s leading startups and enterprises. Companies like Amazon, Hubspot, and Salesforce feature their products in Superhuman. You can learn more about partnering with us here.