- Bot Eat Brain
- Posts
- MIT & Harvard's Automated Social Science designs and runs experiments
MIT & Harvard's Automated Social Science designs and runs experiments
PLUS: Robot with a beefier body and brains
Good morning, human brains, and welcome back to your daily munch of AI news.
Here’s what’s on the menu today:
Step right up; it’s the AI social science clown show 🤡 🎪
MIT and Harvard’s AI system designs and runs social experiments.
AI-developed gene editor that can edit the human genome 🧬👨🔬
Profluent unveiled OpenCRISPR-1. It’s open-source.
The lowdown on the new Phoenix robot 🤖 🔥
Sanctuary AI unveiled the 7th generation of its humanoid robot.
MAIN COURSE
On Wednesday, MIT and Harvard published a paper called Automated Social Science. It’s a system that uses AI models to design and run social experiments.
What’s the point?
Automated Social Science is a system that uses AI models to help turn social science theories into real-world experiments. It can independently propose and test ideas.
What experiments were conducted?
The system was put to the test in various settings, including bail hearings, job interviews, and auctions.
Was it effective?
When directly asked to predict experimental outcomes, it often overestimated results and lacked reliability. However, when it was first given specific experimental data to work with, its ability to accurately predict results based on different variables improved significantly.
Interesting. How does it work?
The system uses structural causal models, which allow it to understand cause-and-effect relationships. These models help the system create agents and scenarios that change based on the factors being tested.
So, what?
This tech could potentially accelerate the efficiency, accuracy, innovation, and cost-effectiveness of research in areas like human psychology, economics, sociology, and more.
SPONSORED BY AE STUDIO
Hire a world class AI team
Engineers who understand AI are expensive and difficult to find, and it can be hard to figure out who to trust. On top of that, 85% of all AI projects fail.
But AE Studio succeeds.
We listen to your business challenge and help you craft and implement the optimal AI solution with our team of world class AI experts from Harvard, Stanford and Princeton.
Our development, design, and data science teams work closely with founders and executives to create custom software and AI solutions that get the job done. The secret to our success is treating your project as if it were our own startup.
SIDE SALAD
Designer babies, here we come 🧬👨🔬
On Monday, Profluent revealed OpenCRISPR-1. It’s the world’s first open-source, AI-generated gene editor.
How do you find the difference between male and female chromosomes? 🤔
Take down their genes... 🧬🤭
What is OpenCRISPR-1?
It’s an open-source, AI-generated gene editor designed for precision editing of the human genome.
How does it work?
OpenCRISPR-1 is designed to produce non-naturally occurring proteins. It leverages AI to analyze biological data and create gene editors. These editors were developed using Nobel Prize-recognized techniques based on CRISPR, a technology that originates from natural bacterial mechanisms.
So, what?
This is the first successful instance of precise human genome editing using customizable gene editors entirely developed through AI innovation. It has the potential to transform the way we fight against diseases and illnesses.
A LITTLE SOMETHING EXTRA
Bot sees Bot 🤖 🔥
On Thursday, Sanctuary AI introduced the 7th generation of Phoenix. It’s a general-purpose robot with hardware and AI software enhancements.
What’s up with Phoenix?
The robot now has improved uptime, faster build/commissioning speeds, and a lower bill of materials.
Beefier body?
This version of Phoenix has a greater range of motion in the wrists, hands, and elbows, along with better durability in the hands. Its hydraulics have been miniaturized, which results in lighter weight, reduced power consumption, and decreased complexity while still exceeding safety standards.
Beefier brains?
The robot’s visual and tactile sensing have been enhanced, which provides higher-fidelity data for AI training. Its new AI control system, Carbon, significantly reduces the time needed to automate new tasks, cutting down the automation process from weeks to less than 24 hours.
RECOMMENDED READING
The Aurorean is a clear, essential newsletter focused on demystifying science and AI. Every edition is designed to distill complex topics into understandable insights in one weekly briefing.
Discover, understand, and master the wonders of STEM with The Aurorean today. It’s free!
YOUR DAILY MUNCH
Tools 🔨
Dart — a project management tool that automates reports, tasks, and more.
Bland AI — create and scale phone agents, handle high call volumes, and more.
Friend — Open-source device that records, transcribes, and lasts over 24 hours.
Think Pieces 🧠
Humane vs. rabbit. The R1 is $199, and the AI Pin is $699 plus $24 a month.
Is Meta making money from AI? Mark Zuckerberg says his AI investments won’t be profitable for years.
Startup News 💰
Google published a report on generative AI's economic effects. How AI enhances growth, productivity, and more.
Snowflake unveiled Artic LLM. Snowflake claims it’s better at coding/SQL generation than Meta’s Llama 2 70B.
Salesforce made Einstein Copilot generally available. It provides contextual, trustworthy answers from enterprise data directly within the corporate cloud.
Research 👨🔬
NeRF-XL — efficiently distributes Neural Radiance Fields across multiple GPUs, enhancing the capture and reconstruction of large-scale, high-detail scenes.
ConsistentID — a method that enhances identity consistency in personalized portrait generation using detailed multimodal ID information.
MultiBooth — enables efficient generation of customized multi-concept images by combining separately learned single-concept modules without additional training.
MEMES FOR DESSERT
TWEET OF THE DAY
Awesome use case for Open Interpreter’s 01 device. It’s making hands-free adjustments in his DAW.
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? |