- Bot Eat Brain
- Posts
- OpenAI's text-to-3D
Good morning human brains, welcome back to your daily munch of AI news.
Here’s what’s on the menu today:
Wendy’s AI drive-thru 🚗
Wendy’s is partnering with Google to test a new AI system that’ll take your orders at its drive-thru.
OpenAI’s new model 🪟
OpenAI just quietly released a paper detailing its own new take on text-to-3D models.
Warren Buffet on AI 💸
Warren Buffet and Charlie Munger are scared of AI, likening it to the atomic bomb.
APPETIZER
Wendy’s AI drive-thru 🚗
Wendy's is partnering up with Google to create an AI chatbot that will take your orders at its drive-thrus.
Wendy’s AI prototype
Wendy's hopes an AI assistant will speed up lines and reduce costs. Wendy's says they want to make workers more efficient, not replace them.
Part of Google's work with Wendy's is ensuring the AI understands Wendy's lingo. This means the bot will have to know that a milkshake is really a Frosty and that a JBC is short for a Junior Bacon Cheeseburger.
According to Google, the bot might struggle to ignore noises from kids in the backseat or understand when change your mind in the middle of an order.
Our Take: As soon as it’s possible to replace a job with a kiosk, that job is as good as gone. Self-checkout didn’t just make workers more efficient, it replaced them. AI kiosks will do the same.
Bad for workers. Good for Wendy’s shareholders and cheaper burgers.
BUZZWORD OF THE DAY
Multimodal
AI that has the ability to process and understand inputs from different modes or modalities, including text, speech, images, and videos. GPT-4 is multimodal because it can process both text and images (a feature which has not yet been released).
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
OpenAI’s Text-To-3D 🥉
OpenAI just released ChatGPT for 3D modelers with Shap-E, a new text-to-3D model that can turn your text into object renders.
At this point, we’ve seen several previous examples of text-to-3d models. So, what’s the big deal? OpenAI joining the fray with its now-massive resources means these models might be about to get a lot better, a lot faster.
Shap-E examples
Here are the highlights:
1/ Shap-E can generate fine-grained textures and complex shapes from text inputs.
2/ It uses neural radiance fields aka NeRFs, the same tech powering AI drone footage and AI interior design.
3/ The process of generating 3D objects from text prompts is significantly faster with Shap-E compared to OpenAI's previous experiment, Point E.
4/ The 3D outputs are low-res and miss key details.
Shap-E is currently available for free on GitHub and will run locally on your PC, though setting up the tool and getting it to run is a mammoth task.
What this means: this isn’t a flashy reveal of new breakthrough tech, but it is OpenAI putting down both feet in the 3D modeling industry. The tech is early, but may soon impact movies, VR, engineering simulations, and other large industries.
A LITTLE SOMETHING EXTRA
Legacy Investors aren’t AI fans 👎️
Wall Street’s most respected billionaire, Warren Buffet, is likening AI to the atom bomb.
At the annual meeting of Berkshire Hathaway, Buffet expressed his concerns about generative AI. More specifically, "when we get something that can do all kind of things, I get a little worried... because I know we will not be able to un-invent it."
Buffet also revealed that he got a personal lesson on GPT from Bill Gates, but he’s still more apprehensive about it than he is impressed.
“I am personally skeptical of some of the hype that is going into artificial intelligence. I think old-fashioned intelligence works pretty well."
Our take: We aren’t sure of the kind of robots Buffet has been hanging out with, but given his historical apprehension of all things tech, we’ll take his financial advice and leave the doomsday predictions for now.
Doomsday predictions from AI researchers on the other hand? Those we’ll entertain.
Buffet’s apprehensions of AI are unlikely to move markets, given that he rarely invests in tech.
MEMES FOR DESSERT
YOUR DAILY MUNCH
Think Pieces
Google and OpenAI are Walmarts besieged by fruit stands.
Tech workers could take labor lessons from Hollywood’s writers.
Startup News
ChatGPT is shaking up the edtech markets.
With DeepFloyd, generative AI art gets a text upgrade.
Research
Fairness in forecasting Linear Dynamical Systems: making sure predictions made by mathematical models are fair and unbiased.
Tools
EzMail: draft your email with AI.
editGPT: proofread, edit, and track changes to your content.
PromptBox: save all your AI prompts in one place.
CometCore: automate tasks & creating multimedia content.
TWEET OF THE DAY
Demoing YOLO-NAS: “A Revolutionary Model for Object Detection”:
AI ART-SHOW
Until next time 🤖😋🧠
What'd you think of today's newsletter? |