By Kim Berard

I can’t believe I missed OpenAI Introducing DALL-E January 2021 – a neural network that can “generate images from textual descriptions” – so I’m sure I won’t miss DALL-E 2 from OpenAI.because of them describe it“DALL-E 2 is a new AI system that creates realistic images and art from descriptions in natural language.” The name, by the way, is a playful mix of animated robots Wali and special artist Salvador Dali.

This is not your father’s artificial intelligence. If you thought it was just about art, think again. If you don’t think it matters for healthcare, you’ve been warned.

Here’s a further description of what OpenAI claims:

“DALL·E 2 can create original, photorealistic images and art from textual descriptions. It can combine concepts, attributes and styles.

DALL·E 2 enables realistic editing of existing images based on natural language captions. It can add and remove elements while taking into account shadows, reflections and textures.

DALL·E 2 can take one image and create different variations based on the inspiration of the original. “

Here is their video:

I’ll leave it to someone else to explain how it does it all, other than saying that it uses a process called diffusion, “It starts with a pattern of random dots and when it identifies a specific aspect of the image , gradually changing the pattern into a picture of the image.” The end result is that the DALL-E 2 “produces a more realistic and accurate image with a 4x increase in resolution” relative to the DALL-E.

Devon Coldway, write on TechCrunchmarvels:

It’s hard to overstate the quality of these images compared to other generators I’ve seen. While there’s almost always the kind of “tells” you’d expect from an AI-generated image, they’re not as obvious, and the rest of the image is much better than the best that anyone else has generated.

Well, it’s true that DALL-E isn’t coming up with artistic ideas of its own, but it’s creating images like never before, like a koala bear dunking or the Mona Lisa Dymohawk. If it’s not the creativity of artificial intelligence, it’s close.

—————

Sam Altman, CEO of OpenAI blog post There are a few interesting thoughts on the DALL-E 2. He started by saying: “For me, this is by far the most enjoyable game we’ve ever created. I’ve found it enhances creativity, helps in many different situations, and in a way I haven’t seen in a while. The joy felt in technology.” I’m a big believer in the Seven Johnson dictum that the future is where people are happiest, so that’s really important to me.

Mr. Altman outlines six things he thinks are worth noting about DALL-E 2:

“1. This is another example of what I think will be a new trend in computer interfaces: you say what you want in natural language or contextual cues, and the computer does it.

2. It does seem to “understand” many levels of concepts and how they relate to each other in complex ways.

3. While I am a firm believer that AI will create a lot of new jobs and make many existing jobs better by doing the boring parts, I think it will make some jobs less relevant more and more to be honest (like technology frequently), it’s important to do)

4. A decade ago, the conventional wisdom was that AI would affect physical labor first, then cognitive labor, and maybe one day it could do creative work. Now it looks like it will do it in reverse order.

5. This is an example of a world where good ideas are the limits of what we can do, not specific skills.

6. While the advantages are great, the model is powerful enough that it is easy to imagine the disadvantages. “

On this last point, OpenAI restricts the images DALL-E has trained on, watermarks every image it generates, censors all generated images, and restricts the use of real faces. They recognize the potential for abuse. Oren Etzioni, Chief Executive Officer Allen Institute for Artificial Intelligence, warn New York Times: “There is already disinformation online, but it is worrying that this will expand it to new levels.”

Mr. Altman said there could be a product launch this summer with wider access, but Mira Murati, director of research at OpenAI, is firm: “This is not a product. The idea is to understand the capabilities and limitations and give us the opportunity to mitigate.”

Prafulla Dhariwal, OpenAI Algorithm Researcher Tell fast company: “Both vision and language are key parts of human intelligence; building models like DALL-E 2 bridge these two fields. This is a very important step for us as we try to teach machines to perceive the way humans do world, and then eventually general intelligence.”

As their video says. “DALL-E helps humans understand how advanced artificial intelligence systems see and understand our world.”

————

I don’t have any artistic skills, but as Mr. Altman suggests, we are building “a world where a good idea is the limit of what we can do, not a specific skill.” In that world, as Altman suggests As Mr Mann also suggested, artificial intelligence could potentially do creative and cognitive work before manual work.we have already Already met Adaan AI-driven “robot artist,” and we’ll see other examples of creative AI.

OpenAI already has OpenAI Codex, an “artificial intelligence system that can convert natural language into code”.There are AI tools that can be written, including One powered by OpenAIand those can compose.

Of course, Google have A range of AI-specific programs for health.

Healthcare in general, and the practice of medicine in particular, has long been viewed as a uniquely human endeavor. Its practitioners claim that it is a combination of art and science that cannot be easily reduced to computer code. If healthcare finally admits that AI is good at, say, Identify radiology imageswhich claims to be a long way from diagnosing patients with complex conditions, let alone advising or comforting them.

Maybe we should have DALL-E 2 draw them a picture of what it might look like.

Kim was the former head of e-marketing for the Blues major program, the late and lamented editor tincture.ionow a regular THCB contributor.