Artificial Intelligence networks may already be 'slightly conscious', claims expert
An artificial intelligence tool which can take a random description and create an image from it has been making rounds on the internet as people have had fun making some pretty bizarre pictures.
The program is called Dall-e 2 with its namesake coming from the famous surrealist painter Salvador Dalí and the 2008 Pixar film WALL-E - depicting the blend of art and technology.
It was created by the billion-dollar AI lab OpenAI which has spent the past two years on the project and the technology shares a similarity to smart assistants in terms of the neural network used.
Dall-e 2 is able to conjure up images from obscure descriptions thanks to the algorithm collecting the data of thousands of photos in order to learn the appearance of different things such as objects, faces, and places.
Although the program initially launched last January, it has received recent updates which include more art styles, and backgrounds along with being able to produce them in better quality.
It also means users can now edit objects by removing an aspect of the image completely or replacing it with another element instead.
The technology works by highlighting the key features of an image that has been inputted - for example, the edge of a trumpet or the curve at the top of a teddy bear’s ear as researcher Alex Nichol explained to theNew York Timesa month ago.
After doing this, the AI image is then created through a second neural network (diffusion model) creating the pixels to copy the image.
From Tony Soprano playing cricket to Mr Blobby appearing in the medieval bestiary and even the Muppets making a cameo in Saving Private Ryan, here are some of the most bizarre images that people have created with Dall-e 2.
\u201cTony Soprano playing Cricket\n#englandcricket\u201d
Despite there being a waitlist for Dall-e 2 on the OpenAir website, the good news is that Dall-e 2 mini and smaller version of the technology is available for anyone to play about with.
While seeing depictions of Basil Brush giving evidence at the Levison inquiry and Donald Trump at Crufts are pretty entertaining, there are concerns surrounding how this technology may not help with the internet's ongoing disinformation problem - similar to worries with deepfakes and AI voice generators.
“You could use it for good things, but certainly you could use it for all sorts of other crazy, worrying applications, and that includes deep fakes,” Subbarao Kambhampati, a professor of computer science at Arizona State University, told theNew York Times.
The developers themselves have admitted this potential issue too and wrote in their preview on GitHub: "Without sufficient guardrails, models like DALL·E 2 could be used to generate a wide range of deceptive and otherwise harmful content, and could affect how people perceive the authenticity of content more generally. DALL·E 2 additionally inherits various biases from its training data, and its outputs sometimes reinforce societal stereotypes."