If you’ve noticed a change in the images that occupy the small space that accompanies each editorial above this column since yesterday, your eyes have it. On first glance
, it may seem a simple switch from thematic illustrations into thematic photos or artwork. But, on closer look, you’d have found yesterday’s tangential images on interest rate
hike (a rupee
cushion) and bankruptcy
code (two scared bankers) pleasantly odd, creatively representative. No, we didn’t get a cushion stitched or hijacked two bankers to get photos taken. We used DALL·E
2, an OpenAI-developed ‘deep learning model’ that generates images from what is prompted to them by written instructions. We key in what we want an image to be, and, in a few seconds, DALL·E 2 pops them out like an extremely talented pop-up artist.
Just to show what DALL·E 2 can do, we prompted ‘Science fiction realistic artwork of elephant in a spacesuit ready for take-off’. The result (above) is something, we believe, Dali would have been proud of. The more detailed the prompt, the better DALL·E’s ‘visual execution’ output. Try it (openai. com/dall-e-2/). It’s like showing off your daydreams.