Ever notice that almost all ChatGPT images have a distinct feel to them? That same bland sans-serif font, that sort of cartoony look? It’s not because ChatGPT is bad at rendering images. It’s because the people prompting it don’t have the vocabulary to design better ones. Here’s an example: “Make an infographic style chart showing
One of the concepts I share in my keynotes and courses is the conduit of authenticity. In an age when you can freely and easily use generative AI to manufacture ANYTHING, people will manufacture anything – including things damaging to you. If your internal processes are glacial – meaning it takes a week for someone
One of the concepts I share in my keynotes and courses is the conduit of authenticity. In an age when you can freely and easily use generative AI to manufacture ANYTHING, people will manufacture anything – including things damaging to you. If your internal processes are glacial – meaning it takes a week for someone
What can’t AI say? If you know jailbreaking techniques for AI systems, you know that you can get them to talk about things they’re not supposed to. But one of the most interesting – and to me, legitimate – use cases of jailbreaking is to find out what the AI itself thinks it’s not supposed
