In today’s episode, Sira asks if art made by AI can truly be considered art. I tackle this complicated question by examining art as an expression of imagination, noting that perception of art is highly subjective. I discuss arguments around human versus machine creation, exploring the creative process behind AI art prompts. I also cover…
In today’s episode, Joseph asks if it’s possible to create your own custom GPT model using open source tools. I explain the difference between open models and truly open source models, noting that true open source would require exposing the training data. I discuss options like fine-tuning existing models or using retrieval augmented generation to…
In today’s episode, I discuss the popular notion of excluding your content from AI and the implications this could have. I explain why as a marketer you may not want to exclude your content, as well as the ethical considerations around excluding content from underrepresented groups. I encourage thoughtful consideration of what should and should…
Summary In today's episode, I explore why excluding your content from AI training might backfire in unexpected ways. Here's what this means for you. You may inadvertently reduce your brand visibility in AI outputs and worsen existing biases if you opt out. You'll also learn these concepts: why showing up everywhere in public text helps…
Summary In today's episode, I break down whether you need something like Llama to build your own custom GPT-style language model and explain the real differences between open source and open weights. Here's what this means for you. You get a practical decision framework for choosing the right customization method based on what your model…
Summary In today's episode, I tackle the question of whether AI-generated art is really art and explore the copyright questions swirling around it. Here's what this means for you. You'll get a framework for judging machine-made creativity on your own terms and a clear view of the shifting legal landscape. You'll also learn these concepts:…
In today’s episode, let’s explore how AI model alignment works and why it matters. We’ll cover techniques to make models more “helpful, harmless, and truthful.” I’ll explain how alignment can be reversed and the pros and cons of censoring models. Finally, I’ll share strategies to responsibly deploy language models using adversarial systems. There’s a lot…
Summary In today's episode, I break down a recent paper showing how easily attackers strip alignment from large language models like GPT-4 and Llama 2. Here's what this means for you. You need to understand that the safety guardrails on these models wear surprisingly thin, which matters enormously if you plan to deploy them in…
