Generative AI 101

In this episode of Generative AI 101, we explore the numbers and benchmarks that make OpenAI's o1 model a standout. From crushing the International Mathematics Olympiad with an 83% success rate to out-coding 93% of humans on Codeforces, o1 isn’t just flexing—it’s proving itself. But it’s not just about math and coding; o1 also excels in reasoning-heavy tasks, earning human preference over GPT-4 fo...

In this episode of Generative AI 101, we pop the hood on OpenAI's o1 model and explore what we know about the inner workings. We’ll break down its advanced "chain of thought" reasoning, its unique training methods like Reinforcement Learning from Human Feedback (RLHF), and the safety measures keeping it from going rogue. From AI boot camp to solving complex problems like a zen master, o1 takes its...

In this episode of Generative AI 101, we spotlight OpenAI’s latest creation - the o1 Preview. Think of it as AI that doesn’t just spit out answers but actually takes the time to "think," like the Sherlock Holmes of machine learning. We’ll break down what makes this new model different, from its ability to simulate human-like reasoning to solving advanced problems in math and science. Whether you'r...

In this episode of Generative AI 101, we explore how companies are transforming their industries with generative AI image tools. From Wayfair’s AI-powered home decor tool to Adobe’s Firefly in Photoshop, businesses are using AI to boost productivity, enhance creativity, and even address global challenges like climate change and poverty. We'll explore examples from Bentley Systems, Microsoft, Stitc...

In this episode of Generative AI 101, we explore how AI image generators like DALL-E, MidJourney, and Stable Diffusion are trained to create stunning visuals from text. We explore the absolute mountains of data—millions of image-text pairs—that fuel these models, the importance of diffusion models in producing reliable, high-quality images, and how self-supervised learning teaches AI to recognize...

In this episode of Generative AI 101, we’re tackling the art of writing image prompts—those little chunks of text that guide AI models in creating the visuals you want. It’s not as simple as saying “a sunset” and getting a masterpiece; writing the perfect image prompt requires precision and creativity. We break down the key elements—like subject, action, environment, objects, color, style, and moo...

In this episode, we explore NightCafe, an AI art studio where anyone can turn their text prompts into visual masterpieces using models like Stable Diffusion, DALL-E 3, MidJourney, and so many more. Whether you're creating a "cyberpunk sunset over Tokyo" or a "cat in a tuxedo riding a unicycle in space," NightCafe makes it easy and fun, with features like Depth-to-Image, Selfie Generators, and even...

In this episode of Generative AI 101, we venture into the world of Stable Diffusion, a groundbreaking AI model that’s democratizing image creation. Released in 2022, this accessible tool allows anyone with a decent GPU to generate stunning visuals, from sci-fi cityscapes to touch-ups on old photos. We’ll break down how it works, exploring the “diffusion” process that adds noise to images before cl...

In this episode of Generative AI 101, we explore DALL·E, OpenAI’s innovative image generator that merges artistic imagination with advanced AI. Explore how DALL·E transforms text prompts into vivid images using a diffusion model, bypassing earlier GAN technology for greater precision and versatility. From Victorian cats in cafés to creative product mockups, DALL·E shines in its ability to visualiz...

In this episode of Generative AI 101, we explore MidJourney, an AI tool revolutionizing the world of digital art by transforming simple text prompts into stunning, high-quality visuals. Using a blend of Diffusion Models and GANs, MidJourney pushes the boundaries of creativity, allowing users to generate everything from futuristic cityscapes to surreal dream-like images that look straight out of an...

In this episode of Generative AI 101, we’re exploring the artistic world of generative AI image generation. Ever imagined describing a scene—like a cat lounging on a floating island—and watching AI bring it to life in seconds? That’s the magic of text-to-image generators. From GANs (Generative Adversarial Networks) to Diffusion Models, we’ll break down how these digital artists work, their quirks,...

In this bonus episode of Generative AI 101, we’re looking at a growing concern - sounding too much like ChatGPT. Ever been told you write like a machine? That’s what we’re unpacking today. We’ll explore how AI, particularly ChatGPT, has started to influence our writing with overused phrases and predictable language. It’s like seasoning your food with the same spice over and over—reliable but event...

In this episode of Generative AI 101, we discover the origins and workings of the Elo Rating System—a clever, adaptable method originally designed to rank chess players but now influencing everything from online gaming to, most importantly for us, AI models. We’ll explore how this simple yet effective system predicts match outcomes, adjusts rankings, and even finds parallels in areas like online d...

In this episode of Generative AI 101, we’re exploring the need-to-know LMSYS Org and their innovative Chatbot Arena. LMSYS Org, a collaboration between top minds at UC Berkeley, UC San Diego, and Carnegie Melon University, is breaking new ground by making powerful AI models accessible to everyone. They’re not just creating AI—they’re building a community where anyone can contribute and evaluate to...

In this episode of Generative AI 101, we explore evaluating Generative AI large language models (LLMs). Just like finding the best restaurant in town means more than judging a single dish, evaluating AI models requires a comprehensive approach. We break down why assessing performance, comparing models, and building user trust are central to evaluation. From authenticity to speed, and fairness to c...

In this episode of Generative AI 101, we’re tackling the curious case of AI hallucinations—when AI creates content that’s completely off the mark. We’ll explore how these digital daydreams happen, why humans aren’t immune to similar slip-ups, and the impact these hallucinations can have. Whether amusing or alarming, AI hallucinations are a phenomenon you’ll want to understand.   Check out this Pap...

In this episode of Generative AI 101, we’re fine-tuning your AI prompting skills. Learn how to refine prompts to turn mediocre responses into top-tier results. We’ll cover techniques like iterative prompting, using context, and applying constraints, along with ethical considerations to ensure fairness. Get ready to make your AI work smarter, not harder.   Connect with Us: If you enjoyed this episo...

In this episode of Generative AI 101, we’re exploring the various flavors of generative AI bias and inaccuracy—those pesky issues that make your AI sound like it’s stuck in a 1950s sitcom. From data-driven biases baked into AI’s training to sneaky algorithmic quirks and the all-too-human flaws we project onto our creations, we explore how to recognize and mitigate these pitfalls. Learn how strateg...

In this episode of Generative AI 101, we catch a wave into the BAB framework (we're totally calling it "BABE")—an effortlessly cool method for crafting AI prompts that turn "meh" into "tubular!" BAB, short for Before, After, and Bridge, guides you from dull, lifeless AI interactions to responses that shine like a pro surfer catching the perfect wave. Learn how to structure your prompts for clarity...

In this episode of Generative AI 101, we explore CARE Prompting—a sophisticated method that prioritizes Craft, Audience, Response, and Evaluation to fine-tune AI outputs. CARE Prompting is designed to create responses that go beyond accuracy, ensuring they resonate deeply and reflect a comprehensive understanding of your goals. If you enjoyed this episode or have questions, reach out to Emily Lair...

In this episode of Generative AI 101, we’re taking a high-speed tour of the RACE Prompting method—where Role, Audience, Context, and Example aren’t just pit stops, but the keys to better communication.  Ready to take your AI skills to the next lap? Connect with Emily Laird on LinkedIn. Stay tuned for more laps around the ever-evolving circuit of AI and prompt engineering.  Connect with Emily Laird...

In this episode of Generative AI 101, we explore the ins and outs of RTF Prompting—an approach that sharpens your AI's focus by defining its Role, Task, and Format. Whether you're tasking your AI with writing a travel blog, analyzing financial data, or extracting historical records, RTF Prompting provides the clarity and structure needed to achieve precise results. We’ll break down how to use this...

In this episode of Generative AI 101, we break down the concept of self-consistency prompting—a technique that enhances AI accuracy by posing the same question in multiple ways and selecting the most consistent answer. By guiding your large language model through varied prompts, this method improves performance on complex reasoning tasks and reduces errors. However, it's not without challenges: it...

In this episode of Generative AI 101, we break down the concept of chain-of-thought prompting—a technique that helps AI models think through complex tasks step by step, improving their logical and accurate responses. By guiding the AI, or your large language model of choice, through detailed instructions, this method enhances performance on tasks like arithmetic and decision-making. However, it’s...

In this episode of Generative AI 101, we explore the example-heavy world of few shot prompting. Imagine describing a dish you love to a chef and then offering them multiple recipes on how to make it. That's few-shot prompting—maximum input, maximum efficiency. We'll explore how this technique can turn your AI into a translation whiz or a sentiment analysis expert with just one example. We'll also...