This Week’s Insights 📰
Claude.ai 3.5 Sonnet Ahead Again in the LLM Races
Anthropic is making waves with its latest AI model, Claude 3.5 Sonnet, promising to outshine OpenAI’s ChatGPT and Google’s Gemini. Claude 3.5 claims faster speeds and lower costs, acing key benchmarks in reasoning and coding. It’s packed with new features like “Artifacts,” a dynamic workspace for real-time collaboration, and enhanced visual data processing. Available for free on claude.ai and the Claude iOS app, this update turns Claude into a collaborative powerhouse. Users are saying the results are INSANE for SEO. Give it a try and let us know what you think!
Would You Like Fries With Your 260 McNuggets?
McDonald’s is pulling the plug on its AI-powered drive-through tests after nearly three years of hilariously frustrating ordering mishaps. Customers took to social media to share their exasperation as the AI misheard orders, like racking up 260 McNuggets or serving nine iced teas instead of one. Despite McDonald’s struggles, competitors like White Castle and Wendy’s report high accuracy with their AI systems. Looks like McD’s AI has some learning to do!
Lights, Camera, and Automation
Looks like Hollywood might be warming up to AI! A new UTA study reveals that 73% of entertainment and marketing creatives believe AI will elevate content, with 75% saying they’re creating higher-quality work thanks to it. While 71% support increased regulation, there’s a growing curiosity and cautious optimism about AI’s potential to enhance creativity. As Joe Kessler of UTA IQ puts it, AI can help creators explore what’s possible and make the most of our precious time.
AI Video Tool Takes Pixar Characters on Unauthorized Vacation
An AI video tool is causing a stir by seemingly borrowing from Disney’s IP. Luma’s new Dream Machine, hailed as the future of filmmaking, generated a trailer featuring an AI-blurred but recognizable Mike Wazowski from Pixar’s Monsters, Inc. Luma’s CEO blamed a user-uploaded image, but questions about the tool’s transparency and training data linger. As Dream Machine joins other text-to-video AI tools, the debate over AI-generated content and copyright infringement heats up. Disney has yet to comment, but the incident highlights ongoing concerns about the AI ecosystem’s propensity for plagiarism.
AI’s Secret Sauce? A Dash of Human Creativity
At ADWEEK House at Cannes, experts from Ulta Beauty, Autodesk, and Vidmob emphasized the crucial role of human input and quality data in AI-driven marketing. Vidmob’s CEO Alex Collmer highlighted the importance of owning and controlling creative data to leverage AI effectively. Autodesk’s CMO Dara Treseder stressed the emerging skill of prompt engineering—knowing how to ask AI the right questions for optimal results. Authenticity in ads remains essential, with experts underscoring the need for human intervention to truly resonate with consumers. In a nutshell, AI in marketing is powerful, but human creativity and data control are still vital for success.
AI Lab Visual Vault 🖼️
Use Case: Using Negative Prompts in Midjourney
Negative prompting in Midjourney allows you to exclude certain elements from your generated images. There are two ways to achieve this: using the –no parameter and using negative weights. Let’s break down both methods with some examples.
1. The –no Parameter
This method is simple and effective for most users. To exclude specific objects from your image, you append –no followed by the object you want to remove.
Example 1: Excluding Birds Let’s say you want to generate a grid of beach images without any birds:
Prompt: beach, sunrise –no birds
Result: You will get four images of a beach at sunrise without any birds in them.
Example 2: Excluding Multiple Objects You can exclude multiple objects by listing them after the –no parameter:
Prompt: beach, sunrise –no birds, people, boats
Result: You will get four images of a beach at sunrise without birds, people, or boats.
2. Negative Prompt Weight
This method provides more nuanced control over the presence of certain objects in your image by adjusting their weights.
Quick Intro to Prompt Weights: You can emphasize or de-emphasize parts of your prompt using the :: syntax. For example, in a prompt with “mountain” and “lake”:
Equal Weight: mountain:: lake::
Increased Lake Weight: mountain:: lake::2
Negative Lake Weight: mountain:: lake::-.2
Result: With a negative weight, the lake is excluded from the image, focusing solely on the mountain.
Using the “–no” parameter (method 1) works the same as applying a negative weight of -0.5 (method 2).
Tools We Used This Week 🧰
HeyGen – Create, repurpose, and revamp studio quality videos at scale with AI-generated avatars and voices. Create your own avatar from your own video and voice!
ChatGPT’s Mobile App – Different features available via your mobile device! From image selection tools to audio conversations, interacting with ChatGPT on your phone is an entirely different experience!
Read.ai – Your AI notetaker and speech coach! Get meeting sentiment analytics, speech tips, and reference transcripts all in one area with Read.Ai
Join the Conversation 🔊
This isn’t just about sending newsletters into the void—we’re building a lively hub here. If you’ve got insights itching for daylight, questions that stump the wise, or ideas that buzz louder than a caffeinated bee, send them flying to info@aismartmarketing.com. Your contributions are the fuel for our content engine, driving us to enlighten and inspire.
See you in the Lab!
Share with a friend or Subscribe!