I’m so glad you’re here.

• Apple Developing Custom Silicon for AI Servers

• Snowflake releases a flagship generative AI model

• SAP Growth Driven by AI Boom

There are a lot of new faces this week, welcome! Let’s dive in 🤿

Did a cool person forward you this email? Sign up for free here

🎧 AI on Screen

Exploring the Power of AI and Problem-Solving with Parth Patil

Takeaways 

• AI has the potential to solve a wide range of problems and provide cognitive superpowers to individuals

• Digitizing processes can lead to faster and more efficient operations

• Analytics and data platforms can provide valuable insights for businesses

• Thinking quickly and finding smarter ways to approach problems can lead to innovation and success

• Language models like ChatGPT can be used to learn programming and build chatbots

💰 AI Money

👀 Everyone is Gabbing About Perplexity

Here is a lil video overview highlighting their latest product offerings, updated valuation metrics, and an intriguing connection to the NBA.

6-month-old AI Startup Raised $175M

Led by Founders Fund, Cognition just launched the first AI engineer, Devin, and is now valued at $2 billion.

• On the SWE-Bench evaluation for software engineering tasks, Devin achieved 13.86% unassisted accuracy, far surpassing the previous state-of-the-art model's 1.96% unassisted accuracy.

• Unlike AI coding assistants like GitHub Copilot that provide recommendations to human developers, Devin works as an autonomous agent taking on full projects.

• Cognition Labs' founders are coding experts Scott Wu, Walden Yan, and Steven Hao who had reportedly turned down $1 billion valuations.

🗺️ AI Announcements

Less is More: The Rise of Small Language Models

Introducing Microsoft's Phi-3, a suite of high-powered small language models trained on curated data, enabling on-device AI with performance rivaling larger models.

• Microsoft has developed a new family of small language models called Phi-3 that can match or outperform larger models on various benchmarks, thanks to innovative training techniques using carefully curated high-quality data.

• The first model, Phi-3-mini (3.8B parameters), is now publicly available.

• Small language models require fewer computing resources, can run locally on devices without cloud connectivity, and provide more privacy by keeping data on-premises.

• Rather than replacing large language models entirely, Microsoft envisions a future where organizations employ a portfolio approach - using small models for simpler tasks and large models for complex reasoning over vast data.

• Small models can offload some work from large models in a complementary way.

✨ Vibe Board

Here are some of my favorite things from this week:

🫶 AI for Good

Tech Giants Unite to Safeguard Children from AI Exploitation Risks

As someone who has little nieces and nephews, this issue is close to my heart, here are the key takeaways:

1. Several leading AI and tech companies, including Google, Meta, OpenAI, Amazon, Microsoft, Anthropic and others have signed on to a set of "Safety by Design" principles aimed at prioritizing child safety when developing, deploying and maintaining their AI systems.

2. By integrating these child safety principles, the companies are protecting children and promoting ethical AI innovation.

3. The key safeguards outlined in the principles are:

   • Vetting training data to remove any illicit child abuse content.

   • Stress-testing AI models to ensure they don't generate dangerous child abuse imagery.

   • Only releasing AI models that have been evaluated for child safety.

👀 Job Board

Here are some awesome open positions in the AI space that I would love to see one of our readers fill:

• Global AI Go-To-Market Strategy Lead, Machine Learning | Microsoft

  • salary: $264,000

• Executive Briefing Center Presenter | NVIDIA

  • salary: $276,000

• Founding Account Executive | Koda Staff

  • salary: $400,000

• Content Manager, GenAI | Meta

  • salary: $223,000

You guys started re-posting my content on X, let’s be friends there too! Thank you 💛

What is IYKYK? (If you know, you know) 💅

I'm Sarah Stoker, a curious soul passionate about weaving the magic of emerging tech into our daily lives. I'm all about demystifying AI, making it feel like a coffee chat topic rather than a textbook chapter. Imagine blending the spark of human creativity with the wonders of AI – that's the adventure I'm on. I’m researching avenues for art that amplify our creative quests, and learning alongside you.