Learn Prompting #4: OpenAI's Chief Product Officer on the Future of AI

Hey there, prompting people!

OpenAI’s Chief Product Officer, Kevin Weil, recently discussed his predictions about where AI is heading. Here’s our 3 takeaways from his talk:

1. AI is Evolving Rapidly—Be Ready to Adapt: AI’s capabilities are growing faster than expected. The key to staying ahead is being flexible and ready to adapt as new features and tools emerge.

2. AI Will Become Cheaper and More Accessible: OpenAI has dropped the cost of its AI models by 99% in just two years. This opens doors for small businesses, individual creators, and hobbyists to tap into tools once only available to big players. It’s time to think about how you can start integrating AI into your workflows.

3. AI Agents Are the Future: Following our previous discussion on AI agents, AI won’t just answer questions—it’ll perform complex tasks autonomously. The future is about AI agents handling entire workflows, managing projects, and collaborating with you on creative tasks. It’s a game-changer for productivity and innovation.

If you didn’t already know, AI Agents are the future… and are already here! Andrew Ng wrote about AI agents at the beginning of 2024:

Want to learn more about AI Agents? We created this beginner-friendly course for you to learn about AI Agents!

Other GenAI Market Updates

• Asana introduced AI Studio, automating workflows like task assignment and project tracking, freeing teams to focus on strategic work—no coding needed.

• Stability AI unveiled Stable Diffusion 3.5, offering faster, customizable text-to-image generation, making it easy to create high-quality visuals on standard hardware.

• Cohere launched Multimodal Embed 3, enhancing search and data retrieval by combining text and image processing.

• ElevenLabs introduced Voice Design, allowing users to generate custom voices and making it easier to create unique audio content quickly.

• Perplexity introduced Reasoning Mode, simplifying complex data analysis with deeper insights from vast datasets for quicker decision-making.

• Claude introduced its Analysis Tool, turning AI into a real-time data analyst, processing datasets and generating insights without external support.

• Runway launched Act-One, transforming live-action performances into animated characters and speeding up production for creators with minimal resources.

Updated Basics Guide

The Basics section of our Prompt Engineering Guide is the perfect place to dive into AI if you’re just getting started. It walks you through essential techniques and tips for using tools like ChatGPT, Claude, and Gemini.

We’ve recently given this section a big refresh, making it even more practical and easy to follow. Check out the updated guide to learn new prompting strategies, pick up useful tips, and try out some interactive tools to get the most out of generative AI!

The Latest AI Models

So many new models this week:

• IBM Granite 3.0 is an enterprise-grade, instruction-tuned model trained on over 12 trillion tokens, excelling in complex workflows and outperforming competitors on key enterprise benchmarks.

• Meta OMat24 provides a large dataset and pre-trained models for materials science, delivering state-of-the-art results in predicting material stability and formation energy.

• Google and University of Maryland Omni×R introduces an evaluation suite for benchmarking top multimodal models, including GPT-4o, setting new standards for model comparison.

• Microsoft BitNet b1.58 enables fast, lossless inference on 1-bit models, optimized for CPU performance with upcoming support for GPUs and NPUs.

• Deepseek Janus unifies visual encoding in a multimodal framework, surpassing task-specific models in both understanding and generation performance.

• DeepMind and MIT FLUID tackles scaling issues in vision models, providing significant advancements in text-to-image generation.

• Meta Spirit LM is Meta's first open-source multimodal model, seamlessly integrating text and speech, directly competing with leading multimodal models like GPT-4o.

AI/ML Red Teaming & AI Safety: Live Cohort

We’re starting the first cohort of the AI/ML Red Teaming & AI Safety masterclass shortly!

We gather 100 AI safety enthusiasts to share:

• Insights from HackAPrompt, the largest AI Red Teaming event ever held

• How to transition into a career as an AI Red Teamer

• 5 weeks of intensive, hands-on exercises + a final project

• A Certificate of Completion, and access to an exclusive AI Security job board on our website

Plus: You’ll also receive a 1-year license to Learn Prompting Plus, which gives you access to 15 of our other Generative AI courses courses ($299 value).

The good news is that you can join us! There are still spots available.

Thanks for reading! We’d love your feedback to make this newsletter even better. Our goal is to share content that’s valuable and relevant to you. Help us fine-tune future editions by sharing your thoughts on this week’s email—it’ll only take a moment!