The Signal: ChatGPT advanced voice mode, how do we 10x GPT-5, and prompt engineering

In partnership with

Hey friends 👋 Happy Sunday.

Here’s your weekly dose of AI and introspection.

Want to feature here? Sponsor this newsletter.

AI Highlights

1. OpenAI releases ChatGPT advanced voice mode.

OpenAI is currently rolling out advanced voice mode to a small group of ChatGPT Plus users. In this mode, you can interrupt the AI with real-time conversations, and the AI can detect and respond to emotional cues.

Alex’s take: The full rollout is planned for autumn 2024, which is later than expected since they announced the feature back in May. That said, voice will transform how we interact with AI, moving past written text prompts.

2. Runway's Gen-3 Alpha turns still images into videos.

Gen-3 Alpha can create highly detailed and realistic videos up to 10 seconds long. This is a major shift in filmmaking, allowing anyone to create highly detailed and realistic videos.

Alex’s take: At $15 per month for one minute of video, the cost of video generation is still high. Runway is currently a front runner in the race to deliver fast, high-quality generative video tools, including the likes of OpenAI’s Sora, Luma AI, and Pika.

3. Nvidia demonstrates humanoid control with Apple Vision Pro.

Nvidia developers are using Apple Vision Pro to capture teleoperated demonstrations, which are then simulated in Nvidia's robotics simulation platform. This trains humanoid robots to perform tasks autonomously.

Alex’s take: I thought this was a neat workflow to reduce the need for extensive real-world data to cut both costs and time. With the announcement of Figure 02 coming next week, August is looking like an exciting time for robotics.

1 Article I Enjoyed

How do we 10x GPT-5?

Ori Eldarov is the Co-Founder & CEO of OffDeal, an AI-enabled investment bank for buying and selling small businesses recently backed by YC.

He wrote a thoughtful piece on “How do we 10x GPT-5?”.

The idea behind the article is that as we move from automating individual tasks to workflows and jobs, AI faces a compounding accuracy challenge.

Ori argues that this is why AI agents have seen limited adoption in enterprises so far. They lack the reliability required for real-world processes.

One question from this article I feel is especially important to highlight: “Is it possible to ‘future-proof’ yourself in a rapidly evolving job market?”

A skill that we’re already seeing become increasingly more important—prompt engineering.

1 Idea I Learned

Prompt engineering.

Prompt engineering is an essential skill to future-proof yourself in the era of AI.

It involves using natural language instructions to tell an AI model what you want it to do.

The more accurate your instructions, or “prompts”, the more accurate your output.

This enables you to harness the full potential of AI systems like ChatGPT.

Prompt engineering will become more and more relevant as tools become increasingly integrated into the workplace.

I feel so strongly about this skill that I spent 2023 developing a course alongside DataCamp to help everyone go from zero to one with prompt engineering.

You can check it out for yourself here.

1 Quote to Share

"Winners focus on winning. Losers focus on winners."

Source: @aubreystrobel on X

1 Question to Ponder

How will humans work alongside increasingly capable machines?

Our human gift of compassion, care and creativity is the most valuable asset to retain.

💡 If you enjoyed this issue, share it with a friend.

See you next week,

Alex Banks

P.S. how your email finds me.