12th issue: 1,000,000,000,000 dollar! 💰

🌯 Wrap for the week

A close look at all the main headlines in AI

One trillion with a T! 🤑 

Nvidia's market cap flirted with $1 trillion this Tuesday, joining the elite trillionaire club of Apple, Alphabet, Amazon, and Microsoft - albeit briefly.

Thanks to some stellar quarterly earnings, Nvidia had Wall Street going crazy. The main reason for such amazing results is the recent surge in demand for GPUs. Once the darling of hardcore gamers, GPUs have found new love with AI platforms like OpenAI’s ChatGPT and Google’s Bard.

On top of that, there have been some amazing announcements from Nvidia at the Computex Taipei Conference this week as well. TechCrunch runs a great coverage of some of the top announcements and you can check further here: https://techcrunch.com/2023/05/28/nvidia-computex-jensen-huang/

However, one thing I do want to point out is how NPC in games could have way better natural conversations with you thanks to the great advancement in Generative AI. The official name of Nvidia for such a feature called Nvidia Avatar Cloud Engine. But to explain how it works, you should check it out how it works 👇️ 

Finally, no more oblivion NPC dialogue like this in the future

News credit:
Nvidia crosses into $1 trillion market cap before giving back gains from The CNBC

Let’s fight against hallucination, step-by-step

OpenAI has declared war on AI "hallucinations".

In an era where the line between AI-generated fake news and the next Avengers sequel plot is blurrier than your vision after a tequila shot, OpenAI is striving to train AI models not to generate falsehoods.

It's like teaching a parrot not to squawk nonsense but with far-reaching implications for the things like the upcoming 2024 presidential election in the US. OpenAI’s new strategy involves AI models rewarding themselves for each correct reasoning step. The new method called “process supervision” could lead to better explainable AI and fewer faux pas. Critics, however, are still sceptical about its efficacy, hence they are calling for more transparency about the approach. A peer review is likely, but when will these changes land in ChatGPT and other OpenAI products? Well, that's still a mystery.

By the way, this method sounds a lot like when I was studying math in primary school. Our teacher always asked for the details steps of how you get the final result in a math problem. I guess the reward system back then was quite similar to this since we could get points for the steps even though my final output is incorrect. I guess this sort of simple trick could do magic on a bigger scale.

News credit:
OpenAI is pursuing a new way to fight A.I. ‘hallucinations’ from CNBC

Other top AI headlines for the week

US colonel retracts comments on simulated drone attack ‘thought experiment’ from The Guardian

This has been shared like wildfire for a couple of days. I am glad to know it is just a thought experiment, otherwise, the progress with AI will certainly be stopped significantly. On the other hand, it is still important to point out that the reason for such a “thought experiment” conclusion largely comes from the fundamental fear of how AI will not progress without aligning with the goal of human beings. The alignment problem is still yet to be solved at this stage.

Andrew Ng also has a thread about this news:

AI Is Writing Code Now. For Companies, That Is Good and Bad. from Wall Street Journal

Just to clarify, based on my reading of the reporting, this is more like a “thought experiment” as well. However, I do understand the concern that since AI can now write code so fast and can be written/requested by anyone, the burden is now on back to the human to distinguish the good vs the bad code and make a decision if any change needs to be incorporated in their codebase. I could see more applications and maybe could be more specialised in this area, i.e. evaluating code quality.

New superbug-killing antibiotic discovered using AI from BBC

Now that is another great example of how AI could help improve the productivity of humans and create amazing outcomes for humanity. I wish to see more of these types of news in many more fields in the future.

🎬️ Guess that movie

Adding a bit of fun to your weekend reading, let’s guess the famous movie from the 10-emoji description created by ChatGPT:

⚾️👨‍💼💻📉📈💡📝��🏆👏

If it is a bit hard to figure out, you can find the answer with some explanation from ChatGPT at the end of this newsletter.

🧰 Toolbox

Check out this week’s tools and resources that give you more superpowers with prompt crafting!

1. 📚️ Improving mathematical reasoning with process supervision
If you want to read in further detail about the approach that OpenAI is taking to combat hallucination, you should check this original blog post from OpenAI about “process supervision”.

2. 📚️ Tree of Thoughts: Deliberate Problem Solving with Large Language Models
Since we are talking about the approach to improve the quality of the output from LLMs, here is another suggestion that could do. Here is the quick differentiation between the different approaches:

3. 🎙️ Ask us anything: AI & ChatGPT from CheckITout
This is a podcast episode from a subscriber of the Prompt Craftsman. It provides a great introduction to ChatGPT and its use cases in the business at the moment. Give it a listen!

👋 Your next Twitter friend

Connect with great people in the industry to supercharge your growth

🖼️ Image of the week

Generative fill within Photoshop has been all the craze this week. I must say the album cover fill is one of the great and fun applications of this feature. Check some amazing results 👇️ 

🎬️ Guess that movie - answer

The answer is "Moneyball" 🥳 

Did you guess that right? Here is ChatGPT’s explanation:

This summary captures the main elements of the story: baseball (⚾️), a businessman or manager (👨‍💼), using computers for analytics (💻), a downturn (📉), an upturn or improvement (📈), a new idea (💡), planning and strategy (📝), financial constraints and success (💰), aiming for a championship (🏆), and finally applause or recognition (👏).