- The Prompt Craftsman
- Posts
- 2nd issue: 🤯 Oh man, the news just keep coming!
2nd issue: 🤯 Oh man, the news just keep coming!
ChatGPT with internet, "an iPhone moment" and much more

🌯 Wrap for the week
A close look at all the main headlines in AI
ChatGPT just got the power of the internet and so much more
OpenAI's ChatGPT can now browse the internet with the help of plugins, which extend the bot's functionality by granting it access to third-party knowledge sources and databases. To be honest, I can’t believe they would release such power-ups this quickly.
My personal favourite is obviously OpenAI's first-party web-browsing plugin. It allows ChatGPT to draw data from around the web to answer various questions posed to it. If you that ChatGPT’s training data is only up to 2021. With this browsing capability, the improvement in these responses is unthinkable.
The plugin retrieves content from the web using the Bing search API and shows any websites it visited in crafting an answer, citing its sources in ChatGPT's responses. Basically, if you have tried Bing, the experience is quite similar to that.
See an example of the web browsing capability below:
Just. Wow.
@OpenAI's just showcased a #ChatGPT plugin for summarizing anything from the web!
Watch it outlining Hackernews' responses to yesterday's `plugins` launch! 🤯👇
— DataChazGPT 🤯 (not a bot) (@DataChaz)
8:13 AM • Mar 24, 2023
Other use cases for ChatGPT with plugins include:
Code interpreter for ChatGPT that provides the chatbot with a working Python interpreter in a sandboxed environment
Using the ChatGPT Code Interpreter plugin to make a simple chess library, play part of a game and then save it as a gif without leaving the UI:
openai.com/blog/chatgpt-p…
— Andrew Mayne (@AndrewMayne)
5:44 PM • Mar 23, 2023OpenTable plugin allows the chatbot to search across restaurants for available bookings
Instacart plugin lets ChatGPT place orders from local stores
Retrieval plugin enables ChatGPT to access snippets of documents from data sources like files, notes, emails or public documentation by asking questions in natural language.
I am still waiting to get access to the plugins and will share my first-hand experience with you all once I can get my hands on it. Keep an eye out for this newsletter!
News credit:
OpenAI connects ChatGPT to the internet from TechCrunch
“The iPhone moment of A.I. has started”, says Jensen Huang, NVIDIA CEO
Nvidia has given companies a way to build their own version of ChatGPT essentially using their new cloud services. It has unveiled a trio of cloud services to help businesses create and deploy their own artificial intelligence models based on individual data and requirements.
The services, called Nvidia AI Foundations, including
Nvidia NeMo, which offers access to foundation models that create text based on user input;
Nvidia Picasso, which can produce images, videos and 3D models.
Nvidia BioNeMo, launched in 2022 and generates scientific text based on biological data.
Companies can customise the foundational models to build applications such as professional content creation, intelligent chat and customer support, and digital simulation.
Adobe, Getty Images, Morningstar, Quantiphi and Shutterstock are among the companies already using the new tools. Nvidia AI Foundations uses the Nvidia AI Enterprise software suite and the company's GPU hardware running in the cloud.
News credit:
Nvidia introduces cloud services for customizing artificial intelligence models from VentureBeats
Other top AI headlines for the week
Google’s answer to ChatGPT, Google Bard, is out from Ars Technica
Unfortunately, Bard is only available in US and UK at the moment. Until I get to try it, I think there still might be a long way for Bard. Here is why I have this kind of thought towards Bard:
Just tried Bard. It's great twitter.com/i/web/status/1…
— Acorn1010 (@theacorn1010)
4:54 AM • Mar 24, 2023
Adobe launches generative AI tools aimed at marketers from TechCrunch
Adobe Firefly, the image generation tool that is part of Adobe Sensei GenAI, is what I found quite interesting. My key takeaway from the comparisons between Firefly and midjouney is getting specific words onto the image itself. If that is your thing, you should certainly check Adobe Firefly out. Here is a video that compares it to midjourney v5:
GitHub Copilot X: Revolutionizing software development with GPT-4 integration from Tech Wire Asia
I personally haven’t used Copilot but knowing the following it does make sense that Github is expanding the capability of Copilot with more AI.
Currently, GitHub Copilot is responsible for generating an average of 46% of code when using the tool, up from 27% of code in June 2022.
If you are curious to see how it gonna work out, check the article from GitHub.
After text and static art creation, it is a no-brainer to consider the creation of video from the text as well. Check some sample results here:
Generate videos with nothing but words. If you can say it, now you can see it.
Introducing, Text to Video. With Gen-2.
Learn more at research.runwayml.com/gen2
— Runway (@runwayml)
12:57 PM • Mar 20, 2023
Apart from text-to-video, there are also other modes that allow you to utilise the AI capability for video creation. Check these interesting modes here from Runway.
📖 Learning note
Concepts, techniques, and ideas I have learnt to become better at crafting prompts
As shared in the tips I have shared this week, being specific with your prompts would always help with getting better results from text generation like ChatGPT and image generation like midjourney.
The way I look at this reminds me of the following image:
Although AI might not be able to build you a swing yet, I think the idea is the same. If you are not specifying that you would like a tyre on the tree, you might get a rollercoaster from AI.
Just to clarify, by being specific, I mean describe what is the problem in detail and include clues regarding what results you would like to see. In the case of ChatGPT, I personally find giving examples works brilliantly. Be sure to try that in your next prompt.
🧰 Toolbox
Check out this week’s tools and resources that give you more superpowers with prompt crafting!
📺 Sam Altman: OpenAI CEO on GPT-4, ChatGPT, and the Future of AI | Lex Fridman Podcast #367
You know it is a good episode when you have a great interviewer, Lex Fried man, and the one in the news spotlight, Sam Altman. I have yet finished this episode but what I have heard so far is really good.
🔧 PromptVine
This is a collection of interesting prompts to try. The list of prompts is still relatively short at the moment, but it is a good concept to see what are the good prompts.📚 The Dall-e 2 prompt book
I am not forgetting the fans of DALL·E 2 in my newsletter. Here is a good free book on crafting prompts for this other amazing tool from OpenAI.
👋 Your next Twitter friend
Connect with great people in the industry to supercharge your growth
I’d like to see your perfect room.
Prompt: compact cozy small kid’s room interior::5 rounded corners::4 3d isometric::6 blender:: light purple pastel ::5 plastic style::4 --v 4 twitter.com/i/web/status/1…
— Kris Kashtanova (@icreatelife)
4:32 AM • Mar 26, 2023
🖼️ Image of the week

Prompt: ChatGPT with super powers
Please ignore midjourney’s best effort to try to write ChatGPT on an image. Other than that, I must say I quite like to see ChatGPT as a chubby super-batman. 😆
That is it for the week! Until I see you next time, stay awesome my friend!
Cheers,
Minjie
Reply