🤝 AidfulAI Newsletter #11: One Year Content Creator Anniversary
Dear curious minds,
Welcome to the latest issue of the AidfulAI newsletter. Today is a special day: I started my content creator journey with a tweet about the BASB book from Tiago Forte exactly one year ago, and since then, I have been exploring and sharing different AI tools and insights with you. Thank you for being part of this journey and for your support and feedback. As a new chapter in my journey, I published my first two Medium articles and would love to greet you also there as a follower.
Major AI News
🎉📷 New feature in Bing Chat: Image Inputs!
Bing Chat, which runs a version of GPT-4 and uses web-search to generate its answers, has rolls-out a new feature that lets you upload an image to your prompt. The image will be analyzed by the chatbot and used to generate a response.
For example, you can upload a picture of a painting and ask the chatbot to write a poem about it. The chatbot will also blur any faces in the image to protect your privacy, but it is not stated if that happens locally in your browser. However, the AI doesn't yet know that the blurred faces are only created for privacy reasons and should not be mentioned in the analysis.
👎🌐 ChatGPT Disables Web-Search Feature for Plus Subscribers Due to Legal Concerns
ChatGPT has temporarily disabled its Browse with Bing feature for Plus subscribers. OpenAI stated that the model sometimes includes the full-text from a webpage in its answer, and this could lead to legal issues related to copyright. Therefore, OpenAI has decided to remove this feature until it finds a way to handle this issue.
So far, the Bing Chat still uses the web-search and my tries to get a full-text from a webpage all failed. It feels somehow strange if Microsoft offers the model with web-search for free and OpenAI disables the feature for their paying customers.
Privacy-Friendly AI
🤖📞 How to use bots to prank telemarketers and scammers
If you're tired of getting annoying calls from telemarketers and scammers, there's a fun and easy way to fight back. A man in California has created a service that lets you use bots to waste their time and money.
The service is called Jolly Roger Telephone Company, and it works by forwarding or merging the unwanted calls to a number that connects them to a bot. The bot uses AI to talk to the caller and keep them on the line for as long as possible, while you can listen in or hang up. The bot can also press keys to get past the automated messages and reach a human scammer. There are different bots with different voices and personalities, such as an old man who rambles about random things, or a mom who yells at her kids in the background. The service costs $25 per year, and you can try it for free for 30 days. It's available in the US, UK, Australia, and New Zealand. You can find out more on the Jolly Roger website, where you can also listen to some hilarious examples of the bots in action. This is a great way to make the scammers regret calling you and prevent that they scam others in the time you keep them busy.
PKM and AI
📚🤖 Claude 100k: An AI Which Understands Books, Reports and Very Long YouTube Videos
Claude 100k are amazing LLMs that can understand and generate natural language of up to 100,000 tokens, which corresponds to about 75,000 words.
In contrast to the GPT models from OpenAI, these models from Anthropic can handle long inputs such as complete books, reports, or video transcripts. You can use them to summarize, rewrite, or create content, especially with super long inputs. If you would like to learn more about these models and especially where you can use them, read my Medium article.
🤖🧠 Using AI to Supercharge the CODE Framework: A New Era in Information Management
In the previous issues, I showed you how you can use AI to apply the CODE framework from Tiago Forte to process information. The CODE framework is a four-step process: Collect, Organize, Distill, and Express and by using AI, you can make each step easier, faster and save time, energy, and attention. I have combined these tips and insights into a slightly updated Medium article.
Podcasts
👩🔬👨💻 Brian Roemmele on AI, Wisdom, and Loneliness
In episode 168 of the Infinite Loops podcast, Jim O’Shaughnessy talks to Brian Roemmele, a scientist, researcher, analyst, connector, thinker who has a passion for AI.
They explore how large language models (LLMs) are like shadows on the wall of Plato’s cave, how AI is a gift and a challenge for humanity, and why we need to run our own AI models locally. They also discuss the use cases and the history of AI, the decline of wisdom and how to preserve it, the problem of loneliness and abandonment in modern society, and how to have a conversation with your local documents. Brian shares his insights on how wisdom can help us shape our future, and how we can recover the lost knowledge of the past. Tune in for a really fascinating and inspiring conversation!
🔮🧠 George Hotz Hypothesis about the Structure of GPT-4
George Hotz is an American hacker, entrepreneur, and software engineer. He is known for developing iOS jailbreaks, reverse engineering the PlayStation 3, and for the subsequent lawsuit brought against him by Sony. He is also the founder of comma.ai, a vehicle automation company that aims to make self-driving cars accessible and affordable. George recently left comma.ai to work on a new deep learning framework called tinygrad. In episode 387 of the Lex Fridman Podcast he elaborates on various AI related topic including his insights about the internal structure of GPT-4, which is not proofed but very interesting.
George Hotz has a creative and visionary mind, even if he sometimes causes controversy or breaks conventions. It is a lot of fun to follow the conversation of him with Lex Fridman, and I recommend you to listen to it yourself.
Disclaimer: This newsletter is written with the help of AI. I use AI as an assistant to generate and optimize the text. However, the amount of AI used varies depending on the topic and the content. I always curate and edit the text myself to ensure quality and accuracy. The opinions and views expressed in this newsletter are my own and do not necessarily reflect those of the sources or the AI models.