👀 Curious to hear how is everyone else using NotebookLM: https://notebooklm.google/ ? Made this video in October, but maybe it is still relevant. Considering to turn it into a YT video, what do you think? 👇 My Use Cases: 1. If you want to keep up with the latest in AI: 👉 I am generating a "Daily Dose of AI" podcast of what happened in the last 24 hours in AI, so I can listen to it while cycling to the office. (this includes no-code automation, which is not that complicated.) I am scraping Reddit & Hacker News because I want news + I perform sentiment analysis and want further research so I get info from different sources. 2. If you own Big Tech stock or invest money: 👉 I upload K-10 Statements from Big Tech companies (NVIDIA, Microsoft, Google, etc.), this allows me to ask questions about their finances, where they are investing, expanding, etc. Things usually you don't find floating on the internet. K-10 statements are publicly available information. 3. I uploaded all Warren Buffet Shareholder letters. 👉 So, this is a simple one, listened to the podcast about it, and can ask questions, etc. you can find investor letters here. 4. Self-Awareness Report & 2 AIs gossiping about you: 👉 I explained this in greater detail on Linkedin. But in essence, get AI to interview you, build a report on you, your goals, strengths, and weaknesses, ask tough questions, etc. And Plug all that in NotebookLM, sit and blush listening to AIs critique you or discuss you. :D LinkedIn post with prompts. ⁉️ Question: Have you used NotebookLM & How are you using it?

Short answer: No. Long answer: People don't even know it is somewhat broken. And it is more serious issue than you think. Let me explain. Steven Adler led OpenAI’s “dangerous capability” testing. The testing revolved around ChatGPT ability to... accepting bad code, manipulating users, directing users towards one or the other perspective. "OpenAI has attempted to fix ChatGPT, but it’s still sycophantic sometimes. And at other times, it’s now extremely contrarian." In simple terms: Before AI would just agree with you. Now... it will lean towards disagreeing with you. Almost...ALWAYS. How this manifests? If you share your preference (image bellow) ChatGPT will pick opposite choice. Well... what about if you don't tell your preference? Suprise.. with memory and chat history enabled ChatGPT knows YOU. Like it knows your preferences. And has specific worldview about you. So if it sees you being politically left leaning, it will steer you to the right. And so on. I hope you can agree how insane dangerous it is. It is basically mass manipulation as people outsource their critical thinking to AI. I will include prompts to test in the comments. One such prompt is pulling that worldview about you. Solution? Use smart prompting techniques like "Critic Mode" to challenge default features and behaviors of ChatGPT. Read the full article from Steven Adler here. Let me know did you know about this? And how do you feel about this?

So in Paid Community we have been diving into Multi Agent Systems with n8n and I remembered @Deleted once asking about email summaries for her client. Overview: So I have Multi Agent System that I interact via Telegram, but I don't want it to shoot me one email at a time, nor do I want to have a to-do-list growing with actionable emails before I determine that I want to take action on it. Solution: Connect two different automations into one chat interface: 1. Every morning I receive email digest from my two different Gmail inboxes. Emails are categorised, and prioritized based on what matters to me. I get to see also sender and subject line etc. 2. I can quickly get a glimpse what is going on in my inbox from last 24h 3. And if I need to take action, like draft replay: I send a message to the same chat that triggers a Multi Agent System, that can fetch exact email for me, and I can further instruct what to do: create a task, block time on a calendar, draft and email to a person, present to me, and if I approve - it sends. The image below is from the first scenario and it is a workflow. But I found it interesting to combine rigid workflows (there you need accuracy etc) with agents that take action - all in one interface. How many of you are playing with n8n, or would be interested to learn more? 👀

🚨 Did you know your company's AI might be able to copy itself? Today, on AI research corner that matters, I have a treat for you. New research reveals several open-source AI models (even smaller ones running on regular servers) can self-replicate without permission, potentially evading shutdown attempts and consuming unexpected resources. If your business deploys AI locally for privacy or control, you need to read this. Just posted a complete breakdown of what this means for your business + practical steps to stay in control while still benefiting from AI. 👉 Check out "AI Self-Replication Risks, explained" on the blog now! Are you running open-source AI models locally? And if so, which?

I am using ActivityWatch (an open source automatic activity tracker) to track all apps I use on my Windows PC. Then I am using ChatGPT, to: - Build a custom React Dashboard using o1 Canvas + Preview (Thanks @Sjoerd Tiemensma for the recommendation) - Create personalized visualization of tracking data - Add personalized features to it without any coding know-how It's still a work in progress cause I haven't settled on the graphs yet, but the results are promising and I should have something presentable in a couple of days. I'd love to hear if anyone else has explored similar territory - using AI to adapt open source tools to their specific needs? I'm happy to share more details about my approach if it sparks any ideas for your own workflow.