How I use AI to organize my thoughts and put them into action? 📝
It all comes down to a simple approach.
I've noticed recently that sometimes I have so many things on my to-do list that I just feel overwhelmed and pressured in an unhealthy way. So I've decided to test an idea that popped into my head and see if it will bring more clarity or just more chaos to the equation.
Basically, this approach is very simple. It comes down to a prompt followed by a conversation or, as I like to call it: the ultimate brain dump (in simpler words, flooding the screen with the stream of your consciousness).
Prompt:
"I am going to type or speak for a while. Please only reply with "ok", until I explicitly tell you that I am finished. Once I'm done, you will:
1. Provide a concise summary of my thoughts a paragraph format, approximately 100 words long.
2. For each distinct thought or idea expressed:
a) Briefly restate the thought (1-2 sentences)
b) List 2-3 actionable steps I can take based on this specific thought
3. Offer 2-3 additional insights or suggestions related to the topic."
After pasting it and hitting enter, just let your thoughts do the work.
You can steal this method and test it out in your reality. You can also build upon it until it fits your needs.
I am very curious about the results you will get.🗝️ Quick Bytes:
Anthropic unveils Claude Sonnet 3.5 AI model
Anthropic has unveiled its latest AI model, Claude 3.5 Sonnet, which is claimed to match or surpass the capabilities of OpenAI's GPT-4 and Google's Gemini across various tasks. This new model, now available on web and iOS platforms, is reported to be twice as fast as its predecessor and outperforms competitors in several benchmarks. Claude 3.5 Sonnet boasts enhanced abilities in coding, multistep workflows, chart interpretation, and more human-like text generation.
Alongside the new model, Anthropic introduced a feature called Artifacts, allowing users to interact directly with Claude's outputs within the app. This addition, coupled with Anthropic's focus on business applications, signals the company's intention to transform Claude into a comprehensive tool for companies to centralize knowledge and work.
Ilya Sutskever is launching Safe Superintelligence Inc.
Ilya Sutskever, a founding member and former chief scientist of OpenAI, has launched a new artificial intelligence company called Safe Superintelligence Inc. (SSI). Announced in a blog post, SSI aims to develop a safe and powerful AI system with a singular focus on balancing safety and capabilities. The company positions itself as a startup that can accelerate AI development without compromising on safety, free from the distractions of management oversight or product cycles that larger corporations face.
Joining Sutskever in this venture are Daniel Gross, formerly of Apple, and Daniel Levy, previously at OpenAI. The announcement comes after Sutskever's departure from OpenAI in May, following his involvement in an attempt to replace CEO Sam Altman last year. SSI's launch also follows the resignations of other prominent AI researchers from OpenAI, including Jan Leike and Gretchen Krueger, who cited concerns over the prioritization of marketable products over safety processes.
Microsoft’s all-knowing Recall AI feature is being delayed
Microsoft has announced a delay in the release of its Recall feature for Copilot Plus PCs, originally scheduled to launch next week. Recall, which captures users' activities on their computers, raised concerns among privacy advocates and security experts. In response, Microsoft had planned to implement security improvements, including making it an opt-in feature, encrypting the database, and requiring Windows Hello authentication. However, the company has decided to postpone the launch to conduct further testing through the Windows Insider program.
The delay comes shortly after Microsoft's vice chair and president Brad Smith emphasized the company's commitment to security in a testimony before the House Homeland Security Committee. Smith stated that security is even more important than Microsoft's work on artificial intelligence and announced that cybersecurity would be incorporated into employee performance reviews. The postponement of Recall suggests that Microsoft needs more time to ensure the feature meets its security-focused promises and high standards for quality and security.
🎛️ Algorithm Command Line
AI use cases outside of work are my favourite. Here are 5 recent ones:
I spent last week far from the screen, with the goal to reset my brain. The days were full of activities - including high intensity hiking, comfy food, and listening to the sound of nature.
Many people struggle to identify where they can use AI and LLMs outside of the work environment.
So, I want to share five simple use cases that I've spontaneously used in the last couple of days.
Planning the carb intake for hiking
As a diabetic, my physical activity requires careful planning. I used LLMs to create personalized plans for hiking, focusing on carb intake during training.
It took off a lot of mental strain and saved me time. AI helped me determine the optimal amount and timing of carbs to consume based on the intensity and duration of my hikes.
Analyzing the performance and predicting the blood glucose
By combining data from my Garmin watch and my CGM, I created a routine in a demanding environment that resulted in more predictable blood sugar levels.
LLMsanalyzed patterns in my data and provided insights on how different factors, such as exercise intensity and food intake, affected my blood glucose levels.
Personalized insulin regimen
To avoid injecting too much insulin during hiking, I used AI to generate precise, reduced doses based on my data. This resulted in easier blood glucose management throughout the trip.
Models took into account factors like my activity level, carb intake, and historical blood glucose trends to recommend optimal insulin doses.
Meteograms and weather at a glance
I asked models to analyze meteograms and blend the information with the clothing I wanted to prepare for the trip.
This resulted in more efficient packing and significantly reduced the weight of my backpack.
Guesstimating the macros when eating out
The hardest part for me is when I can't weigh my food or access labels, which is obvious when eating out.
GPT handled this task well, guesstimating the amount of carbs, fats, and protein for dinner after the activity. While it struggled sometimes, no one's to blame here.
Here's the thing:
You don't need to create new, shiny use cases each time. For me, it comes down to being intentional and recognizing where I can offload some of my usual mental load.
When you allow yourself to experiment more and sometimes even do stupid things, you can build effective habits faster.
What do you think?
How have you recently used AI outside of your work? Have you found any particularly helpful or creative applications?
🗞️ Longreads
Fascinating take from Lenny’s Newsletter, and an insight into how Perplexity builds their product. Worth a read if you’re working on AI-driven products, and need some inspiration.
Some things here surprised me.
