- AI Fire
- Posts
- 🖥️ 10-septillion-year Problem Solved in 5 Mins?
In partnership with
Plus: Your AI doctor is here?
Read time: 5 minutes
Believe that? Maybe you are living in the multiverse (like Marvel’s world). This is just the beginning. What a time to be alive! Scroll down to learn what is happening!
What are on FIRE 🔥
AI INSIGHTS
🤖 GenAI is Being Adopted at Double The Pace of PCs or the Internet
A working paper shows that generative AI has been adopted faster than personal computers or the internet. In just two years, its adoption rate is 39.5%, while the internet and PCs reached 20% at the same time.
Key takeaways:
28% of employed individuals used generative AI at work, with 10% using it daily, with 24.2% using it at least weekly.
Adoption is faster compared to technologies like PCs and the internet.
Between 0.5% and 3.5% of work hours are assisted by generative AI.
Generative AI could boost labor productivity by up to 0.875% under current adoption levels.
Generative AI adoption outpaces both PCs and the internet, largely due to lower costs and broader accessibility.
Why it matters: Understanding the speed and intensity of generative AI’s adoption is essential for gauging its economic impact. As AI becomes an increasingly integral tool for workers across occupations, it offers the opportunity to significantly boost productivity—potentially improving labor efficiency by up to 0.875% points at current usage levels. This could result in far-reaching economic benefits, particularly in terms of workforce innovation and performance enhancement.
PRESENTED BY CONVERGENCE AI
Meet your own personal AI Agent, for everything…Proxy
Imagine if you had a digital clone to do your tasks for you. Well, meet Proxy…
Last week, Convergence, the London based AI start-up revealed Proxy to the world, the first general AI Agent.
Users are asking things like “Book my trip to Paris and find a restaurant suitable for an interview” or “Order a grocery delivery for me with a custom weekly meal plan”.
You can train it how you choose, so all Proxy’s are different, and personalised to how you teach it. The more you teach it, the more it learns about your personal work flows and begins to automate them.
TODAY IN AI
AI HIGHLIGHTS
4️⃣ Day 4 of 12 days of OpenAI: Canvas is now fully available on the web and the ChatGPT desktop app for Windows! This feature lets you collaborate with ChatGPT to draft, edit, and refine writing or code. It’s accessible to all users in the 4.0 model.
💾 Google’s Willow chip solved a problem in 5 minutes that would take today’s fastest supercomputer 10 septillion years—10,000,000,000,000,000,000,000,000 years (yes, longer than the universe has existed).
Google’s new quantum chip has 105 qubits with advanced error correction, focusing on stability, a major leap for practical quantum computing. It could revolutionize medicine, AI, and energy (think fusion).
The twist: Google believes this breakthrough supports the multiverse theory, where computations happen across parallel universes, as predicted by David Deutsch.
P.S.: It is true that Google has made new advancements in quantum computing. However, during every Bitcoin uptrend season, similar news emerges, implying that quantum computers could break Bitcoin (this started about 4 years ago). Google's advertising versus what it can actually achieve is still far apart.
⚽ Manchester City partnered with Puma and invited fans to design an AI-created kit, the winning design becomes their 2026-2027 third uniform.
🚀 Nubank used Devin to refactor millions of code lines, reducing ETL migration from 1.5 years to 2 months. Devin boosted engineering efficiency by 12x, minimizing developer workload as Nubank scaled to serve 110+ million customers.
🩺 Elon Musk says: “You can upload anything from a phone picture of your blood test results to an MRI or X-ray and Grok will analyze it for you”.
🧠 Early adopters are moving from Google to AI chatbots like Perplexity (15 million users) and ChatGPT, reducing Google use by 90%. OpenAI smartly plans to introduce ads on GPT search, positioning itself ahead of the trend and disrupting the search economy.
DAILY AI FUNDRAISING
Former OpenAI researcher Alexis Conneau (who created OpenAI’s famous voice bot) has raised $40 million for WaveForms AI, a startup focused on developing emotionally intelligent audio AI.
AI SOURCES FROM AI FIRE
TOGETHER WITH VULTR
Deploy AMD Instinct™ MI300X on Vultr
AMD Instinct MI300X accelerators are now available on the Vultr. With thousands of GPUs ready for use & 32 data center locations worldwide. Vultr ensures reliable & high-performance computing for clusters of any size without vendor lock-in.
NEW EMPOWERED AI TOOLS
📅 Pickle creates a clone to stand in for you during video meetings
🤖 Webhooks in Notion let you trigger automation directly from your documents and databases
📝 uList is a smart list that organizes anything and plans everything
🎬 Magic Clips by Submagic turns long videos into viral shorts instantly with AI
💼 Peek AI builds a professional, shareable online portfolio in seconds with AI
AI QUICK HITS
🤖 China’s new AI-powered robot cop tracks criminals with facial recognition (Link)
🩺 AI tool could improve breast cancer detection by 21%, study finds (Link)
🎶 AI quiz tool gives real-time feedback and even sings a triangle rhyme (Link)
🎉 Congrats to 2024’s CEO of the year: Lisa Lu (leader of AMD) (Link)
🎙️ YouTube’s AI-powered dubbing is now available to many more creators (Link)
AI MUST-READ
Key Differences:
o1-preview: A general-purpose model, useful for a wide range of tasks.
o1-mini: Specialized in STEM (Science, Technology, Engineering, Math) tasks, like coding, making it particularly useful for solving technical problems step by step.
4o (GPT-4): While it's a powerful AI, GPT-4 can't break down complex problems into smaller parts or follow a step-by-step logical flow on its own, which is where o1 models shine.
Why o1 is Different:
After the release of GPT-4, OpenAI implemented a "Chain-of-Thought" process where the AI breaks down complex problems like a human would. This is done by providing step-by-step training data, allowing the model to mimic human-like thinking. This stepwise approach is key to making o1 more intelligent and efficient.
Example in Action:
The writer describes how they were able to provide o1-mini with a task to create a new module for a SaaS product, which would normally take a day to solve manually. By giving o1-mini detailed documentation, it provided a step-by-step guide on how to approach the feature development, including code, security, and deployment instructions.
After integrating the code from the AI, the result was successful with no errors, which greatly impressed the writer and saved them significant time.
Why Use o1-mini:
The o1-mini model is highly efficient for coding tasks. It can break down complex issues, provide accurate solutions, and generate code with minimal errors.
It's recommended to use the API version rather than ChatGPT for tasks that require flexibility and context because the API allows you to set your own system prompt, while ChatGPT has more restrictive guidelines that may limit its effectiveness.
Key Takeaways:
o1-mini is especially effective for coding and problem-solving in technical areas.
Providing as much context as possible helps the AI generate better solutions.
The API version of models (like o1-mini) is preferred over ChatGPT due to its more flexible system prompt.
The overall theme is that o1-mini is a breakthrough in AI’s ability to assist with complex technical tasks efficiently, saving time and effort for developers.
AI JOBS
We read your emails, comments, and poll replies daily
How would you rate today’s newsletter?Your feedback helps us create the best newsletter possible |
Hit reply and say Hello – we'd love to hear from you!
Like what you're reading? Forward it to friends, and they can sign up here.
Cheers,
The AI Fire Team
Reply