malik muhammad jahangir’s Post

Humans have always had the edge over AI when it comes to thinking 🧠... But OpenAI just changed the game with their new model, o1. This AI doesn’t just give answers—it thinks. It breaks down questions into steps, just like we do. But here’s the catch: it’s better than most of us. While GPT-4o solved only 13% of tough problems, o1 hit 83%. In coding, it’s even more impressive, landing in the 89th percentile. This makes o1 a game-changer for science, coding, and math. It’s not just a step forward—it’s a leap towards AGI. And this could shake up industries, replacing millions of jobs. If you’ve got a ChatGPT Plus subscription, you can try it out today. What do you think—how close are we to AGI? 🤖

Humans have always had the edge over AI when it comes to thinking 🧠.... But OpenAI just changed the game with their new model, o1. This AI doesn’t just give answers—it thinks. It breaks down questions into steps, just like we do. But here’s the catch: it’s better than most of us. While GPT-4o solved only 13% of tough problems, o1 hit 83%. In coding, it’s even more impressive, landing in the 89th percentile. This makes o1 a game-changer for science, coding, and math. It’s not just a step forward—it’s a leap towards AGI. And this could shake up industries, replacing millions of jobs. If you’ve got a ChatGPT Plus subscription, you can try it out today. What do you think—how close are we to AGI? 🤖

In our second blog post of the "Code Green" blog series, we dive into the mechanics of AI and large language models (LLM's), like ChatGPT. We explore the technology behind AI, breaking down complex concepts such as tokenization, vectorization, and attention mechanisms. Understanding these foundations is crucial for software developers and IT managers looking to leverage AI responsibly and to sustainably apply it into your software development practices. Read our blog post on how we connect AI innovation with sustainable software development practices. Your code can make a difference! 🌍💻 🔗 Read the full post here: https://lnkd.in/e5-SxvsN

𝐎𝐩𝐞𝐧𝐀𝐈'𝐬 𝐨1: 𝐓𝐡𝐞 𝐀𝐈 𝐓𝐡𝐚𝐭 "𝐓𝐡𝐢𝐧𝐤𝐬" 𝐁𝐞𝐟𝐨𝐫𝐞 𝐈𝐭 𝐒𝐩𝐞𝐚𝐤𝐬! 🧠 OpenAI just dropped a game-changer: o1 (Project Strawberry/Q*). This isn't an incremental update - it's a whole new ballgame in AI problem-solving. 💡Key highlights: - Outperforms experts on PhD-level science questions - Solves 83% of International Math Olympiad problems - Topped the Norway Mensa IQ test with a score of ~120 - Ranks in 89th percentile for competitive programming What's o1's secret sauce? Advanced reasoning. Using reinforcement learning and chain-of-thought processing, o1 "thinks" before it speaks, mirroring human problem-solving. 🔍 AGI Progress: Word on the street is o1 might be hitting Level 2 on OpenAI's 5-level AGI blueprint. The implications? Huge. From scientific breakthroughs to creative innovations, we're entering uncharted territory. Exciting times, but also full of challenges. So, what's your take on this AI leap? How do you see it shaking up your industry? (FYI: o1 is now available to try out for ChatGPT Premium and Teams users!)

In this video, we will discuss the prompt engineering free course! https://lnkd.in/gieHR-rF In this you will Discover how to create different LLM applications easily. Get started with prompt engineering, explore open-source LLMs, and learn the best practices and techniques. Master the art of Prompt Engineering and unlock the full potential of Large Language Models (LLMs) like ChatGPT and Open Source LLMs. This course is perfect for anyone who wants to leverage the power of AI for tasks like content creation, code generation, and more! #analyticsvidhya #datascience #machinelearning #generativeai

It is good that some universities, including my current one, focus on actual industry skills. I can't recall the last college major I utilized while building anything during my undergraduate studies. I don't know when regular Indian colleges apart from HEIs will understand that to make LLMs, you don't have to write the internal code of RNN on paper. I am thankful to be part of such an esteemed institution that believes in the practical approach to learning. In the past, during undergraduate studies, we were told to write pure Python code for ANN on pen and paper once before coding it on a system because we didn't learn Python in college but wanted to use it for the final year project. And saying that "these are the basics" just mocks the students. You don't need Tensorflow's entire internal code at your fingertips; you need an overview of its workings and knowledge of how to read the documentation. #HigherEducation #SkillDevelopment #MachineLearning #AI #PracticalSkills #IndustryReady #DataScience #Python #LearningByDoing #FutureOfWork #TechEducation #LLMs #StudentVoices #InnovativeLearning #CareerSkills

A balanced approach to AI involves using it as a tool to supplement human expertise, not replace it. By letting AI take on routine or time-consuming tasks, we can free up time to focus on complex problems and develop deeper insights, ultimately improving our expertise and knowledge. It is equally important to stay updated with new AI advancements to maintain relevance in a rapidly evolving field. 

The new ChatGPT-4o model, named ChatGPT-4o-latest (2024-08-08), has been tested as "anonymous-chatbot" in the Chatbot Arena, where it reclaimed the #1 position. Now available also via API but only for research and evaluation, this update builds on GPT-4o with bug fixes, performance improvements, and notable advancements in coding, instruction-following, and handling hard prompts, based on experimental results and user feedback. https://lnkd.in/gyed_pgX

Do you think GPT-4 is the best coder? HERE comes Claude 3. I'd suggest everyone to watch the video in the comments. It shows how Claude can not only pass medium and hard Competetive Programing questions but also Expert level😱 I remember when chatgpt first came out in Dec 2022, my fellow coder friends were like "it can solve easy coding problems but no way it can solve Expert ones." BAM!!! AI supremacy!!!! Think what's gonna happen in the next 1-2 years? Let me know what you think in the comments. Think Turbo

Stanford just uploaded their new "Building LLMS" lecture. It's a must watch. "This lecture provides a concise overview of building a ChatGPT-like model, covering both pretraining (language modeling) and post-training (SFT/RLHF). For each component, it explores common practices in data collection, algorithms, and evaluation methods." https://lnkd.in/ePZEaZMW ↓ Are you a developer? Check out https://AlphaSignal.ai to get a daily summary of breakthrough models, repos and papers in AI. Read by 200,000+ devs.