Some quick observations from trying out the o1-preview from OpenAI. 1. These new models reduce the need to master prompt engineering will make conversational AI tools more useful for a broader audience 2. The new models reach a "final" good response quicker even though these models take longer to think by avoiding the need for longer conversations 3. The responses sound more convincing, making errors and mistakes in responses harder to identify There are lots of technical innovations and improvements in these models. If you'd like to see a discussion on the importance and challenges with reasoning check this excerpt from a longer podcast with Yann LeCun. https://lnkd.in/gi-6XdFQ
Kamran S.’s Post
More Relevant Posts
-
LLMs are fundamentally parrots. Would you let a parrot run your key business processes? No. Would you let it do very specific, well-defined tasks with careful oversight? Maybe. Narrow scope is key to effective #automation using #AI.
Can LLMs reason? | Yann LeCun and Lex Fridman
https://www.youtube.com/
To view or add a comment, sign in
-
🚀 Ever wondered if AI can truly "think" like humans? 🧠 Yann LeCun, a pioneer in AI, sheds light on this intriguing question! LeCun explains that current Large Language Models (LLMs) excel at tasks that require instinctive, automatic responses—what he calls System One thinking. Examples include simple pattern recognition and routine tasks. However, when it comes to System Two reasoning—tasks that require deliberate, planned thought—LLMs fall short. These include complex decision-making and strategic planning, where a deeper understanding and more thoughtful consideration are needed. The current models predict the next word in a sequence, which works well for generating text but lacks the depth needed for true reasoning. They are not capable of advanced planning or optimizing their answers in an abstract, meaningful way. LeCun envisions future AI systems overcoming these limitations through energy-based models. These models will measure how well an answer fits a prompt, optimizing responses efficiently and thoughtfully. This approach promises a significant leap forward in making AI more capable of reasoning like humans. 🌟 Curious about AI's capabilities and limitations? Watch the video and join the conversation!
Can LLMs reason? | Yann LeCun and Lex Fridman
https://www.youtube.com/
To view or add a comment, sign in
-
NEW: Watch the latest #KempnerInstitute Seminar Series talk: Yoav Goldberg of Bar-Ilan University and Ai2 describes his work on the limits of LLMs, and how new abilities become possible when LLMs are embedded in larger systems. https://lnkd.in/ejH3Z2gQ #ML #AI
(Some) Open Frontiers with LLMs with Yoav Goldberg
https://www.youtube.com/
To view or add a comment, sign in
-
Let Hevi do the heavy lifting! 💪 Check out our latest demo where we show you how to ask the right questions to Hevi's AI chat feature to get pinpoint accurate answers. 🔍 What You'll Learn: How to ask general vs. specific questions Formatting queries for the most precise answers Creating question sets for ongoing use Demo video here: https://lnkd.in/gFAYj88t
Unlock Precise Answers from Hevi: The Power of Asking the Right Questions!
https://www.youtube.com/
To view or add a comment, sign in
-
Last month, PhD researcher Dyah Adila told Snorkel researchers about her work on ROBOSHOT, a novel approach to get better performance out of foundation models without fine-tuning. This work shows promise and could meaningfully impact how enterprise AI teams approach FM applications. Watch the video here: #airesearch #foundationmodels
ROBOSHOT: better foundation model performance without fine-tuning (Stanford researcher presentation)
https://www.youtube.com/
To view or add a comment, sign in
-
Was so good to sit down with Lord Tim Clement-Jones - lead on AI governance and regulation in the House of Lords - to chat about my book God-like (which he was so generous in praise for) and his excellent primer on responsible AI, Living with the Algorithm - which I definitely recommend. https://lnkd.in/ec8ghuc6
Lord Tim Clement Jones - Living with the Algorithm: Servant or Master? - Video
https://www.youtube.com/
To view or add a comment, sign in
-
Worth a worth a read.
Making data fit for human use. Keeping it real (ontologically). Building tools for the future of infrastructure.
Worth a read.
To view or add a comment, sign in
-
I am sharing some screenshots from his lecture series. #artificialintelligence
AGI Speech by LeCun
link.medium.com
To view or add a comment, sign in
-
Does the debate between open-source and closed-source AI still matter today? In this episode of AI Rising, hosts Leslie D'Monte and Jayanth N Kolla explore this with Sayandeb Banerjee, CEO of MathCo. They discuss resistance to open-source AI, the rise of Scalable Language Models (SLMs) like Gemma 2, and how its portability sets it apart. Get insights on AI's real-world applications and what it means for businesses looking to scale with AI-driven solutions! 🎧 Listen Now : https://shorturl.at/aQyPs #HTSmartcast #AIRising #OpenSourceAI #ClosedSourceAI #ScalableAI #DataDriven #AIModels #BusinessIntelligence #MachineLearning
Gemma 2 and the Open-Source Debate: A Deep Dive with MathCo's Sayandeb Banerjee
https://spotify.com
To view or add a comment, sign in
-
Answer the following questions in Part 1 below in atleast 500 words altogether.
Answer the following questions in Part 1 below in atleast 500 words altogether.
https://academicfocustutors.blog
To view or add a comment, sign in
EVP @ Klick | Driving AI Adoption and Business Solutions
3moThe above comparison was looking quickly at GPT 4o vs o1. Claude Sonnet 3.5 does pretty well compared to o1. It’s faster and also input tokens are $3/million vs the $15/million for o1. That 5x price difference makes it important to think about the use case, accuracy improvement and cost vs benefit. I’ve been using Sonnet a lot more for coding assistance in tools like Cursor. Gemini has been much better for recently published material, APIs and long document(s) summarization. Gemini also tends to have more guarails on topics it won’t cover. My habit is pulling up GPT otherwise. There still isn’t “one ring to rule them all”, there will unlikely ever be one and that’s not a bad thing. It can feel overwhelming at times, it’s easy to question if we’re on the right path, making the right choices. These models are all extremely capable, for most use cases using any of them will be sufficient. Just take time periodically to try others and make decisions based on your needs, cost of switching and the benefit. You can also invest in building your infrastructure, including tracing and evaluation pipelines, to me multi LLM if you’re beyond the learning and prototyping phase.