​AI Tech Circle​

Hey Reader!

This weekend, my attention was drawn to the lecture from Yann LeCun, a Professor at NYU and Chief AI Scientist at Meta, titled ‘Objective-Driven AI: Towards AI systems that can learn, remember, reason, plan, have common sense, yet are steerable and safe’ and four design patterns for agents from Andrew Ng.

The holiday period has started, and there will be no newsletter next week. An advance Eid Mubarak to whoever is celebrating.

The first slide of 97 Slidedeck starts with the title ‘Machine Learning sucks! (compared to humans and animals); I just want to leave every one of you to spare an hour and go through this complete slide deck of a master class on Artificial Intelligence.

Slide from Objective-Driven AI – Yann LeCun

Second good read on four design patterns introduced by Andrew Ng for AI Agent Workflows:

• Reflection: The LLM examines its own work to come up with ways to improve it.
• Tool use: The LLM is given tools such as web search, code execution, or any other function to help it gather information, take action, or process data.
• Planning: The LLM comes up with, and executes, a multistep plan to achieve a goal (for example, writing an outline for an essay, then doing online research, then writing a draft, and so on).
• Multi-agent collaboration: More than one AI agent work together, splitting up tasks and discussing and debating ideas, to come up with better solutions than a single agent would.

Andrew Ng @AndrewYNg Last week, I described four design patterns for AI agentic workflows that I believe will drive significant progress this year: Reflection, Tool use, Planning and Multi-agent collaboration. Instead of having an LLM generate its final output directly, an agentic workflow prompts… https://twitter.com/i/web/status/1773393357022298617 8:54 PM • Mar 28, 2024 499 Retweets 2406 Likes Read 62 replies

​

Weekly News & Updates…

This week’s AI breakthroughs mark another leap forward in the tech revolution.

1. ​Jamba: AI21’s SSM-Transformer Model, enhances the Mamba Structured State Space model (SSM) technology with elements of the traditional Transformer architecture; Jamba compensates for the inherent limitations of a pure SSM model. Offering a 256K context window
2. ​Grok-1.5 is announced from X with improved reasoning capabilities and a context length of 128,000 tokens.
3. ​VoiceEngine from OpenAI: Navigating the Challenges and Opportunities of Synthetic Voices
4. ​Third-party testing as a key ingredient of AI policy: In this article, Anthropic explores the concept of third-party testing, including its importance and the research that led to advocating for this policy stance. Additionally, they examine how testing intertwines with broader AI policy issues, like the availability of open models and concerns about regulatory capture.
5. ​Lumiere is a space-time diffusion research model from Google Research. Using fine-tuned text-to-image model weights, Lumiere can generate videos in the target style from a single reference image.

The Cloud: the backbone of the AI revolution

• ​An AI application that can chat with any service
• ​Enhancing OCI Data Science: Unveiling the New Autoscaling Feature for Model Deployment

Favorite Tip Of The Week:

Here’s my favorite resource of the week.

• ​Enterprise Search and Retrieval Demystified: A Guide for RAG Users. Cohere has answered the search and retrieval questions for RAG use cases.
• ​A little guide to building Large Language Models in 2024, from the co-founder of HuggigFace

Potential of AI

• ​TensorRT-LLM running on NVIDIA H200 Tensor Core GPUs, the latest memory-enhanced Hopper GPUs, has shown the fastest performance running inference in MLPerf’s biggest test of generative AI. This benchmark was used on the giant version of Llama 2, packing 70 billion parameters. The model is more than 10 times larger than the GPT-J LLM first used in the earlier benchmarks.

Things to Know

• ​Third-party testing as a key ingredient of AI policy: In this article, Anthropic explores the concept of third-party testing, including its importance and the research that led to advocating for this policy stance. Additionally, they examine how testing intertwines with broader AI policy issues, like the availability of open models and concerns about regulatory capture.
• ​You Transformed the World,’ NVIDIA CEO Tells Researchers Behind Landmark AI Paper. Look at the people behind the research paper “Attention Is All You Need.”

The Opportunity…

Podcast:

• This week’s Open Tech Talks episode 126 is “Meet AI Teacher: The Future of AI in Education Unveiled with Dr Pauldy Otermans and Dev Aditya”

Apple | Spotify | Google Podcast | Youtube

Courses to attend:

• ​GitHub Foundation Certification Series: An introductory session to explore the basics of GitHub as well as foundational concepts on Git with hands-on examples on how to use both
• ​Introduction to Data-Centric AI: This class covers algorithms to find and fix common issues in ML data and to construct better datasets, concentrating on data used in supervised learning tasks like classification
• ​Machine Learning Engineering for Production (MLOps): a course from Andrew Ng on MLOPs
• ​JavaScript RAG Web Apps with LlamaIndex: Learn how to build a RAG application in JavaScript, and use an intelligent agent

Events:

• ​MEOUG MAJLIS 2024, Annual conference of Middle East Oracle User Group, April 18, Dubai, UAE
• ​GISEC, Global, 23-24 April, Dubai, UAE

Tech and Tools…

• ​Gradio: Build Machine Learning Web Apps in Python. It’s an open-source Python package that allows you to build a demo or web application for your machine-learning model
• ​Fuel provides your machine learning models with the data they need to learn. Interfaces to common datasets such as MNIST, CIFAR-10 (image datasets), Google’s One Billion Words (text), and many more

Data Sets…

• ​World Bank Open Data: Free and open access to global development data
• ​US Statistics & Data Sets for Health Care and Public Health

Other Technology News

Want to stay on the cutting edge?

Here’s what else is happening in Information Technology you should know about:

• ​OpenAI holds back public release of tech that can clone someone’s voice in 15 seconds due to safety concerns
• ​US Congress bans staff use of Microsoft’s AI Copilot, Axios reports.

Earlier Edition of a newsletter

That’s it!

As always, thanks for reading.

Hit reply and let me know what you found most helpful this week – I’d love to hear from you!

Until next week,

Kashif Manzoor

The opinions expressed here are solely my conjecture based on experience, practice, and observation. They do not represent the thoughts, intentions, plans, or strategies of my current or previous employers or their clients/customers. The objective of this newsletter is to share and learn with the community.