​AI Tech Circle​

Stay Ahead in AI with Weekly AI Roundup

Read and listen on AITechCircle.com​

Welcome to the weekly AI Newsletter, where I provide actionable ideas and tips to assist you in your job and business.

Today at a Glance:

• Key takeaways from the article of McKinsey, Implementing Generative AI with Speed and Safety
• AI Weekly news and updates
• Generative AI Usecase of the week for you to explore
• Favorite Tip Of The Week & Potential of AI
• Things to Know: a framework for Large Language Model evaluations from the UK AI Safety Institute
• Open Tech Talk Podcast, the latest episode on AI & the Art of advertising
• AI/ML/GenAI courses to attend
• AI Events
• Tech Tools to try out
• Datasets for your next project
• Other Tech news

​McKinsey research has reported that Generative AI could contribute as much as $4.4 trillion to the global economy and boost the overall effectiveness of all AI technologies by 15 to 40 percent.

As organizations prioritize how quickly to implement Generative AI use cases in an enterprise, the McKinsey article "Implementing Generative AI with Speed and Safety" targets this area and presents very actionable tips to avoid risks.

​

A few of my key takeaways:

To manage essential inbound Gen AI risks by the organizations, follow these 4 steps:

Further, these risks are categorized into 8 main areas, and every organization has to develop its framework to understand and manage these risks based on the implementation of different use cases.

8 Risk Categories:

1. Impaired fairness - algorithmic bias, etc
2. IP Infringement - on copyrighted material
3. Data privacy and quality - unauthorized use or model training on inaccurate data
4. Malicious use - harmful, deep fake, hate speech, etc
5. Security threats - Vulnerabilities in Gen AI systems
6. Performance and explainability - model output & inaccuracies explanations
7. Strategic - non-compliance to regulations, societal and reputational risks
8. Third-party - linked with the use of third-party AI tools

​

Source of the above & read the complete article over here​

Weekly News & Updates...

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

1. ​RadOnc-GPT: Mayo Clinic has introduced RadOnc-GPT, which uses Meta Llama 2 LLM. This model has the potential to greatly improve the accuracy, speed, and quality of radiation therapy decision-making, which can benefit both healthcare providers and their patients. The model was trained on a large dataset of radiation oncology patient records from the Mayo Clinic in Arizona, and no patient data was shared outside of a secure network. The model was trained locally using Llama 2 and a local GPU server.
2. ​BLIP3 is a series of large multimodal models (LMMS) developed by Salesforce Al Research. BLIP3 is a new SOTA model under 5B on few-shot learning and multimodal benchmarks.
3. ​AlphaFold 3 predicts the structure and interactions of all of life’s molecules. It's a new AI model developed by Google DeepMind and Isomorphic Labs. it has accurately predicted the structure of proteins, DNA, RNA, ligands, and more, and how they interact.

The Cloud: the backbone of the AI revolution

• ​Oracle Code Assist, an AI code companion, helps developers increase velocity and enhance code consistency. It is based on the large language models (LLMs) running on OCI and optimized for Java, SQL, and application development on OCI, Oracle Code Assist is planned to provide developers with context-specific suggestions that can be tailored to an organization’s best practices and codebases.
• ​Establishing an AI/ML center of excellence. CoE plays an integral role by partnering with business units and end-users to identify AI/ML use cases aligned to business strategy, implement a company-wide vision, and deploy an AI/ML platform on appropriate hardware and software.

Gen AI Use Case of the Week:

Generative AI use cases in the supply chain industry:

Evaluation of proposals submitted by the suppliers in response to RFP with LLMs.

Areas covered are:

1. Business Challenges
2. AI Solution Description
3. Expected Impact/Business Outcome
4. Required Data Sources
5. Strategic Fit and Impact Rating
6. Practical Example of Implementation

Kashif Manzoor @kashifmanzoor One of the critical Generative AI use cases in the supply chain industry Evaluation of proposal submitted by the suppliers in response to RFP with LLMs Areas covered are: 1 - Business Challenges 2 - AI Solution Description 3 - Expected Impact/Business Outcome 4 - Required Data… https://twitter.com/i/web/status/1789568699461960144 12:10 PM • May 12, 2024 0 Retweets 0 Likes

​

Favorite Tip Of The Week:

Here's my favorite resource of the week.

• ​Trustworthy Retrieval-Augmented Generation with the Trustworthy Language Model. This tutorial goes in-depth to showcase how you can replace the Generator LLM in any RAG system with Cleanlab’s Trustworthy Language Model (TLM) to score the trustworthiness of answers and improve overall reliability.

Potential of AI

• Watch this video of Marc Andreessen hosting Ben Horowitz. It covers various topics, including: Will the "God models" get 100x better? Challenges for AI applications? Does Internet data represent average human activity? Lessons learned from the Internet era, etc.

Things to Know

Inspect, a framework for Large Language Model evaluations created by the UK AI Safety Institute, has been publicly released as open-source.

​Inspect is a software library that allows testers, ranging from start-ups and academia to AI developers and international governments, to evaluate the specific capabilities of individual models. Based on the results, Inspect generates a score. This library can assess models in various areas, such as their core knowledge, reasoning abilities, and autonomous capabilities.

The Opportunity...

Podcast:

• This week's Open Tech Talks episode 134 is "AI and the Art of Advertising with Robert Brill." He’s the Chief Executive Officer of Brill Media and has been immersed in the advertising industry since 2003.

Apple | Spotify | Google Podcast | Youtube

Courses to attend:

• ​Stanford XCS224U: Natural Language Understanding, taught by Professor Christopher Potts, this course covers theoretical concepts from linguistics, natural language processing, and machine learning.
• ​LangChain for LLM Application Development from Deep Learning AI. This course enables you to apply LLMs to your proprietary data to build personal assistants and specialized chatbots
• ​Building Systems with the ChatGPT API from Deep Learning AI. This 1-hour course will assist you in building multi-step systems using large language models.

Events:

• ​Oracle’s Data and AI Forum, May 23, 2024, Theatre of Digital Art (TODA), Dubai, UAE
• ​AI World Congress 2024, May 30-31, London, UK
• ​The AI Retreat 2024, June 13, Dubai, UAE
• ​GITEX GLOBAL, Oct 14-18, 2024, Dubai, UAE

Tech and Tools...

• ​ScrapeGraphAI is a web scraping Python library that uses LLM and direct graph logic to create scraping pipelines for websites and local documents (XML, HTML, JSON, etc.)
• ​AnythingLLM: A full-stack application that enables you to turn any document, resource, or piece of content into a context that any LLM can use as a reference during chatting

Data Sets...

• ​Labeled Faces in the Wild is a public benchmark for face verification, also known as pair matching. This database of face photographs is designed for studying the problem of unconstrained face recognition. The data set contains more than 13,000 images of faces collected from the web. Maintained by the University of Massachusetts.
• ​Indoor Scene Recognition is a challenging open problem in high-level vision. The database contains 67 Indoor categories and a total of 15620 images. Maintained by the MIT.

Other Technology News

Want to stay on the cutting edge?

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

• An excellent step from TikTok is to start marking the content generated with AI. They will use the metadata to content attached by the Content Credentials as digital watermarking. This feature is from the Coalition for Content Provenance and Authenticity. This metadata will enable us to recognize and label the AI-generated content. Reported by CNBC.
• Apple plans to use M2 Ultra chips in the cloud for AI, as reported by The Verge. They will offload the complex queries to M2 Ultra on cloud servers.

AI First Community to Learn & Share...

Have a question or need some assistance with your AI project, or maybe you want to be part of the thriving community to learn AI together,

​Click here to join the AI Tech Circle - It's your Community on Discord​

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