​AI Tech Circle​

How to find and prioritize an AI use case in your organization.

Stay Ahead in AI with Weekly AI Roundup; read and listen on AITechCircle: June 01, 2024

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

Before we start, share this week's updates with a friend or a colleague:

Today at a Glance:

• Selection of Pilot AI Use case in your organization
• Generative AI Usecase: Enhancing Customer Service in Healthcare with Large Language Models
• AI Weekly news and updates, Favorite Tip Of The Week & Things to Know
• Open Tech Talk Podcast, the latest episode on Unlocking Startup Success

---

One of the fundamental questions discussed in several conversations with colleagues and friends in the AI ecosystem is how to select AI and Gen AI use cases to headstart organizational efforts. With the current wave of AI globally and regionally, there is a mindset that I should not be left behind; therefore, I need to hunt some use cases that I can bring to my organization. This is vital, and the benefits that AI is bringing to businesses and on a personal level are beyond the hype. Today, I am discussing how we can try to make this whole process in a structured way rather than just running around to do something with AI.

​

Over the last few weeks, we have started the journey of how AI will be used in an organization. This was based on the notion that you have been assigned to lead or contribute to the efforts. Embedding AI and generative AI in an organization correctly is vital for success.

Let's go ahead and explore leading AI efforts within your organization.

First Step:

Build a 90-day plan for your AI initiative. This will help you put all your efforts behind this without going on an ad hoc basis to look at a few use cases and that's it. The formal plan will help you ride the wave of bringing AI into your organization more successfully and beneficially for your organization's business objectives.

So, you will execute the 90-day plan for the organization to introduce the first AI pilot project with tangible benefits aligned with solving business problems or bringing innovations to the business.

An earlier article, "Chief AI Officer’s Playbook: Executing the First 90 Days," provides a more in-depth coverage of this area.

Second Step:

Start building the AI strategy, which is primarily for your organization and will bring value rather than just following the crowd or market hype. I know the strategy is sometimes considered bookish, and people argue that let's get practical and go for the use case. However, running behind only the use case will not give you the exceptional benefits that AI and Gen AI bring to the business.

Therefore, treat this step as important as implementing an AI use case in your organization.

This topic is covered in detail in the earlier. "Lost in AI use cases, don't forget to put your AI Strategy first and align it with business."​

Third Step:

To identify quick wins to execute in alignment with the 90-day plan and ensure your strategy is in place, which AI early projects can you take up to showcase the benefits to the business? This will give you a head start on preparing for the rest of your AI strategy implementation in the organization.

It is essential to select a use case that is aligned with the business objective and contributes to factors such as revenue, user experience, operations, process, and cost.

Below is the table to summarize the example use case of AI-powered fraud detection covering the above areas.

By following the above three steps, you have put the AI initiative in motion at your organization and positioned it to play an integral role in achieving the business objectives. This can also be enhanced, so please share your comments and feedback.

Let's lay the groundwork, and we can all benefit from it as the AI community.

---

Weekly News & Updates...

Last week's AI breakthroughs marked another leap forward in the tech revolution.

1. How does the Voice engine work, and what is the safety of OpenAI? The model text-to-speech (TTS) gets human-life audio from text. OpenAI gives more insight into how they are working on the safety aspects while working on these models. Link​
2. ​PwC has partnered with OpenAI to provide access to all employees with ChatGPT enterprise. In contrast, EY has developed its own LLM, 'EY.ai EYQ.'
3. Meta has shared the update on the vision-language model (VLM), link​
4. Nomic-Embed-Vision is a high-quality, unified embedding space for image, text, and multimodal tasks. link​

The Cloud: the backbone of the AI revolution

• Serving LLMs using HuggingFace and Kubernetes on OCI. link​
• To tune or not to tune? A guide to leveraging your data with LLMs. This article covers different approaches to interacting with your data and the LLMs. link​
• IT trends show customers need computing power to take advantage of AI. link​

Gen AI Use Case of the Week:

Generative AI use cases in the HealthCare industry:

Enhancing Customer Service in Healthcare with Large Language Models

Business Challenges:

High Volume of Inquiries: Handling many customer questions about claims, insurance coverage, and plan details.

Inconsistent Information: Ensuring that customers receive accurate and consistent information across various channels

Resource Intensive: Requiring significant human resources to manage customer service operations efficiently

Response Time: Reducing the time taken to respond to customer inquiries to improve satisfaction.

AI Solution Description:

Using Large Language Models (LLMs) to automate responses to customer questions about the claims process, insurance coverage, and other plan details can streamline customer service operations.

Here’s how it works:

Data Ingestion: Collect and integrate data from customer inquiries, FAQs, claims databases, and insurance policy documents

Natural Language Understanding: The LLM processes customer queries in natural language, understanding the context and extracting key details.

Automated Responses: The model generates accurate and contextually relevant responses based on the ingested data, providing customers with the information they need

Continuous Learning: The system continuously learns from new inquiries and feedback to improve response accuracy and relevance over time

Expected Impact/Business Outcome:

Revenue: Lower operational costs by reducing the need for extensive human customer service teams. Improved customer satisfaction can lead to higher retention rates and reduced churn.

User Experience: Faster response times enhance customer satisfaction and trust, and providing consistent and accurate information reduces customer frustration

Operations: Automating responses frees up human agents to handle more complex inquiries. Easily scales to handle increasing volumes of customer inquiries without additional human resources.

Process: Ensures standardized responses, improving the overall quality of customer service 24/7. Provides around-the-clock customer support, enhancing accessibility and convenience

Cost: Significant reduction in the costs associated with staffing and training customer service teams.

Required Data Sources:

• Historical and real-time data on customer questions and inquiries.
• Comprehensive FAQs and knowledge base articles.
• Detailed claims process information and insurance policy documents.
• Data on customer feedback and satisfaction levels to fine-tune responses

Strategic Fit and Impact Rating:

Strategic Fit: High

Aligns with strategic goals of improving customer service efficiency, reducing costs, and enhancing customer satisfaction

Impact Rating: High

Significant positive impact on revenue, user experience, and operational efficiency, making it a precious investment for healthcare organizations

Favorite Tip Of The Week:

Here's my favorite resource of the week.

• OpenAI has released the architecture for AI-trained supercomputers. It utilizes cloud infrastructure, Kubernetes for orchestration, Azure Entra ID, key management, IAM, CI/CD security, etc. Read the full details.

Potential of AI

• COMPUTEX 2024: Watch the keynote delivered by NVIDIA founder and CEO Jensen Huang as he shares how the era of AI is driving a new industrial revolution across the globe. Link​

Things to Know

Anthropic has released details of how they introduced 'Policy Vulnerability Testing' to counter Elections-related Risks.

How will models respond to questions related to elections? Here are 4 steps process:

• Policy Vulnerability Testing (PVT)
• Automated Evaluations
• Implement Mitigation Strategies
• Retest to Measure the Efficacy

It covers 3 stages:

Planning: Collaboratively select policy areas and potential misuse applications to test. Elections-related PVT could include election administration, political parity, targeting voters, or creating disinformation.

Testing: Experts create and test prompts on models, starting with non-adversarial questions and progressing to adversarial attempts. Partners document model outputs and compare them against policies, including industry benchmarking.

Reviewing Results: After each testing round, meet with partners to discuss findings, identify policy gaps, and prioritize mitigation areas. Collaborative sessions ensure actionable test results.

---

The Opportunity...

Podcast:

• This week's Open Tech Talks episode 137 is "Unlocking Startup Success: Insights from Leopold van Oosten." The CEO of Amsterdam Standard has built 18 startups in 20 years and has won 3 significant awards through his organizations.

Apple | Spotify | Google Podcast | Youtube

Courses to attend:

• ​Advanced Natural Language Processing (Fall 2020): Lectures from the Fall 2020 offering of CS 685 (advanced natural language processing) at UMass Amherst
• ​AI Agents in LangGraph: You will learn to build an agent from scratch using Python and an LLM and then rebuild it using LangGraph, learning about its components and how to combine them to build flow-based applications.

Events:

• ​The AI Retreat 2024, June 11, Dubai, UAE
• ​GITEX GLOBAL, Oct 14-18, 2024, Dubai, UAE

Tech and Tools...

• ​Transformers.js enables you to Run machine-learning Transformers directly in your browser. It is designed to be functionally equivalent to Hugging Face's transformers Python library, meaning you can run the same pre-trained models using a similar API.
• ​Mesop: Helps you to build web apps in Python.

Data Sets...

• ​KITTI - Karlsruhe Institute of Technology and Toyota Technological Institute Dataset: This dataset is captured by driving around the mid-size city of Karlsruhe, in rural areas, and on highways. Up to 15 cars and 30 pedestrians are visible per image.
• ​GLUE: The general Language Understanding Evaluation (GLUE) benchmark is a collection of resources for training, evaluating, and analyzing natural language understanding systems.

Other Technology News

Want to stay on the cutting edge?

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

• From Gen AI 1.5 to 2.0: Moving from RAG to agent systems, VentureBeat has summarized the AI boom into three categories. Gen AI 1.0 with the launch of Large Language Models. Gen AI 1.5 was dominated by the discussion and architecture patterns for Retrieval Augmentation Generation, embedding models, and vector databases. Now Gen AI 2.0 is targeting Agents
• Meta adds AI-powered features to WhatsApp Business ap,p as reported by TechCrunch​

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​

Download 100+ Gen AI use cases:

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

---