?? The AI & Data Digest: Edition 8 ??

Hello LinkedIn network!

?? After a short holiday break, we are back and ready to share a whole heap of new AI & data news with our readers. I'm excited to present the 8th edition of my newsletter, The AI & Data Digest. This edition is packed with the latest insights, news, and expert opinions on AI, data, and digital transformation.

Let’s dive in and stay ahead of the data & AI curve!

?? Top 10 News Beats:

1?? Anthropic Launches Claude 3.5 Sonnet: Anthropic has unveiled Claude 3.5 Sonnet, the newest addition to their Claude family. This model surpasses its predecessors in intelligence, speed, and cost-efficiency, making it ideal for complex tasks and customer support. Available on Claude.ai, the Claude iOS app, and via Anthropic API, Amazon Bedrock, and Google Cloud's Vertex AI, Claude 3.5 Sonnet includes a new feature called Artefacts for real-time content generation and editing. ??

2?? Apple and Meta Explore AI Partnership: According to the Wall Street Journal, Apple and Meta are discussing integrating Meta's generative AI model into Apple's AI system for iPhones. This potential collaboration could enhance Apple's AI capabilities across its apps, including Siri, while maintaining user privacy. ??

3?? Apple Delays AI Features in Europe: Apple has postponed the launch of three AI-powered features—Phone Mirroring, SharePlay Screen Sharing enhancements, and Apple Intelligence—in Europe due to EU competition rules. These features will debut in the US this autumn but won’t be available in Europe until 2025. ???

4?? OpenAI Acquires Rockset: OpenAI has acquired enterprise analytics startup Rockset to enhance its AI infrastructure. This acquisition aims to improve OpenAI’s data retrieval capabilities, transforming data into actionable intelligence for its products. ??

5?? Ilya Sutskever Starts New AI Company: Ilya Sutskever, co-founder and former chief scientist of OpenAI, has launched a new AI startup, Safe Superintelligence. The company focuses on developing safe superintelligent machines that surpass human intelligence. ??

6?? Cyber Attack Exposes Sensitive Patient Data: A significant cyber attack by the Qilin gang targeted multiple London based NHS hospitals, publishing nearly 400GB of sensitive patient data. This breach disrupted over 3,000 hospital and GP appointments, highlighting the critical need for robust cybersecurity measures in healthcare. ??

7?? Nvidia Becomes World’s Most Valuable Company: Nvidia’s market value has surged to $3.34 trillion, driven by its dominance in the AI chip market. This remarkable rise has positioned Nvidia ahead of tech giants like Apple and Microsoft. ??

8?? McDonald's Ends AI Drive-Through Tests: McDonald's has terminated its nearly three-year partnership with IBM to develop AI-powered drive-through ordering systems due to numerous customer complaints about inaccurate orders. The fast-food giant is exploring other AI collaborations despite the challenges. ??

9?? Meta Delays AI Rollout in Europe: Meta has postponed the launch of its AI offerings in Europe due to privacy concerns and regulatory barriers. The delay follows opposition from European authorities regarding Meta's new privacy policy for AI model training. ??

?? AWS Removes Data Transfer Fees for Cloud Migration: AWS has eliminated some data transfer fees, allowing customers to migrate data without paying egress fees. This change aligns with the European Data Act and enhances customer flexibility and compliance with new regulations. ??

?? Gain Actionable Insights:

This week, I explored the importance of Navigating the Risk Landscape of AI Systems. With AI's rapid adoption, understanding and managing associated risks is crucial. Our guide dives into the multi-layered risk landscape, defining AI system risks, and providing a risk assessment playbook to help organisations make informed decisions, improve compliance, and build trust.

Additionally, I delved into LLMOps on AWS, highlighting how to implement Large Language Model Operations using Amazon Bedrock. This article covers the principles of LLMOps, the key components of Bedrock, and best practices for deploying and managing large language models at scale. This was an important link to the change in track around our AI success stories section for this week's edition:?

?? Explore AI Failure Stories - A Stark Reminder on AI Safety:

This week, we are changing track to discuss AI Case Studies Gone Wrong. These stories highlight the importance of human oversight and continuous monitoring in AI systems:

1. Air Canada’s Virtual Assistant Error: After a virtual assistant provided incorrect information regarding bereavement fares, Air Canada was ordered to pay damages. The impact was significant, leading to a tribunal ruling and emphasising the need for accurate AI systems.

2. McDonald's AI Drive-Through Mishaps: McDonald's ended its AI drive-through tests due to numerous customer complaints about inaccurate orders, highlighting the challenges in implementing AI in customer service.

3. Sports Illustrated’s AI-Generated Authors: The publication faced backlash for using AI-generated authors and articles, demonstrating the potential risks of deploying AI in content creation without proper oversight.

4. Uber Eats AI-Generated Food Images: Uber Eats used AI-generated images for menu items, leading to customer confusion and mistrust when the actual food did not match the images.

5. Samsung Employees Leaking Confidential Data to ChatGPT: Samsung engineers accidentally leaked confidential information by inputting it into ChatGPT, highlighting the need for stringent data handling policies and training.

These cases underscore the necessity for robust MLOps/LLMOps practices to control AI system development and capture errors effectively. As well as common sense and ethical assessments about how customers and consumers of AI enabled services interact with content and material. Just because we can, doesn’t mean we should after all!?

?? Guest Blog:

Our guest blog this week is from Sokratis Kartakis and Heiko Hotz. The blog discusses the operationalisation of generative AI applications using MLOps, LLMOps and FMOps principles on AWS and is a really detailed and interesting read covering people, process & technology considerations. Key topics include:

- Model Optionality: Ensuring the right model is used for the right problem.

- Data Privacy: Protecting sensitive data within AI systems.

- Scale & Resilience: Building scalable and resilient AI systems.

- Monitoring and Optimisation: Continuously tracking performance metrics to maintain high-performing models.

These considerations are crucial for organisations looking to leverage AI responsibly and effectively.

?? AI Research Paper:

This week's featured research paper is "How can neural networks learn the rich internal representations required for difficult tasks such as recognising objects or understanding language?" by Yoshua Bengio, Yann LeCun, and Geoffrey Hinton. The paper explores the mechanisms through which neural networks develop complex internal representations, providing key insights into advancing neural network capabilities and improving AI model performance. Organisations interested in these concepts can gain valuable knowledge to enhance their AI strategies.

?AI Risk & Ethics:

With political agendas gearing up in the UK, it's been a surprisingly quieter period than usual on the AI regulation front. This week, we just have two relevant updates to share:

?? Norm Secures $25m Investment: Norm AI has raised $25 million to expand its AI-powered regulatory compliance platform. This funding will help improve efficiency, accuracy, and effectiveness in compliance management, addressing the challenges of different regulatory landscapes. Given the heightened regulatory overheads for organisations, this is a domain that will likely see further investment in the years ahead.

?? US Tightens Controls on Foreign Investments: The US has proposed new regulations to restrict outbound investments in critical technologies like AI, quantum computing, and semiconductors to protect national security. This positioning by the Biden administration would seem to ensure investment in US AI services and systems and curtail investment in competitive nations. The recent appointment of Retired U.S. Army General Paul M. Nakasone to OpenAI's Board of Directors underscores the significance of cybersecurity and AI safety, reflecting a strategic move to safeguard AI advancements within the US.

??? Upcoming AI Events:

July seems to be heating up in terms of AI events… There’s now even an AI Beach Party! ???

?? AI in Financial Services: Panel Event: London - 11th July

?? AI Summer Beach Party - 2nd July:?

?? Is Your AI Good Enough to Save Humanity - 3rd July?

There are a ton more events on the AI and data front this month and we will share more with you next week.

?? Jobs Spotlight:

This week we have 5 hot new Data & AI jobs from our friends over at Otta. Check them out below.

?? Data Scientist Manager, Mastercard (London): Focusing on AI and data-driven solutions. ??

?? Machine Learning Engineer, Prevail Partners (London): Junior, Mid, and Senior level positions available on a contract basis.

?? Analytics Engineering Lead, Monzo (London): Working on AI-driven banking solutions. ??

?? Senior Data Engineer, Checkatrade (London): Leveraging data and cloud technologies to develop innovative data products. ??

?? Applied Scientist, Amazon (London): Key accountabilities for designing and optimising some of Amazon retails key algorithms. . ??

?? Connect With Me!

Engage with me further on these topics and more by connecting on LinkedIn.

?? Like & Share

Enjoyed this newsletter? Hit like and share it within your network to help others stay on top of the latest in AI and data! Also, if you have any feedback on other bits to include, then let me know! Thanks as ever for reading!

Thanks for reading and looking forward to reconnecting again next week!

Ben

?? Please Like | ?? Please Share | ?? Please Comment