?? The AI & Data Digest: Edition 9 ??
Hello LinkedIn network! It’s a big weekend for sport in England with the Men’s football team set to take on their Spanish contemporaries in the final of the European Championships. I went a bit football crazy on Friday and have another story to share with you on Monday…without giving anything away… I’ve been doing a model bake-off. I’ll leave it at that for now! ?
?? To that point, I am pleased to share the 9th edition of my newsletter, The AI & Data Digest. With a brand-new logo! This edition is packed with the latest insights, news, and expert opinions on AI, data, and digital transformation. So, let’s get to it!
?? Top 10 News Beats:?
Graphcore Acquired by SoftBank??
?? The BBC is reporting a significant development: British AI chip designer Graphcore has been acquired by Japanese conglomerate SoftBank. This deal, although undisclosed in value, is speculated to be substantially lower than the ￡2bn valuation Graphcore achieved in 2020. The acquisition is viewed as a major endorsement of Graphcore's team, despite raising concerns about the UK's ability to nurture tech firms to compete globally in the booming AI chip market. SoftBank, which previously acquired another British chip designer Arm, plans to keep Graphcore headquartered in Bristol, aiming to drive growth and investment in the UK.
OpenAI's 'Strawberry' Project Targets Advanced Reasoning??
?? In an exclusive report by Reuters, OpenAI is working on a novel AI technology under the code name 'Strawberry.' This initiative aims to enhance AI's reasoning capabilities, enabling models to plan ahead and perform complex tasks autonomously. The project involves a specialised post-training phase designed to improve AI's ability to conduct deep research and navigate the internet reliably. While details remain under wraps, this development underscores OpenAI's commitment to advancing AI reasoning, a critical factor for achieving human-like intelligence.
Microsoft & Apple Drop Observer Role on OpenAI Board Amid Regulatory Scrutiny??
?? In response to regulatory concerns, Microsoft has relinquished its observer seat on OpenAI's board, a role that did not include voting rights. This move comes as competition regulators, including the UK's Competition and Markets Authority and the US Federal Trade Commission, scrutinise Microsoft's partnership with OpenAI. Additionally, Apple will no longer be able to appoint an executive to a similar position. OpenAI has announced a new strategy to inform and engage key partners through regular stakeholder meetings, reinforcing collaboration on safety and security.
AWS Unveils Five New Generative AI Innovations?
?? At the AWS Summit in New York, Amazon Web Services (AWS) announced five major innovations to democratise generative AI. Highlights include the launch of AWS App Studio, a low-code service enabling users to create secure applications swiftly, and Amazon Q Apps, which allows non-technical users to build AI-powered applications through simple prompts. Additionally, Amazon Bedrock has been enhanced to facilitate the customization of AI models, and new certifications are introduced to upskill individuals in cloud computing. AWS has also surpassed its goal of training 29 million people by 2025, reaching 31 million learners globally.
US Lawmakers Seek Intel Assessment on Microsoft's Investment in UAE's G42??
??? Reuters is reporting that Republican lawmakers are urging the Biden administration to conduct an intelligence assessment on Microsoft's $1.5 billion investment in UAE-based AI firm G42 due to concerns about the transfer of sensitive technology and G42's historical ties to China. Representative Michael McCaul and John Moolenaar requested a briefing on the deal's implications for U.S. national security. They highlighted the potential risks of exporting restricted AI semiconductor chips and model weights. Microsoft has stated that national security remains a priority, while the Chinese Embassy has criticised the U.S. for undermining international cooperation.
Google's Gemini AI Enhances Robotic Intelligence??
?? In a piece covered by the Verge, Google's DeepMind is using its Gemini AI to significantly enhance the intelligence and capabilities of its robots. By utilising the long context window of Gemini 1.5 Pro, the AI can process extensive information, enabling robots to navigate and complete tasks based on video tours of environments. The technology allows robots to understand and execute complex commands through natural language interactions. Preliminary tests showed a 90% success rate in a variety of tasks. This advancement paves the way for more sophisticated, context-aware robots that could assist in everyday environments.
Google Transitions DeepMind Focus from Research to AI Products?
?? This one is about a month old, but I still wanted to share the story. Reportedly, Google is shifting its DeepMind unit from primarily research to developing commercial AI products. This move has reportedly caused frustration among employees, who are now required to adhere to strict guidelines. The consolidation of Google's AI teams under the DeepMind banner aims to leverage the company's vast AI expertise to enhance product development. However, challenges with the Gemini AI model and the unique demands of generative AI are creating a learning curve for both research and product teams. This transition underscores Google's commitment to advancing AI while balancing innovation with practical applications.
AI in K-Pop: Innovation or Intrusion???
?? The BBC reports that the K-pop industry is increasingly incorporating AI in music production, sparking mixed reactions among fans. Groups like Seventeen and Aespa are experimenting with AI-generated music videos and lyrics. While some fans appreciate AI's ability to aid creativity, others fear it may erode the personal connection with artists. Industry insiders see AI as a tool for innovation but acknowledge that it cannot replicate the unique emotional input of human songwriters. As AI's role in K-pop evolves, the balance between technological advancement and artistic authenticity remains a critical discussion point.
AT&T Hit by Massive Data Breach??
?? Hackers have stolen call and text records from nearly all 109 million AT&T Wireless customers, as disclosed by the company. The breach, which occurred in April, involved data from May to October 2022 and included call and text records but not the content. AT&T stated that the stolen data could be used to identify workplaces, home locations, and personal contacts. The breach follows a previous incident where data from 73 million customers was compromised in 2019. AT&T is notifying affected customers and advising caution regarding emails or texts requesting personal information.
Social Media Reports Circulating that Disney Has Been Hit As Well??
?? Social Media is rife with reports that Disney has reportedly been hacked by a group named "Nullbulge," which claims to have accessed 1.1TB of data, including information on unreleased projects, concept art, and various login details. The hackers allegedly gained access through Disney's internal Slack. As of now, Disney has not commented on the breach, and none of the stolen content has been released online. This incident could potentially reveal significant information about future offerings from Marvel Studios and Lucasfilm. The situation draws parallels to the 2022 Rockstar Games breach, which compromised development data for Grand Theft Auto VI.
Treat this with a pinch of salt until more accurate reports are covered.
?? Gain Actionable Insights:
This week, I had to have a bit of fun with the England Men’s team heading to the final of the European Championships. You can read more about what Claude 3.5 Sonnet had to say in terms of its predictions about the winners of tonight's big match here!?
Other things I shared some thoughts on this week:?
I am also hosting a LinkedIn Live event at 2pm on the 13th of August where I will discuss:
?? ???? ???????????? ???? ???????????????????? ???? ???????????????? ???? ?????? ???????????????????? ??
Feel free to join me for an exclusive LinkedIn Live session where I will delve into the true reality of generative AI adoption in enterprises over the past year. This event will highlight key lessons learned, success stories from the frontline and insights into the future of generative AI adoption.
?? ?????? ?????????????
?? Discover practical insights from generative AI adoption in real-world enterprise scenarios.
?? Understand what works and what doesn't when it comes to your organisations generative AI journey.
领英推荐
?? Hear first hand experiences to understand what capabilities your business needs to invest in to scale generative AI in a safe and ethical way.
?? Hear how organisations are looking at generative AI to make money, save money and reduce risk in their businesses.
?? Date: Tuesday 13th August 2024
? Time: 2:00-2:45pm
?? Platform: LinkedIn Live
You can register here for the event.
?? Explore Captivating AI Stories:
AI Outperforms Clinical Tests in Predicting Alzheimer's Progression ??
Cambridge scientists have developed an AI tool capable of predicting whether individuals with early signs of dementia will progress to Alzheimer's disease in four out of five cases. This tool is significantly more sensitive than current methods, relying on non-invasive, low-cost data from cognitive tests and MRI scans.?
This AI model was tested on over 1,900 individuals from the US, UK, and Singapore, demonstrating its ability to accurately distinguish between those who would remain stable and those who would develop Alzheimer's within three years. The algorithm showed an accuracy rate of 82% in identifying those who would develop Alzheimer's and 81% for those who wouldn't, outperforming standard clinical markers and diagnoses.
The model's accuracy in predicting disease progression can help improve patient outcomes by enabling early interventions. It can also stratify patients into groups based on how quickly their condition is expected to deteriorate, which can guide the appropriate clinical pathways for those with symptoms due to other causes like anxiety or depression.
This AI tool was developed by a team led by Professor Zoe Kourtzi from the University of Cambridge's Department of Psychology and is expected to significantly reduce the need for invasive and costly diagnostic tests, ultimately improving patient care. The team plans to expand the model to other forms of dementia and incorporate additional types of data, such as blood test markers.
This research was funded by Wellcome, the Royal Society, Alzheimer’s Research UK, and several other prestigious institutions. The findings are published in eClinical Medicine.
For more details, you can read the full research paper by Lee, LY & Vaghari, D et al., titled "Robust and interpretable AI-guided marker for early dementia prediction in real-world clinical settings" in eClinical Medicine.
?? Guest Blog:
Given the big announcements at AWS Summit New York this week, I have opted to cover a piece by Swami Sivasubramanian, VP for AI & Data at Amazon Web Services from the AWS Machine Learning Blog. In the piece, Swami covers several new capabilities aimed at helping customers rapidly build, customise, and deploy generative AI applications securely. These enhancements are part of AWS's growing generative AI stack, which includes Amazon Q for AI-powered applications, Amazon Bedrock for building and scaling AI apps, and AWS's resilient infrastructure layer with purpose-built AI chips.
Amazon Q, is an advanced generative AI assistant, and it now includes features like Amazon Q Apps, enabling employees to create AI-powered applications through simple natural language prompts. This tool helps automate tasks and improve productivity across various industries. Additionally, AWS App Studio (preview) allows technical professionals to develop and manage enterprise applications quickly using natural language descriptions.
AWS also introduced new features for Amazon Bedrock, enhancing its capabilities for customising AI applications with business-specific data. This includes fine-tuning models like Anthropic’s Claude 3 Haiku and leveraging data sources for Retrieval Augmented Generation (RAG). Amazon Bedrock now supports vector search to enhance data retrieval speeds and accuracy.
To address concerns about AI-generated misinformation, AWS has added contextual grounding checks to its Guardrails for Amazon Bedrock, improving the accuracy and reliability of AI responses.
Furthermore, the piece goes on to explain that AWS continues to expand its partner network, with new collaborations like Scale AI for model customization and evaluation. AWS is really pushing hard to democratise generative AI, making it accessible and beneficial for a wide range of businesses and employees, fostering innovation and efficiency across industries.
?????AI Risk & Ethics:
??EU's AI Act Published with Transition Periods Announced
The EU's AI Act has been officially published in the Official Journal of the European Union, marking the conclusion of extensive negotiations since its proposal in 2021. This regulation establishes harmonised rules for AI across the EU, with a risk-based approach and phased implementation. Key deadlines include the applicability of prohibited practices from February 2025 and general purpose AI model regulations from August 2025. The AI Act aims to ensure responsible AI development, providing legal clarity and guidelines to support compliance and enforcement across the 27 member states.
It feels like we have been talking about the EU AI Act for a really long time now. However, it seems that it is set to come to a compliance and regulatory team near you immediately! For those who aren’t familiar with the text, key highlights and milestones for the act include:?
The AI Act will enter into force on August 2, 2024, under ~16 days from today. The timeline for implementation is as follows:
?? February 2, 2025: Prohibited AI practices must be withdrawn from the market.
??May 2, 2025: Codes of practice will be ready.
?? August 2, 2025: General purpose AI (GPAI) models must be in compliance. The governance structure (AI Office, European Artificial Intelligence Board, national market surveillance authorities, etc.) must be in place.
?? February 2, 2026: The European Commission will adopt the Implementing Act, which lays down detailed provisions establishing a template for the post-market monitoring plan and the list of elements to be included in the plan.
??August 2, 2026: All rules of the AI Act become applicable, including obligations for high-risk systems defined in Annex III (list of high-risk use cases). Member States must ensure that their competent authorities have established at least one operational AI regulatory sandbox at the national level.
??August 2, 2027: Obligations for high-risk systems defined in Annex I (list of EU harmonisation legislation) apply.
??? Upcoming AI Events:
There are some new and interesting AI & data related events to take in across London this month:?
??Responsible AI & Music Workshop: University of Arts -? London - 11th July ??
??AI & Tech Networking: London, Leicester Square - 23rd July ??
??Data Engineers Meetup: Primrose Street - 17th July ????
??Databricks Meetup: IDEALondon - 18th July ??
?? Jobs Spotlight:
This week we have 5 hot new Data & AI jobs from our friends over at Otta. Check them out below.
?? Senior Analytics Engineer, Monzo (London): Working on AI-driven banking solutions. ?? Apply here
?? Senior Machine Learning Engineer, OVO (London): Powering human progress with clean affordable energy. ?? Apply here
?? Lead Data Scientist, Spotify (London): Leading commerce analytics and parental leave cover for 13 months. ?? Apply here
?? Senior Data Scientist, Field (London): Renewable energy infrastructure, investment, and trading. ?? Apply here
?? Analytics Engineer, Starling Bank (London): Building AI-driven financial services. ?? Apply here
Don't miss out on these exciting opportunities! Apply now and take your career to the next level. ??
?? Thanks for reading!
Thanks as ever for checking in and reading this week's blog. As an FYI, I have opted to drop down the frequency of the newsletter to once a fortnight. I’ll still be sharing the 10 Big AI & Data Newsbeats weekly though! Whilst I will also be looking to host more LinkedIn live events in the months ahead to share insights with you all more frequently through a different channel!?
Stay connected for more updates and insights by following The AI & Data Digest. As ever, please Like, Comment, and Share to keep the conversation going!
Thanks,
Ben
?? Please Like | ?? Please Share | ?? Please Comment
Professional video editer at JMD Group
3 个月I looking for video editing work from home and i have 2 year experience of editing videos, shorts also My editing tools like Adobe Premiere Pro And After Effect Motion tracking With a normal color grading Davinci resolve or more If anyone need a video editor then connect me and I am from India here is my contact email is [email protected]