Trusted Business Optimization & Automation (BOAT) partner in the GCC for AI and IoT-powered digital transformations for businesses.
Latest edition of One More Thing in AI Newsletter.
In this Edition
🤖👥 AI -Humanity: AI Paradox: Meta’s Chief Scientist Weighs in on the Existential Debate
🎵📜 Music: The Fine Line: AI, Lyrics, and Copyright Battles
🛠️🖨️ 3D Printing: The Blueprint of Tomorrow: AI Build’s Smart 3D Printing Software
💳🤖 Payments: Square’s AI Playbook: A New Chapter in Business Automation
🌍🤝 UAE: Bridging AI Horizons: OpenAI and G42’s Strategic Middle East Expansion
📊🤖 LLMs: The David vs. Goliath of AI: Why Small Language Models Could Be the Next Big Thing
🌟🛒 Sponsored: Yadget Data Generator
The AI Paradox
Meta’s Chief Scientist Weighs in on the Existential Debate
Picture a world where AI is as ubiquitous as smartphones, yet the debate rages on: friend or foe? Meta’s Chief AI Scientist, Yann LeCun, recently stepped into the ring to share his perspective.
🔍 The Existential Question:
Since the release of OpenAI’s ChatGPT-4, the AI conversation has shifted towards its potential risks. While some, like Geoffrey Hinton, have left their positions to warn about AI’s dangers, LeCun takes a different stance.
🏢 Big Tech and Regulation:
LeCun warns that premature regulation could stifle innovation and solidify Big Tech’s dominance. It’s like putting training wheels on a race car; you won’t get very far, very fast. This stance is ironic since LeCun comes from Meta, and you can argue that Meta is late to the AI party compared to Microsoft with its investment in Open AI. Regulation usually helps the leaders, not the laggards.
Regulatory Capture:
According to LeCun, the term “AI safety” is being used as a smokescreen for what he calls “regulatory capture.” It’s as if we’re putting a leash on AI research under the guise of safety, potentially holding back progress.
🤖 AI’s Future Role:
LeCun envisions a future where AI mediates our interaction with the digital world. Think of it as your digital butler, making your life easier. The debate on AI’s existential risk is far from settled.
While some see it as the proverbial wolf at the door, others, like LeCun, view it as a kitten that has yet to grow claws. One thing is clear: the conversation is just getting started, and it could shape the future of AI and its impact on our society.
Music
AI, Lyrics, and Copyright Battles
🎵 Imagine you’re humming along to the Rolling Stones’ “You Can’t Always Get What You Want,” and you realize you’ve forgotten the lyrics. You prompt your AI assistant to help, and voila! The lyrics are there. But what if that seemingly innocent act lands someone in a legal quagmire?
📜 The Legal Landscape: Universal Music Group and other music publishers have taken Anthropic, an AI company, to court. The bone of contention? Anthropic’s AI model, Claude 2, can generate lyrics almost identical to copyrighted songs. It’s not just about responding to specific song requests; even generic prompts can result in the AI spitting out copyrighted lyrics.
🔍 The Underlying Issue: It’s about the intellectual property that fuels the music industry. Many platforms pay to license lyrics, but Anthropic has been accused of sidestepping this crucial step. The company is also alleged to have used copyrighted material to train its language models, raising questions about the ethical boundaries of AI training.
🤖 AI Ethics and Investment: Anthropic isn’t a small player; it has significant investment backing, including $4 billion from Amazon and $300 million from Google. The company claims to take “trust and safety” seriously and even talks about “constitutional AI,” a framework for rule-based AI systems.
🔮 Looking Ahead: As generative AI becomes more sophisticated, the line between innovation and infringement is getting blurry. This case could set a precedent for how intellectual property laws adapt to the age of AI. It’s not just about the music industry; it’s a wake-up call for anyone leveraging AI in their business operations.
🎤 The Mic Drop: This case is a cautionary tale for those navigating the evolving landscape of AI and data. As this case unfolds, it’s clear that the next big hit in the AI world might be a gavel striking down.
The Blueprint of Tomorrow
Ai Build’s Smart 3D Printing Software
Imagine you’re an engineer, and you’ve just designed a complex structure. Now, you need to bring it to life. You could go the traditional manufacturing route or leverage 3D printing. But what if the 3D printing software could think, adapt, and optimize? Enter Ai Build, a London-based startup, just secured $8.5 million in Series A funding.
📈 The Investment Landscape: Nikon has jumped in as a strategic investor in Ai Build’s Series A round. IQ Capital led the round and saw participation from existing investors like Superseed and ACT Venture Partners. The additive manufacturing (AM) industry is pegged at $16 billion and is growing at over 20% per year. Ai Build aims to unlock this sector’s full potential.
🎨 From Architects to Innovators: The founders of Ai Build, Daghan Cam and Michail Desyllas, started their journey as architects, solving complex geometric problems. Their experience led them to identify a niche in optimizing 3D printing processes, which are often error-prone. They developed AiSync, a machine learning-based software that perfects the toolpath for depositing melted materials.
🤝 Partnerships and Expansion: Ai Build has joined forces with industry giants like Boeing and Rolls-Royce. They collaborate with about 25 hardware manufacturers and have diversified their material range to include plastics, metals, cement, and epoxies.
🗣️ Natural Language Commands: One of Ai Build’s most recent innovations is the ability to command the software using natural language. This feature is designed to lower the barrier to entry and make 3D printing accessible even to those with minimal engineering knowledge.
🎯 Takeaway: The Series A funding will be used primarily for software development and expansion into the United States. Ai Build aims to make 3D printing more efficient and redefine how we think about manufacturing itself.
Square’s AI Playbook
A New Chapter in Payments
🤖 Picture this: you’re running a bustling café, and you’re swamped with orders, inventory management, and customer queries. What if AI tools could take some of that load off your shoulders? The US-based financial services company Square is making this a reality with its new suite of generative AI features.
🛠️ The Toolbox Unveiled: Square has rolled out ten generative AI tools to automate operations and speed up workflows. These tools are integrated directly into Square’s existing business software. The goal? To act as an extra set of hands for businesses grappling with multifaceted demands.
🍽️ On the Menu: One of the standout features is a Menu Generator for restaurants. With a few clicks, businesses can create a full menu, saving valuable time when launching operations on Square’s platform.
📸 Picture Perfect: E-commerce sellers can now elevate their websites with AI-generated backgrounds and style prompts via the latest version of the Photo Studio app.
📧 Communication Overhaul: From generating personalized email copy to sending out team announcements, Square’s AI features are making communications more efficient and tailored for the business.
🌐 SEO and More: Square’s AI-assisted copy generator helps enhance websites and boost SEO, from setup to refresh and from headlines to blog posts.
🤝 Personal Touch: Advanced AI responses in Square Messages allow for personalized messaging to buyers, enhancing customer engagement.
📊 Data-Driven Decisions: Square’s new AI-powered starter library suggests items for businesses to adopt based on sales insights, making inventory management smarter.
🔄 Switching Made Easy: For those looking to switch booking platforms, Square Appointments now auto-imports service names, descriptions, durations, and prices during onboarding.
UAE’s AI Ambitions
OpenAI and G42’s Strategic Middle East Expansion
🌍 OpenAI, the entity behind the renowned ChatGPT, and UAE’s tech holding group G42 have recently partnered to expand AI capabilities in the Middle East region. G42 has the UAE government backing it through its Mubadala Investment arm.
🤝 The Partnership: OpenAI and G42 intend to deploy generative AI models across various sectors, notably financial services, energy, healthcare, and public services. The collaboration aims to simplify the integration of advanced AI capabilities for organizations in the United Arab Emirates and beyond.
🔗 Strategic Synergies: G42 has committed to leveraging its substantial AI infrastructure to support OpenAI’s regional operations, mainly through Microsoft Azure data centers. OpenAI’s CEO, Sam Altman, emphasized that G42’s industry connections would facilitate AI solutions finely tuned to the region’s unique characteristics.
🌐 Global Context: This partnership comes when AI developments are gaining traction globally. For instance, Saudi Arabia has recently announced a collaboration with Chinese universities to develop an Arabic-based AI system. These developments occur against increasing scrutiny by U.S. regulators over AI semiconductor chip exports, including to the Middle East.
📊 Regulatory Landscape: The U.S. has been ambivalent about AI exports to the Middle East. New rules have expanded licensing requirements for exporting advanced chips to countries under a U.S. arms embargo, which includes several Middle Eastern nations.
🎯 Takeaway: Sam Altman suggests that this partnership will benefit the Middle East and advance generative AI globally. These types of partnerships unlock new markets and offer a nuanced approach to AI adoption, considering regional specificities. In this particular deal, Microsoft, Open AI, and the UAE government win.
David vs. Goliath of AI
Why Small Language Models Could Be the Next Big Thing
Imagine you’re a chef in a kitchen. You could have all the fancy gadgets in the world, but sometimes, a simple knife and cutting board do the job more efficiently. In AI, particularly language models, bigger doesn’t always mean better.
🚀 Quick to Train: Time is of the essence in any project. Smaller language models (LLMs) are like quick recipes; they offer faster training, fine-tuning, and results. The shorter feedback loop means you can iterate and improve much faster.
💻 Runs Anywhere: Think of smaller LLMs as a Swiss Army knife. They can run in various environments, even locally, on your laptop, making them incredibly versatile.
🔧 No Special Hardware: Smaller LLMs are a breath of fresh air in a world where the latest and greatest chips are hard to come by (I wrote about this in an earlier newsletter). They rarely require specialized hardware for training or inference.
⚡ Low Latency: In a fast-paced world, speed is crucial. Smaller LLMs can generate completions faster and are well-suited for low-latency environments.
🌐 Edge Computing: Smaller LLMs are like pocket-sized travel guides; they’re perfect for running at the edge due to their low latency, smaller file size, and quick startup times.
🤝 Ensemble Power: Rumor has it that GPT-4 is an ensemble of eight smaller models. This strategy has been effective for years in machine learning, proving that teamwork often trumps individual prowess.
🤔 Bonus Points: Smaller LLMs might offer better interpretability and enhanced reproducibility. Understanding a 7-billion parameter model is likely easier than deciphering a 60-billion parameter behemoth.
🎯 Takeaway: In the AI arena, it might just be that the smaller contenders are the ones to watch. These LLMs will help create a level playing field for Small Businesses and Startups who want to use the power of AI in their solutions.
