From Moore's Law to Scaling Law - Why your company must adapt now

Hi Futurist,

You are actually receiving this Dip one day late. I know. I had finished the new Dip. But then Anthropic decided to launch a new version of Claude. One that, according to them, is even better than GPT-4. I want to provide you with the best, most recent, and most comprehensive information from the past two weeks within marketing technology, AI, and Web3. Therefore, I am sending you insights, inspiration, and innovation straight to your inbox. Let’s dive into the depths of the digital future together and discover the waves of change shaping our industry.

In this post, we're dipping in:

• 📣 Byte-Sized Breakthroughs: Three (technological) developments in the digital landscape

• 🎙️ MarTech Maestros: one webinar, podcast, or interesting paper I believe you should watch, listen to, or read too

•  🧐 Last but not least: Things that I like

Do you have tips, feedback or ideas? Or just want to give your opinion? Feel free to share it at the bottom of this dip. That's what I'm looking for.

Prefer to listen rather than read?

Always on the move and no time to read? No problem! Now you can listen to our Digital Dip wherever and whenever it suits you. Happy listening. The link to the podcast is only available to subscribers. If you haven’t subscribe already, I recommend to do so.

📣 Byte-Sized Breakthroughs

'Byte-Sized Breakthroughs' briefly highlights (technological) developments in the digital landscape

Headstory: From Moore's Law to Scaling Law - Why your company must adapt now

In early May, OpenAI claimed the top spot as the best AI model with its latest release. Yet, a mere six weeks later, they’ve already ceded that position to Claude 3.5 Sonnet. This rapid turnover highlights a pivotal shift: we are no longer in the era of Moore’s Law, but firmly in the Scaling Law of AI models.

Claude 3.5 Sonnet, launched yesterday, raises the industry bar for intelligence. It outperforms competitor models like GPT-4o and Claude 3 Opus on a wide range of evaluations, combining speed and cost-effectiveness. Available for free on Claude.ai and the Claude iOS app, with higher rate limits for Pro and Team subscribers, it also integrates seamlessly via the Anthropic API, Amazon Bedrock, and Google Cloud’s Vertex AI. Its performance and cost metrics make it ideal for complex tasks such as context-sensitive customer support and multi-step workflows.

Claude 3.5 Sonnet excels in areas like graduate-level reasoning, coding proficiency, and visual reasoning, setting new industry benchmarks. Its ability to independently write, edit, and execute code with sophisticated reasoning and troubleshooting capabilities is particularly noteworthy.

For companies, this transition is monumental. It’s a clear signal to stop relying on a single AI model. Instead, they must develop solutions that are model-agnostic, anticipating the future trajectory of AI advancements. The pace at which AI models are evolving demands this flexibility.

Multimodal agents represent the next significant leap. The blistering speed of AI model improvements underscores this, pointing to a future where versatile AI agents will dominate. Companies must prepare their strategies accordingly, ensuring they remain at the forefront of this technological evolution. For CTOs and CEOs, the message is clear: adaptability is crucial. Embrace the Scaling Law, innovate continuously, and build a resilient, future-proof AI landscape.

To be well-prepared for the future of AI agents, organizations must:

1. Optimize data management processes: Streamline data workflows and ensure that data is managed efficiently and effectively.

2. Centralize all data: Ensure data consistency and accessibility by integrating all data sources into a single, unified system.

3. Enhance data quality: Implement robust data quality measures to ensure the accuracy and reliability of data for AI-driven insights.

Starting early is key. Organizations that wait until the future is upon them will find themselves outpaced by competitors who have already adapted. Ensuring robust data management now lays the foundation for leveraging advanced AI agents effectively. By acting proactively, companies can position themselves at the forefront of change and innovation, staying ahead of the competition.

Apple Intelligence: Apple's AI Revolution

TL;DR

Apple has officially entered the AI arena with Apple Intelligence, emphasising privacy and integration across native and third-party apps. Siri leads the interface, with the ability to incorporate OpenAI when needed. Developers can now integrate these AI features, with iOS 18 launching this summer and MacBooks following in the fall.

Read it yourself?

Sentiment

Apple’s AI announcement has been met with mixed reactions. Initially, the stock dipped slightly, but it rose again the next day, puzzling many. While some see it as a timely move, others feel it lacks innovation compared to competitors. However, the focus on privacy and personalized AI integration has garnered positive attention. Apple is clearly aiming to redefine AI interaction on its devices.

My thoughts

Apple has also renamed existing features from competitors, a classic move that Apple enthusiasts love. Despite the rebranding of existing features, this is classic Apple strategy. The biggest update, Siri's enhancement, will only be available in 2025 and initially in English. It's an exciting time as Apple layers AI across its OS, making interactions more intuitive and integrated than ever before.

NVIDIA released their own open-source model

TL;DR

NVIDIA introduces Nemotron-4 340B, a model family for generating synthetic data to train large language models (LLMs) across various industries. This open model pipeline includes base, instruct, and reward models optimized for NVIDIA NeMo and TensorRT-LLM, enhancing data quality and performance.

Read it yourself?

Sentiment

NVIDIA’s release of Nemotron-4 340B is a significant leap in the AI landscape, especially for industries requiring high-quality training data. The response has been positive, highlighting its potential to democratize access to robust LLM training datasets.

My thoughts

It all seems so obvious now. NVIDIA not only provides the hardware for training AI models but also addresses the scarcity of high-quality data. It's ingenious that NVIDIA has developed an AI model allowing other companies to generate synthetic data, which they can then use to train their own models on NVIDIA's equipment. Do you get it?

More byte-sized breakthroughs:

• Midjourney introduces model personalization for AI images
  Midjourney has launched a groundbreaking feature, model personalization, allowing users to create custom AI-generated images tailored to their specific needs. This innovation empowers creators to generate unique visuals by training the AI with their own data. With model personalization, Midjourney continues to push the boundaries of creative possibilities in the AI art space.

• China's open-source model challenges the US
  DeepSeek unveiled Coder V2, an open-source model designed for developers, supporting 338 programming languages. This model surpasses GPT-4 Turbo in coding and math tasks. By excelling in benchmarks and offering advanced capabilities, DeepSeek Coder V2 aims to disrupt the current AI landscape and highlight China's growing influence in the field.

• Luma Labs introduces Dream Machine
  Luma Labs' Dream Machine generates high-quality, realistic videos from text and images. This model produces 120 frames in just 120 seconds, delivering smooth, eventful shots with consistent characters and accurate physics. Users are thrilled with its performance, creating everything from hyper-realistic clips to entertaining memes, showcasing a tool that rivals the quality of OpenAI's Sora.

🎙️ MarTech Maestros

In 'MarTech Maestros,' I share one webinar, podcast, or interesting paper I've encountered, which I believe you should watch, listen to, or read too.

How AI is reshaping business strategies in 2024

Discover the transformative power of AI in today's business landscape. Our latest study reveals how top companies are leveraging AI to drive growth, streamline operations, and enhance customer experiences. From budget allocations to the most impactful tools, find out what sets successful businesses apart in the AI era.

 🧐 Last but not least: Things that I like

• Meta adds AI-powered features to WhatsApp Business app: Meta introduces AI tools for WhatsApp Business to create ads, generate responses, and enhance customer interactions, aiming to boost conversational commerce.

• Freepik introduces 'Freepik Designer': An easy-to-use online design tool featuring AI tools and a vast resource library, fulfilling user requests.

• Kling AI features high-resolution video generation: Kling AI generates high-resolution videos up to two minutes long, using advanced 3D space-time attention and diffusion transformers for realistic motion and interactions.

• NOX on technology and humanity: NOX emphasizes using technology to improve humanity by teaching love, intentionality, and achieving the impossible.

• DuckDuckGo announces AI Chat: DuckDuckGo has introduced a new feature called DuckDuckGo AI Chat. This service allows users to access popular AI chatbots privately.

• Adobe overhauls terms of service to address AI training concerns: Adobe is updating its terms of service to clarify that it will not use customers' work to train AI models.

• Elon Musk threatens to ban Apple devices over ChatGPT integration: Elon Musk threatens to ban Apple devices from his companies due to privacy concerns over Apple's integration of ChatGPT in iOS 18.

• Meet the Google TV Network: Google introduces a new TV network offering targeted, in-stream video ads across 125+ channels on Google TV and Android TV OS devices.

• BeReal acquired by Voodoo for €500M: Voodoo acquires BeReal for €500 million, aiming to leverage its expertise to scale the social network focused on authentic, unfiltered experiences.

• TikTok rolls out image search in TikTok shop: TikTok introduces image search capabilities in TikTok Shop, challenging Google's dominance in search and expanding its e-commerce efforts.

• Elon Musk drops lawsuit against OpenAI: Elon Musk has withdrawn his lawsuit against OpenAI and its co-founders, citing a lack of formal written agreement.

• Microsoft retires GPT Builder: Microsoft will retire GPT Builder on July 10, 2024, and delete all associated data by July 14, 2024, focusing on commercial and enterprise scenarios.

• AI bot runs for mayor in Wyoming: Victor Miller's campaign for mayor of Cheyenne, Wyoming, involves an AI bot, VIC, making decisions, raising legal and ethical questions.

• LinkedIn introduces AI enhancements for job seekers and businesses: LinkedIn launches AI-powered tools for job searches, personalized learning, and business growth, enhancing productivity and success for users.

• OpenAI and Microsoft team up with Oracle to boost ChatGPT's compute capacity: This partnership allows OpenAI to utilize the Microsoft Azure AI platform on Oracle’s infrastructure to enhance the compute capacity required to run ChatGPT.

• Spotify launches in-house creative agency: Spotify introduces Creative Lab to help brands create custom marketing campaigns and tests generative AI ads for advertisers.

• AI video regeneration that will save you hours: Make edits by typing, or choose a different language, and Descript will generate your AI voice and sync your lips. You’ll never re-record again.

• Microsoft delays Windows Recall feature: Microsoft postpones the Windows Recall feature for Copilot+ PCs due to security concerns, with a future update planned.

• Picsart partners with Getty Images for custom AI model: Picsart collaborates with Getty Images to develop a custom AI model for generating licensed, commercially usable images for its 150 million users.

• Brave integrates search results with Leo AI assistant: Brave integrates its search results with the Leo AI assistant, enhancing user experience with up-to-date information and privacy-focused features.

• Meta has paused its plans to train AI using data from European users: Meta had planned to use public content from Facebook and Instagram to train its AI, but faced obstacles due to GDPR regulations.

• Leonardo AI announces Phoenix model: A foundational model with exceptional prompt adherence and coherent text generation for banners, posters, and logos.

• YouTube is testing server-side ad injection to combat ad blockers: This method injects ads into videos before they reach the user's device, making it harder for ad blockers to detect and block them.

• Stable Diffusion 3 Medium released: Stability AI offers improved text-to-image generation with enhanced performance and resource efficiency, available for non-commercial use.

• Runway introduces Gen-3 Alpha: Runway's Gen-3 Alpha model generates highly detailed videos with complex scenes and cinematic choices.

• Google DeepMind unveils V2A technology: Google DeepMind's V2A technology adds sound to silent clips, matching scene acoustics and on-screen action.

• Microsoft Research launches AutoGen Studio: AutoGen Studio allows developers to build, test, and share multi-agent AI solutions with minimal coding using Microsoft's open-source AutoGen framework.

• TikTok introduces AI avatars for ads: TikTok is launching generative AI avatars and an AI dubbing tool to enhance branded content and ads on its platform.

• Apple suspends high-end headset work: Apple halts development on its next high-end Vision Pro headset, focusing on a cheaper model for late 2025 release.

• NVIDIA becomes world's largest firm amid AI rally: Nvidia's market capitalization reached $3.3 trillion, surpassing Microsoft and Apple, driven by the ongoing AI boom.

• Disney introduces advergames and shoppable ads on Hulu and ESPN: Disney launches interactive advergames and shoppable ads on Hulu and ESPN, enhancing viewer engagement and providing direct product links.

• Snap previews real-time image model for AR experiences: Snap previews a real-time image diffusion model for generating AR experiences, with new generative AI tools for AR creators launching soon.

• Former Snap engineer launches AI-human social network Butterflies: Butterflies, a new social network by a former Snap engineer, allows users to create AI personas that interact with humans and other AIs.

• Nokia unveils 3D spatial audio for 5G calls: Leveraging 5G technology, this innovation enhances voice clarity and provides an immersive audio experience during phone calls.

• Ilya Sutskever Founds Safe Superintelligence Inc.: Ilya Sutskever, co-founder of OpenAI, launches Safe Superintelligence Inc. to focus on the safe development of artificial super intelligence.

This was it. Our fourteenth digital dip together. It might seem like a lot, but remember; this wasn't even everything that happened in the past few weeks. This was just a fraction.

I can imagine it is difficult to keep track of which AI model is the best and for which use case. But also how your organization takes the first step. Let's talk about that, together. Hit reply!

Looking forward to what tomorrow brings! ▽

-Wesley