### WEEK30 ###

We went from 0 to 410 subscribers in the first week—that’s truly humbling, and thank you for supporting this newsletter. The past week has been eventful, from model wars to endless talks of the AI bubble. Here is the top news from week 30.

OpenAI Unveils SearchGPT: is Google's dominance over?

Shots fired and Google is the target. OpenAI just announced SearchGPT, a new AI-powered search engine that could reshape how we find information online. But is this the Google killer we've been waiting for, or just another player in an increasingly crowded field??

Key points:

• SearchGPT uses GPT-4 models to organize and make sense of search results
• Currently in "prototype" stage with access limited to 10,000 test users
• Designed to be integrated directly into ChatGPT in the future

Why it matters: This move puts OpenAI in direct competition with Google and startups like Perplexity, potentially disrupting the search market. However, it also raises questions about AI's role in curating information and the potential for bias or misinformation.

DeepMind's AlphaProof: Cracking the Math Olympiad Code

While ChatGPT struggles to count the r's in "strawberry", Google DeepMind's AlphaProof is solving Olympic-level math problems. This AI system has just achieved a near-gold medal performance at the International Mathematical Olympiad (IMO). Is this the dawn of truly intelligent AI, or just a highly specialized calculator?

Key points:

• AlphaProof solved 4 out of 6 IMO problems, just one point shy of a gold medal
• It's a neurosymbolic system, combining neural networks with symbolic AI
• Unlike GPT-4 or Claude, AlphaProof can handle complex mathematical reasoning

Why it matters: This breakthrough demonstrates a significant leap in AI's problem-solving capabilities. While large language models like GPT-4 and Claude excel at language tasks, they often fumble with basic math. AlphaProof's success in tackling complex mathematical problems showcases a new approach to AI that could pave the way for systems capable of advanced reasoning and scientific discovery. It's not just about solving equations – it's about creating AI that can truly think.

Meta's Llama 3.1 405B: The Linux Moment for AI?

Meta dominated the news this week by releasing Llama 3.1 405B, a colossal open-source AI model with 405 billion parameters. But the real story isn't just its size - it's Meta's attempt to create the Linux of AI. Could this be the moment that democratizes advanced AI capabilities?

Key points:

• Llama 3.1 405B is Meta's largest "open" AI model to date
• It's competitive with leading proprietary models like GPT-4 and Claude 3.5 Sonnet
• Mark Zuckerberg compares it to Linux, aiming for it to become an industry standard

Why it matters: By releasing a state-of-the-art model as open-source, Meta is challenging the closed-source dominance of companies like OpenAI and Anthropic. This move could democratize access to advanced AI capabilities, potentially leveling the playing field for researchers and smaller companies. If Llama becomes the "Linux of AI" as Zuckerberg envisions, it could shift the AI landscape from a proprietary model to an open, collaborative one. However, it also raises questions about the responsible development and use of powerful AI technologies when they become widely accessible.

AI Agents Set to Collaborate: Capgemini's 2025 Prediction

Imagine a world where AI agents work together, communicating and solving complex tasks without human intervention. According to Capgemini, this sci-fi scenario could become a reality as soon as 2025. Are we on the brink of an AI revolution, or is this another over-optimistic forecast??

Key points:

• Prediction of "multi-agent AI" systems by 2025
• 82% of companies plan to integrate AI agents within 1-3 years
• Potential for AI agents to facilitate automation and handle repetitive tasks

Why it matters: If accurate, this prediction signals a fundamental shift in how work is done, potentially revolutionizing industries and redefining human roles in the workplace. However, it also raises concerns about job displacement; read more about this in my in-depth article: https://bit.ly/TheDeathoftheSpecialist

The AI Innovation Plateau: Is Business Data the Next Frontier?

The pace of AI innovation seems to be slowing down. Are we hitting the limits of what current AI can do, or is this just a temporary lull before the next big breakthrough? Some experts argue that the key to unlocking the next wave of AI advancements lies in harnessing business data.?

Key points:

• Current AI models are running out of good public training data
• Business data could provide higher quality, more specialized information
• Challenges include data privacy, security, and companies' willingness to share?

Why it matters: This potential shift could reshape the AI landscape, giving businesses a central role in advancing AI technology. It also highlights the growing importance of data as a strategic asset and the need for robust data governance frameworks.

AI Bubble Concerns: Is History Repeating Itself?

As AI stocks soar (Nvidia anyone?) and valuations reach insane heights, some analysts are drawing parallels to the dot-com bubble of the late 1990s. Are we witnessing the birth of a new economic paradigm, or are we on the brink of a spectacular market correction?

Key points:

• The S&P 500 has advanced 45% since January 2023, largely due to AI stocks
• Tech stocks are currently trading at 30 times forward earnings
• Experts suggest a market downturn is likely but may be less severe than the dot-com crash?

Why it matters: Understanding the potential for an AI bubble is crucial for investors, businesses, and policymakers. While AI undoubtedly has transformative potential, realistic expectations, and careful analysis are needed to navigate this rapidly evolving landscape.

EU AI Act Set to Reshape the AI Landscape

?Mark your calendars: The EU AI Act is set to come into force on August 1, 2024. This landmark legislation aims to create a harmonized legal framework for AI in the EU. But will it foster innovation or stifle progress?

?Key points:

• Phased implementation with full applicability from August 2, 2026
• Risk-based approach categorizing AI systems into different risk levels
• Hefty penalties for non-compliance, up to €35 million or 7% of global turnover

Why it matters: This regulation will have far-reaching implications for AI development and deployment in Europe and beyond. Companies operating in or selling to the EU market will need to ensure compliance, potentially reshaping global AI practices and standards.

I'm truly humbled by the support this newsletter has received in its first week. Reaching 410 subscribers is beyond what I could have imagined, and it's all thanks to you. As a new creator in this space, I'm learning and growing with each edition, and your engagement is what keeps this going.

Now, I'm setting an ambitious goal: to reach 1000 subscribers. But I can't do it alone – I need your help. If you've found value in this newsletter, here's how you can support this fledgling project:

? ?? Share this recap with friends or colleagues who might enjoy it - your recommendation means the world
? ?? Give me feedback - what do you like? What could be better? Your insights will help shape future editions
? ?? If you haven't already, consider subscribing - every new subscriber is a huge motivation

Your support, big or small, makes a real difference to a starting creator like me. Whether it's sharing the newsletter, providing feedback, or simply reading each week, you're helping to build something special.

Thank you for being part of this journey. I'm excited to see where we can take this together, and I'm grateful for every bit of support along the way.

See you next Sunday for another round of AI insights!

Leandro