Welcome back, AI’ers!

Here’s What’s On The Menu For Today

• OpenAI’s GPT-5 math ‘breakthrough’ blunder

• Andrej Karpathy’s reality check on AI agents

• Real-world subtitles arrive for smart glasses

• Why rivals from Google and Meta mocked OpenAI

AI ON MY MIND

A MESSAGE FROM OUR PARTNER

Master ChatGPT for Work Success

ChatGPT is revolutionizing how we work, but most people barely scratch the surface. Subscribe to Mindstream for free and unlock 5 essential resources including templates, workflows, and expert strategies for 2025. Whether you're writing emails, analyzing data, or streamlining tasks, this bundle shows you exactly how to save hours every week.

Subscribe for Your Free Bundle

TODAY’S TOP STORIES

OpenAI's Math 'Breakthrough' Blunder

The Recap: OpenAI researchers announced that GPT-5 had solved famously difficult math problems, only to be publicly corrected that the model had simply found existing solutions. The misstep quickly drew ridicule from competitors at Google DeepMind and Meta.

Unpacked:

• The excitement began when OpenAI's VP of Science, Kevin Weil, posted that GPT-5 solved 10 unsolved Erdős problems, a claim that was amplified by other researchers before the posts were deleted.

• Mathematician Thomas Bloom, who runs the official Erdős Problems database, quickly clarified that the problems were only listed as "unsolved" because he personally was unaware of a solution, not that they were truly unsolved by the math community.

• The blunder was highlighted by rivals, with Google DeepMind CEO Demis Hassabis calling the incident "embarrassing" and Meta's Yann LeCun mocking the OpenAI team for the mistake.

Bottom line: The incident underscores a key distinction: GPT-5 demonstrated an incredible ability for literature search, not novel mathematical discovery. It also serves as a stark reminder of the intense competition and public scrutiny that now surrounds every major AI development.

Karpathy's Agent Reality Check

The Recap: Esteemed AI researcher Andrej Karpathy delivered a sobering assessment of AI agents, projecting a decade-long timeline before they can deliver on their current hype and stating they 'just don't work' yet.

Unpacked:

• Karpathy attributes the current failures to fundamental gaps in AI, including insufficient intelligence, multimodal limitations, and a lack of continual learning.

• He also called reinforcement learning a 'terrible' and 'noisy' technique, suggesting it only seems effective because previous methods were much worse.

• The critique prompted a public challenge from Elon Musk on X, though Karpathy said he would rather collaborate with a model like Grok 5 than compete against it.

Bottom line: As one of the most respected researchers in the field, Karpathy's words serve as a major technical reality check on the 'Year of the AI Agent' hype. Still, systems that fail to impress a top mind may already offer massive productivity boosts for the other 99% of users.

MY FAVORITE AI TOOLS

Relevance is my favorite AI tool right now (even more so than N8N).

Here’s why:

It’s an AI agent building tool that doesn’t require you to build out the logic using nodes. Instead, you use natural language to build out these agents. But the best thing imo is that you can build AI Agent teams.

Like agents talking to each other completing a goal until its done. It’s nuts. No other tool on the market is doing this. You have to check it out.

(Try it out →Link)

👾 AI Tools We Use Everyday

➡️ N8N for AI Agents - Literally use this everyday to write this newsletter (link)

➡️ Axiom AI - AI Browser automation to scrape anything on the web (link)

➡️ GummySearch - AI Reddit research (link)

Tools are sourced from our database and are not sponsored or affiliated.

THE LEARNING CENTER

The step-by-step way you can become a highly paid AI Consultant. Get in touch to learn more.

MORE NEWS

Real-World Subtitles Arrive

The Recap: A new device called TranscribeGlass is bringing AI-powered transcriptions directly to smart glasses, creating a real-time subtitles experience for everyday conversations.

Unpacked:

• The device works by connecting to your phone to display real-time transcriptions of conversations directly in your field of view.

• Users can customize the text display to their preference, making it easier to follow along in crowded rooms or group settings.

• The roadmap includes plans to add new capabilities like live language translation and even emotion detection to further enhance communication.

Bottom line: This technology moves transcription from a post-event activity to an in-the-moment assistive tool, greatly improving accessibility. As augmented reality hardware improves, applications like this show how AI can directly enhance daily human interaction.

OTHER NEWS YOU MAY LIKE

 Google integrated live Maps data into its Gemini API, giving its AI real-time access to over 250 million venues worldwide for location-aware responses.

Alibaba unveiled its Aegaeon system for AI cloud computing, claiming the new architecture can reduce the number of required Nvidia AI chips by up to 82%.

Jack Clark described modern AI as a “real and mysterious creature” rather than a simple tool, with the Anthropic co-founder expressing fear about AI designing its own successors.

JOIN OUR BETA PROGRAM

🚀 Want to Make Money from AI?

We’re quietly launching a private beta program that trains a small group to become AI Implementers — People who help other founders install AI systems and get paid for it.

If you're interested in joining this early group…

👉 Reply with the word “implementer”

AI ON SOCIAL

 AI ART

RANDOM PROMPT OF THE DAY

Want more prompts?

We have a database of over 8,000 of the best prompts we’ve collected, which you can access by sharing Best of AI using the link below.

THAT’S IT FOR TODAY

Until next time, stay safe.

-Best of AI Team

Remember to click to unsubscribe if you don’t want us to land in your inbox anymore.