• Everyday AI
  • Posts
  • AI’s First Chapter: Why Generative AI Is Only the Beginning

AI’s First Chapter: Why Generative AI Is Only the Beginning

OpenAI unveils Codex AI coding assistant, Google One surpasses 150M users, Anthropic apologizes for Claude legal citation issue and more!

Author

Everyday AI
May 16, 2025

Outsmart The Future

Today in Everyday AI
8 minute read

🎙 Daily Podcast Episode: We’re at Day Zero of AI. Find out what that means from an industry vet.

🕵️‍♂️ Fresh Finds: Sam Altman wants ChatGPT to remember your entire life, xAI speaks out on Grok mishap and AWS says the UK needs more nuclear for AI power. Read on for Fresh Finds.

🗞 Byte Sized Daily AI News: OpenAI unveils Codex AI coding partner, Google One surpasses 150M users and Anthropic apologizes for Claude legal citation issue. Read on for Byte Sized News.

🧠 Leverage AI: We’re in Day Zero of AI. Got it. How do you keep up and get ahead? Keep reading for that!

↩️ Don’t miss out: Did you miss our last newsletter? We talked about Salesforce acquiring Convergence.ai, U.S. and UAE strengthening AI ties, Microsoft testing Copilot voice commands and more. Check it here!

AI’s First Chapter: Why Generative AI Is Only the Beginning

Think AI is hitting a wall? 🧱

Nope.

This is just the start.

Actually, we're at Day Zero.

Here's what that means, and how you can move your company ahead. 

Also on the pod today:

Day zero of AI and future prospects 📅
Reinforcement learning advancements 🧠
Emergent reasoning capabilities in AI 🤖

It’ll be worth your 33 minutes:

Listen on our site:

Click to listen

Subscribe and listen on your favorite podcast platform

Listen on:

Here’s our favorite AI finds from across the web:

New AI Tool Spotlight – No Cap is an AI investment project that auto-invested $100K with AI, Cuckoo is a global AI translator for teams and Open Gig is an AI product manager.

OpenAI — Sam Altman’s goal is for ChatGPT to remember your whole life.

Google – Google is adding a new generative media experience in Google AI Studio.

xAI – xAI blames its recent Grok mishap as an unauthorized modification.

Trending in AI – AI Godfather Geoffrey Hinton talks about how he uses AI models like ChatGPT.

AI in Society – AWS says that the UK needs more nuclear to power AI.

AI Safety – AI-powered school surveillance aims to keep kids safe but raises big privacy questions

1. OpenAI Unveils Codex: The Next AI Coding Partner 🧑‍💻️

OpenAI has launched Codex, a new AI coding assistant available now to ChatGPT Pro, Enterprise, and Team users, aiming to become the “virtual coworker” engineers have been waiting for. Unlike ChatGPT, Codex can autonomously fix bugs, run tests, and optimize code over extended periods, all while running securely in a sandboxed environment.

Powered by the specialized codex-1 model, it’s designed to complement existing tools like Cursor and Windsurf—highlighting OpenAI’s expanding footprint in AI-driven software development.

2. Google One Surpasses 150 Million Subscribers, Boosted by New AI Tier 📈

Alphabet’s Google One subscription service has jumped 50% since early 2024, now boasting over 150 million users, fueled by a fresh $19.99/month AI-powered plan unveiled just last month. This surge highlights Alphabet’s push to grow beyond ads, as AI features become a major draw despite traditional storage plans remaining available at lower prices.

With AI reshaping search habits and subscription models becoming key revenue drivers, Google’s strategy signals a shift in how cloud and AI services monetize in a fast-evolving tech landscape.

3. Anthropic’s AI Citation Mix-Up Sparks Legal Spotlight ⚖️

Anthropic has admitted to an “honest citation mistake” after its Claude AI chatbot generated a flawed source in a recent legal filing defending against copyright claims from major music publishers. The error, which included incorrect authors and titles despite linking to a real publication, drew sharp criticism from Universal Music Group’s attorney, who labeled the source a “complete fabrication.”

This incident highlights ongoing challenges with AI hallucinations in legal contexts—following similar courtroom controversies where AI-generated briefs contained bogus references.

4. Microsoft CEO Predicts AI Will Transform Every Business 🏢

Microsoft’s CEO recently declared that AI is poised to reshape every company, including Microsoft itself, signaling a seismic shift in how businesses operate. He emphasized that AI isn’t just a tool but a fundamental change that will redefine productivity and innovation across industries.

This moment is critical for professionals and entrepreneurs aiming to stay competitive, as adopting AI could soon be as essential as having an internet presence. The message is clear: integrating AI now can mean the difference between leading the market or falling behind.

5. Windsurf Launches SWE-1 AI Models 💨

Windsurf, known for its AI-powered “vibe coding” tools, just unveiled its own family of AI software engineering models—SWE-1, SWE-1-lite, and SWE-1-mini—marking a major shift from relying on third-party models to building proprietary ones. While SWE-1 competes well with GPT-4.1 and Claude 3.5 on coding benchmarks, it still trails the latest frontier models, highlighting Windsurf’s focus on broader software engineering tasks beyond just writing code.

This move comes as OpenAI reportedly closes a $3 billion deal to acquire Windsurf, signaling the startup’s intent to expand influence in both AI application and model development.

🦾How You Can Leverage:

AI can now reason about reasoning.

Meta, right? 

Advanced GenAI systems are designing cancer-fighting proteins over a weekend while many business leaders think LLMs are just a LinkedIn post generator.  

And according to a CTO who's been in the AI game since the 90s who joined us on today’s Everyday AI show, we haven't even started the real race yet.

Fact check shorties: we’re barely at Day Zero in AI. Like…. We’re just getting started. 

Ron cut the fat and dished how orgs need to be prepared for AI innovation to hockey stick real soon. 

(We’ll wait here.) 

Back? 

Bet. 

Now, here’s what ya need to know to move past Day Zero.

1. Your AI Has Developed Metacognition 🧠

Traditional AI needed examples of EVERYTHING to learn anything. No more, Ron said. 

Today's models can think about their own thinking. They've developed metacognition—something humans thought was ours alone.

(Dang robot. You took that from us, too?!) 

Need proof? 

Ron uses Deep Research from OpenAI to tackle problems so complex they'd take most humans days to unravel. 

Comprehensive analysis in minutes.

This isn't just faster research. It's an entirely new approach to problem-solving.

The shift from supervised learning to reinforcement learning has opened capabilities Ron never thought possible in his three decades of AI work.

Try This: 

Take your most complex business challenge and feed it to Claude 3.7 or GPT-4.5. 

Structure your prompt to request three distinct approaches, then ask for a detailed comparison of tradeoffs between them. 

Ron uses this exact technique to compress days of strategic planning into hours. 

Start with the free version before investing in enterprise licenses.

2. Hackers Just Solved What Scientists Couldn't 🕵️‍♂️

Protein folding stumped scientists for decades.

(Like folding your wife’s blouse. Is that thing a shirt or a dress? Does it even FOLD!?) 

Now college students are designing novel cancer-fighting proteins over weekend hackathons.

For reals. 

Ron detailed a mind-blowing example of AI’s progress from the University of Texas. Their BioML group organized a 48-hour hackathon with participants from 62 countries who developed 20,000 potential cancer-fighting protein sequences.

Five years ago, this was an OPEN THEORETICAL QUESTION. Now it's a weekend project.

Whuuuuuuut. 

This pattern repeats across industries. 

Ron’s company helped build a computer vision system that predicts breast cancer risk five years in advance—at superhuman accuracy levels. He said the model is currently awaiting FDA approval.

(You shorties seeing that hockey stick curve?) 

The common thread for the sudden uptick in capabilities? 

Ron said one reason was that domain-specific AI with "one or two amazing superpowers" often delivers more immediate ROI than general-purpose tools.

Try This:

Identify your company's most valuable proprietary dataset—the one competitors can't access. 

That yummy paydirt. 

Instead of building a general AI chatbot, develop a narrow-focused AI solutions trained specifically on that data. 

Ron said to start with a simple prediction task: customer churn, inventory needs, or price optimization. 

Start simple and measure AI’s impact where it hurts. 

If those students can find potential cancer-fighting proteins, your tech dream team can use LLMs to address your pain points. 

3. The Corporate AI Implementation Gap 🏢

The disconnect between what's possible and what most companies are doing? 

Yuuuuuge. 

The room always splits 50/50.

Half have revolutionized their workflow. The other half think it's not worth their time.

At that point, Ron said companies move into the Day Negative 1. 

Day Zero has moved from experimentation in 2024 to implementation in 2025. 

If you’re not already implementing LLM solutions top to bottom, you’re not even at Day Zero. 

This isn't just a technological gap. 

More like a business extinction event in slow motion.

Ron said that within 12-36 months, we'll see systems that can reason at such sophisticated levels they'll need to "dumb down" explanations for humans to understand them.

Sweet?

Try This: Run this 1-hour experiment tomorrow. 

Have two teams tackle the same programming challenge—one using AI coding assistants, one without. 

Time both groups and compare not just completion speed but code quality. Green has run this test with clients and consistently sees 3-5x productivity improvements. 

After the test, implement mandatory daily "AI pairing sessions" where developers work alongside tools like Copilot for at least 45 minutes to overcome initial resistance.

Reply

or to participate.