• Everyday AI
  • Posts
  • Day Zero of AI: Why Generative AI is Just the Start

Day Zero of AI: Why Generative AI is Just the Start

Google drops huge Gemini and Gemma updates, Apple's AI being buggy, China's Manus and Qwen team up.

Author

Everyday AI
March 12, 2025

👉 Subscribe Here | đź—Ł Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI

Outsmart The Future

Sup y’all! 👋

Apparently, today’s show was a banger.

Oh, and tomorrow’s show?

Hot dang.

You won’t wanna miss it. (With someone who kinda helped develop AI as we know it today.)

✌️

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: Companies going crazy over DeepSeek, a new YouTube-AI hack, and a new $1 billion AI investment from Salesforce. Read on for Fresh Finds.

đź—ž Byte Sized Daily AI News: Google drops huge Gemini and Gemma updates, Apple's AI being buggy, China's Manus and Qwen team up. 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 OpenAI is opening up its computer-using API, Squarespace releases AI website builder, Meta's creating its own GPU as tech giants struggle to secure compute and more. Check it here!

Day Zero of AI: Why Generative AI is Just the Start

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 đź“…
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:

Tomorrow’s Everyday AI Show

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.

AI Usage — Half of Americans now use AI like ChatGPT, with many saying it feels surprisingly human—personality, humor, and all. What’s driving this rapid adoption?

Chinese AI – China’s DeepSeek AI is sparking a frenzy, with industries racing to integrate it

Google AI Studio – Google’s newest updates bring a pretty useful YouTube hack.

Salesforce and AI – Salesforce is investing $1 billion in Singapore, cementing the city-state's rise as a global AI hub.

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

AI Health Studies – Mass General Brigham’s new AI tool analyzes sleep brainwaves to predict dementia years early

AI Oversight – Spain's new AI bill cracks down on unlabeled deepfakes and harmful practices with fines up to $38M—transparency just got serious.

1. Google’s Gemma 3 AI Model Ups the Game 🎮

Google has unveiled Gemma 3, an upgraded AI model that promises cutting-edge performance while running on minimal hardware. With support for over 35 languages and the ability to analyze text, images, and short videos, it’s being touted as the “world’s best single-accelerator model,” outperforming rivals like Meta’s Llama and OpenAI on single GPUs.

Key improvements include a more robust vision encoder for high-res images and the ShieldGemma 2 classifier to filter harmful content. While still restricted under Google’s specific licensing terms, the release could make sophisticated AI tools more accessible for developers and businesses alike.

2. Google’s Gemini AI Takes Robotics to the Next Level 🦾

Google has unveiled two new AI models, Gemini Robotics and Gemini Robotics-ER, powered by its advanced Gemini 2.0 technology, marking a major leap into robotics. The models enable humanoid robots developed with Apptronik to perform physical tasks like zipping bags and packing lunchboxes based on spoken commands.

Google aims to make robots adaptable, interactive, and dexterous—qualities essential for real-world use—and is collaborating with top robotics firms like Boston Dynamics and Agile Robots. This move signals a growing race in AI-powered robotics, as competitors like OpenAI and Tesla also push boundaries in merging AI with physical action.

3. Apple’s AI Won’t Take No for an Answer! 🙅‍♂️

Apple's latest iOS 18.3.2 update has sparked frustration, as it automatically re-enables Apple Intelligence—even if users had previously turned it off. According to CNET, the beta-stage AI platform has faced criticism for dodgy summaries, unpredictable image generation, and delayed improvements to Siri, leaving many wondering why it’s being forced back on by default.

While turning it off remains relatively simple, the features can hog up to 7GB of storage, making it a headache for users who don’t see the value in Apple’s AI offerings.

4. Google Gemini 2.0 Flash Unleashes Conversational Image Editing 🖼️

Google is stepping up its AI game, with Gemini 2.0 Flash now offering conversational image editing and multimodal capabilities. This update lets users generate and refine images through natural dialogue while preserving context, making it ideal for tasks like visual storytelling or step-by-step recipe illustrations.

Gemini’s ability to combine text and images seamlessly leverages enhanced reasoning and world knowledge, giving it an edge over standalone models. Developers can dive into this experimental feature today in Google AI Studio, though daily limits apply.

5. China's AI Agent Manus Joins Forces With Alibaba’s Qwen 🤝

Manus AI, the viral autonomous agent developed by a Tencent-backed startup, is teaming up with Alibaba's Qwen team to create a localized version for Chinese users, according to the South China Morning Post. This collaboration is part of China’s AI arms race, with Manus leveraging Alibaba’s open-source models to ensure features run on domestic infrastructure.

Notably, Manus is already drawing comparisons to top-tier AI players like DeepSeek, while Alibaba claims its new reasoning model outshines competitors like OpenAI. With Alibaba pushing for full AI adoption among merchants by 2025, this partnership could ramp up innovation and reshape how Chinese businesses integrate AI into everyday operations.

🦾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 prepare 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 đź•µď¸Źâ€Ťâ™‚️

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 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.

⌚

Numbers to watch

$50 million

OpenAI launched a $50 million fund to accelerate scientific breakthroughs.

Reply

or to participate.