You won’t get fired. You’ll just get less relevant. The AI won’t replace you all at once. It’ll do 5% of your job, then 15%, then 40%. And at every step, your manager will say, “It’s just helping.”. But eventually, you’re no longer the operator. You’re the overseer. Then the.

Personality cloning (mimicking a style of speech) is the next dimension of speech models. Many subliminal layers of realism here.

Meta's proposed Hyperion Data Center size compared to Manhattan

No scenario in which this company doesn’t end up getting acquired for billions by one of the big 4.

The best part about AI’s amazing capabilities . is that it’ll be so laughably limited compared to what we’ll have in 2 years. …and so will that when compared to what we’ll have in 5 years.

Just used to build a real full-stack app from a single sentence - with Grok 4. No devs. No code. No BS. Type what you want → It builds it. Frontend + backend. Now with:.✅ Google Auth.✅ Built-in LLMs (no API keys!).✅ Security + scale agents.✅ Mobile

The Manhattan Project cost 5% of GDP. You can get Grok 4 for the price of Chipotle.

We taught rocks how to think, and now they're teaching us.

I’ve been using @contextsuite (at my real job) for a few weeks now and it’s wild how much time it saves. Context is like if Notion, Google Docs, ChatGPT, and your file system all merged, but with an AI that actually knows your work. Here’s what I’ve done with it so far:.

4 PhD-level agents working in tandem. easiest $3000 I ever spent

At this rate, we’re going to run out of benchmarks in 1–2 years. Every new model will max out any human-generated capability metric and will evolve into something we don’t know how to measure. It’ll be like trying to weigh a supernova with a kitchen scale.

Somebody check on Gary Marcus.

Sneak peak - SuperGrok Heavy is $3,000/yr

I had over a dozen people email me today after I sent out a newsletter titled “Grok 4 Release Today”, calling me a Nazi or a Nazi supporter for attempting to “normalize” Grok. It’s absurd, but also familiar. The same people calling Elon a Nazi now are the ones who used to make.

This week's sponsor of the AI Breakfast newsletter is Tanka AI - which is like Gemini in Gmail, but for all of your apps. It can reach into Slack, Whatsapp, Telegram, Notion, Outlook, Calendars, etc via MCP and query all of them based on your prompt. Still crazy to me that

Apparently you can completely derail AI's ability to reason when you throw in "cat facts" or other out-of-context materials into the prompt. Error rate goes up 10x. When you are facing down a Terminator, just tell it that cats sleep for most of their lives.

MARBLE exposes the gap between mimicking intelligence and real cognition. Planning, perception, execution is still human. Scale won't solve it. Solve MARBLE, and you build the next trillion-dollar model. link to paper:

Reasoning is worse. The CUBE task has 188 million possible solutions. Even when perception is bypassed, models collapse under combinatorics. They cannot hold multi-step chains. They cannot search.

Perception is the bottleneck. MLLMs can’t even transcribe a simple 5x5 grid from an image. The best models got 0% accuracy on full-shape recognition. They see… but don’t understand.