Hi fam!
AI apparently had another quiet week.
Just kidding.
While we were trying to get some actual work done, Google went after lawyers, Anthropic put money into studying what AI is doing to us, OpenAI lost another senior executive, and AI somehow found its way into designing new materials.
So, business as usual.
This week, we're looking at what actually matters, sharing one tiny prompt that can rescue you from bad first ideas, and exploring a much smarter way to use AI when you're staring at a task you really don't want to do.
The future of AI feels Like Magic, and it’s here!
AI news this week isn't dominated by one giant model launch—and that's actually what makes it interesting.
Instead, we're seeing AI move deeper into real professions, scientific research, corporate infrastructure, and even questions about our own wellbeing.
Here are five stories worth your attention:
Google has launched Gemini Enterprise for Legal, bringing specialized AI agents and tools to law firms. The system can assist with contract review, regulatory tracking, document searches and other legal workflows, while integrating with established legal platforms including Thomson Reuters, Harvey, Relativity and iManage.
Why this matters: We're moving from “here's an AI that can do everything” toward AI built around specific professions and workflows. For businesses, specialized AI may ultimately prove more valuable than another slightly smarter general chatbot.
Anthropic announced a $5 million grant program for independent research into AI wellbeing—essentially studying how increasingly capable AI systems affect the people interacting with them. Research areas include emotional reliance, human-AI relationships and how AI influences human behavior.
Why this matters: We've spent years asking whether AI is safe. Now the question is becoming more personal: What happens to us when talking to AI becomes a normal part of everyday life?
MIT researchers have developed an AI-driven approach that doesn't just generate theoretical new materials—it helps identify designs that can actually be manufactured and function in the real world. The method combines AI generation with physics-based constraints to narrow millions of possibilities into useful candidates.
Why this matters: Some of AI's biggest breakthroughs may have nothing to do with chatbots. AI is increasingly becoming a tool for discovering things humans might never have thought to design in the first place.
OpenAI's head of data centers, Chris Malone, has left the company following a reorganization of its infrastructure division. His departure comes amid a broader wave of senior exits as OpenAI simultaneously scales enormous infrastructure projects and prepares for its next phase as a business.
Why this matters: The AI race isn't only about models anymore. It's also about chips, electricity, data centers and the people capable of building infrastructure at an unprecedented scale.
More than 80 performers and public figures have backed the UK's Save Our Voices Now campaign, calling for stronger legal protection against unauthorized AI voice cloning. The campaign argues that people should have enforceable rights over digital replicas of their own voices.
Why this matters: Voice cloning is becoming good enough that “Who owns your voice?” is no longer a hypothetical question. Laws around identity, consent and AI-generated replicas are now racing to catch up with the technology.
The “Give Me the Bad Version First” Prompt
AI has a problem.
It's desperate to be helpful.
Which sounds wonderful until you're brainstorming and every idea it gives you feels like it was approved by a committee of extremely polite LinkedIn consultants.
So try something slightly backwards.
Instead of asking:
“Give me five great ideas for this.”
Ask:
“Give me five deliberately bad, obvious, cliché versions first. Then explain what makes them predictable and create five ideas that avoid those patterns completely.”
Why does this work?
Because you're forcing AI to identify the clichés before generating the answer.
Say you're naming a sustainability startup.
Without this trick, prepare yourself for:
EcoSphere.
GreenFuture.
EcoNova.
Somewhere, a leaf-shaped logo is already being generated.
But once AI has identified those obvious patterns, you can tell it to deliberately move away from them.
It's useful for headlines, campaign concepts, product names, presentations, scripts, visual ideas and anything where the first AI answer feels painfully... AI.
Sometimes the fastest route to a good idea is making AI show you exactly what not to do first.
Stop Doing the Task. Change the Task.
Most of us use AI after we've already decided how a task should be done.
We ask it to summarize faster, write faster, analyze faster or automate another step.
But there's a better question:
Do you need to do the task that way at all?
Give AI your goal and existing workflow, then ask it to identify what can be eliminated, combined or automated before you start.
Sometimes the biggest productivity win isn't turning a three-hour job into a two-hour job.
It's discovering that most of the job never needed to exist.
From the Archive: Automate Time Consuming Tasks: AI is scripting, you use it
Back in 2025, we looked at a surprisingly simple idea: you don't need to know how to code to automate boring work anymore.
Give AI a repetitive task, explain what you want to happen, and let it help you build the script.
A year later, that idea feels even more relevant. AI can now automate increasingly complex workflows—but the principle hasn't changed:
If you're doing the same boring thing over and over again, maybe you shouldn't be doing it manually at all.
A perfect throwback for this week's theme of questioning the work before simply doing it faster.
Text generation | Image Generation | LMAI recommends |
|---|---|---|
Like Magic AI NFT 🏞️
Our master plan is to publish an NFT image in each newsletter and hand it out to our subscribers. The earlier you subscribe, the smaller the series are. It's a future collectible, a piece of digital art that captures the essence of this moment in time.
Thank you for being a valued subscriber. Together, let's embrace the magic of AI and creativity!
LMAI228-28082026
Was this email forwarded to you? Sign up here 👇
