Cassie Kozyrkov

A huge thank you to all of you for encouraging me to write! I just noticed that on Medium, my community of followers is 70% the size of Barack Obama’s. Whoa!

I’m honored and humbled by all the love this amazing community has given me. I don’t know if it’s despite my being a cheerful weirdo or because of it, but thank you! And thank you for being unresponsive to my job title when I turned it off and on at random several times, with no effect on any of the metrics — it means a lot that you’re here for my thoughts and not my labels. Especially since it has been almost 10 years since I’ve had a career change and who knows what kind of madcap adventure I’ll pick when I eventually decide that a change is a good as a holiday.

It was been a great pleasure to share helpful musings with you, but now that I’ve published 180 blog posts, many of you have told me you’re drowning in all my content and I need to index it better. Turns out it’s very confusing for newcomers to my blog to sort through all the different topics I write about. I hear you! Not everyone is here for all the things. Eventually, I’ll prepare a well-curated site to help you out, but in the meantime, let me take the first step towards a fix by adding standardized supertitles to all my articles. That way you’ll know what category you’re dealing with each time so you can dive right to the ones you care about and skip my musings on random esoterica. In essence, it’ll be as if I have mini-publications for you to chose from.

But if you’re a bit more narrowly focused, hopefully this new index will add some rhyme and reason to your knowledge feast:

https://lnkd.in/efMrrQNy

The categories:
🎲 Decision Skills
🎲 Data-Driven Leadership and Careers
🎲 Making Friends with AI
🎲 Irreverent Demystifiers
🎲 Statistical Thinking
🎲 Making Data Useful
🎲 How To Hack Yourself

Please let me know in the comments which one you enjoy best! 💖

Two truths and a lie:

🧠 OpenAI's CTO Mira Murati resigns just as the company moves to a for-profit model.

🧀 Scientists made skin of mice see-through by rubbing a solution of tartrazine on them. You may know tartrazine from its greatest hits, including Doritos and mustard.

💸 Offering GenAI to the GenPublic is profitable on a per query basis. State of California to punish nonprofit-themed jokes with fines.

Edit: Award for best comment so far goes to Rakesh Sookay. “This pic is proof that AI can take your job.“

How often does a technology come along where the C-suite not only needs to be championing its adoption (by other people) but should personally start using it immediately? Yes, I'm talking about LLMs like GPT-4.

GPT-4 is to large language models (LLMs) what a cutting edge laptop is to computers. As we saw with the jump in usefulness from ChatGPT to today’s ChatGPT Plus, the tech is evolving quickly. Honestly, it’s my hope that this post ages as badly as a poem praising the first Macbook. Innovation never sleeps, so you can expect that the future will bring us far better LLMs than GPT-4.

The important thing to know is that LLM technology is here to stay and, no matter who you are, it would be a dodgy career move to ignore it, which is all the more reason for you to have a hands-on sense of what it’s about, so do try it. You don't need any expertise and you can't do anyone harm if you keep your (un-fact-checked) output to yourself. Follow the link in the article to get set up in minutes.

One of the most audacious things about the LLM revolution is that almost no one’s work is spared a potential productivity boost. Yes, this technology has its place on the CEO’s own laptop, right next to email and Solitaire.

I’m suddenly reminded of a factoid I heard about how secretaries in the 1970s feared they’d be replaced by robots… but were instead replaced by their own bosses. Previously accustomed to dictating all their correspondence, managers now felt that clacking away on a computer was, unlike a typewriter, no longer beneath them and absorbed many of their secretaries’ duties.

That’s the kind of revolution we’re about to see, folks. Not job replacement by machines but rather an unprecedented spike in individual productivity.

Maybe this post won't mean much to most of you who follow me, but it means a lot to me.

Today, on Thanksgiving, I'm naturally very grateful for all that makes life worth living: my family, my friends, my health, and so on. But since you guessed all those before you even started reading, let me share something I'm excited about and grateful for that you might not have guessed.

Since childhood, I've been two kinds of nerds. You know me for the achievements of *one* of my nerdy sides, the data-decisions-AI one. After all, the majority of you are technologists, STEM enthusiasts, and leaders who share my passion for AI and decision-making.

Many of you have asked me how, coming from a quantitative background, I got so good at public speaking. My secret is that I've been harboring a second inner nerd this whole time. I'm obsessed with theatre. Public speaking has been my career-appropriate way of finding an outlet for that passion, but it goes much deeper than that.

But, while I spent many years chasing degrees in STEM, I've never graduated from any kind of theatre program.

So I'm beyond grateful that I've been accepted into one of the most prestigious actor training programs, the Shakespeare and Company January Immersive, following in the footsteps of names like Bill Murray, Sigourney Weaver, Keanu Reeves, and my personal favorite, John Douglas Thompson, whom I've had the privilege of seeing on stage four times.

If I'm not posting as much as you're used to this winter, it's because I'll be immersed in the three week immersive like a kid in a candy store. (And if you do see me post, it'll be because I've spent the preceding weekends pre-writing content like a maniac.) I'm even -- *gasp* -- giving up Davos to do this.

Don't worry, I'm not changing careers, just nourishing the nonobvious side of what makes me me. When you see it shine through in my work as an AI leader, advisor, and speaker, you'll be in on my secret.

Meanwhile, I'm beyond excited!

What’s different about today’s #AI? Hint: The AI revolution isn't an AI revolution, it's something else entirely...

And the answer is UX.

Long before the generative AI art apps and chatbots splashed their way onto the tech scene, artificial intelligence was powering your online experiences and optimizing many of the businesses you enjoy offline. AI was everywhere, but it was hidden in plain sight - on purpose! - by user experience (UX) designers following a very intentional strategy.

Last decade’s UX design philosophy for AI prioritized seamless, reliable solutions with minimal user tinkering.

What’s happening today is a radical pivot in UX design philosophy; suddenly, the user is encouraged to interact directly with an AI component that’s designed for usefulness rather than seamless correctness.

That’s right. Behind the scenes, UX designers are permitting themselves to serve you an unprecedented user interaction: an AI system that’s allowed to serve garbage… as long as it’s useful garbage.

The last few months weren’t about AI capabilities. The AI capabilities revolution has been going for a few years now and the public hardly batted an eyelid.

Many writers are correctly observing that GPT (the core tech) models were available several years before ChatGPT (the interface that let non-programmers interact with the core tech easily), hence today’s crescendo is about users… but there’s more to it than that.

It's about more than user interfaces.

It’s about user expectations.

What do I mean by that?

Read on in my blog: https://lnkd.in/eeH4mcqd

Once you take user expectations as your starting point for analysis, so much of the ecosystem will make instant sense. It’s a design revolution and it’s just getting started.

https://lnkd.in/eeH4mcqd

Datasets are not all created equal and people are not intuitively good data designers. It takes knowhow to design the data collection so it’s quicker and easier to make that data useful. The good news is that there are resources to help you get better at it (see the link). The bad news is that the industry is still suffering from far too much snake oil.

The data careers space has been disgustingly hyped up. Not overhyped (there is incredible value to be had from data), but more like “mishyped” — a lot of people are generating data buzz for all the wrong reasons.

The right reasons to be excited relate to the old adage: knowledge is power: the power to improve your business, your work, your personal life, and the world around you. With all of the technological improvements in storing and processing the raw materials of knowledge, there’s so much potential just waiting to be unlocked. That’s worth several truckloads of hype.

But I hope you’ll join me in saying no to the mishype: don’t equate data with magic. Garbage doesn’t turn into gold, no matter how much math you throw at it. Industry’s fascination with data alchemy is as embarrassing as the 15th centuries fascination with muttering Latin mumbo jumbo over iron filings.

I wish we’d all stop pronouncing data with a capital ‘D’. Data aren’t magic — just because you have a spreadsheet full of numbers doesn’t guarantee that you’ll be able to get anything useful out of it. The GIGO principle applies as strongly as ever.

GIGO: Garbage in, garbage out.

Data isn’t truth. Far from it. In the link, I'll illustrate this with a humorous (but true) example...

https://lnkd.in/dsWYaKKq

#datascience #datadesign #machinelearning #ai #statistics #dataanalytics

If only machine learning had the equivalent of a Swiss Army knife that was 80% faster to use than the regular toolbox... Good news, as of today it does! 🥳

I usually try to keep things civil on here, but this time I invite you to have at it.

My friend who is a genius at what she does -- but what she does is not tech -- recently quit her awesome job to join an AI startup (in a non-tech role) where the engineer who just left was the only one who knew how the code worked (because he wrote it) but he didn't document the code at all. Surprise, surprise, now no one can make it work in prod.

Have at it, my lovelies. Let's have a therapy session since we've all lived this if we've been in tech for a minute. I explained to her about code documentation (easy) and data documentation (hard), but there's only so much one person like me can help. More stories will help her feel less alone, so please share yours. It sucks to experience this for the first time if you're a coder, but I can only imagine how unpleasant it must be for a non-coder like my friend who trusted that the engineer would do his job so she could do hers.

Comments, lessons, and words of encouragement welcome, painful lessons are funny for those who haven't lived them (and also generally funny when they're funny), go forth.

And also, if people have good rules of thumb they can offer non-technical folks for checking whether they're dealing with the real deal and how to spot a bad engineer (ahem, no comments in the code is one dead giveaway, but do share more advanced tells), throw them in there, pretty please. ## Let's get this post well-commented!

And here's hugs for anyone who's had to deal with this!

Moral: AI doesn't just magically work in prod. You need good engineering. And higher than usual standards of documentation. Lots of people are selling lies. This is the reality.

DATA + ASSUMPTIONS = PREDICTION

When your data are the size of the moon while your assumptions are the size of Jupiter, good manners demand that you paste an apology to your readers on Every. Single. Line.

If you're making assumptions and you're not a domain expert, they're probably extra pungent garbage assumptions. Apologize several times on every line of your report.

If your report is mostly apology, so much the better.

Video link below.

Added a few more definitions to the Stats Gist List here: https://lnkd.in/eWGwWYQz

Enjoy scedasticity explained as sausages vs trumpets, plus a bunch more. 🌭🎺

Unsupervised learning is an algorithmic way to implement “birds of a feather flock together.“

To do supervised learning, you supply the right answers. Not so with unsupervised learning...

Learn more:
Supervised - https://lnkd.in/ddNJipu
Unsupervised - https://lnkd.in/dezvvjJ

#datascience #ai #machinelearning #analytics

So proud of my friend Emma Haruka Iwao for setting a new badass π record: 100 trillion (!!!) digits of pi calculated.\n\nhttps://lnkd.in/gVbJT7zG\n\n(Same Emma as the one from the dress-shopping story in my article https://lnkd.in/etyhe7n, by the way. She's a rockstar.)

It's official, 🦾Jepson Taylor and I are co-founding an AI company and we are so excited! I know, I know, it was cruel to tease y'all by releasing the first of these photos on April Fool's Day, but keeping you guessing was the extent of the prank... our startup is very real indeed.

I'm thrilled to have started this journey with Jepson, our CTO and a great friend whom I met just 5 days short of 5 years ago (April 10, 2019) -- can't wait to share all the details of what we're building as soon as we exit stealth mode.
(Investors who'd be interested in hearing a pitch in a few weeks, please go here: https://lnkd.in/eW7t9REQ )

🛳 🛳 🛳 Oh, and water seems to be a theme with us... if you happen to be at Summit at Sea where we're celebrating Jepson's birthday, do come say hi! 🛳 🛳 🛳

When should you buy AI vs build AI?

Most teams start with the wrong assumption: that building means innovation. In reality, it usually means distraction.

Buying wins when speed, scale, and integration matter. Building only pays off when your data itself is a competitive moat: proprietary, compounding, and inaccessible to others. Without that, “custom AI” just burns time and budget on something vendors already do better.

Building makes sense when your advantage depends on capabilities outsiders can’t deliver. That happens when:

🔸 You have genuinely proprietary data.
🔸 AI is core to your product differentiation.
🔸 Your operations are so specific external tools won’t fit.
🔸 The solution you need doesn't exist yet.

The smart move isn’t to build more... it’s to build what's unique to your business.

Oh, and never recreate the untrained intern. If anyone can do it, AI will do it better and your specialized vendor's AI will do it better than yours.

Your turn: Do you think most companies overestimate how truly unique their data is? Type "Yes" or "No" below. Your comments make the (digital) world go around and your reposts ♻️  make my day.

#AILeadership #DecisionIntelligence #GenerativeAI #AIinAction
#Strategy #DigitalTransformation

Don't forget to mash, ahem no, daintily tap that follow button for more.

Happy Halloween, friends! Nerd points to you if you can figure what I'm dressed as. 🎃

AI won’t rescue you from the laws of innovation.

Two truths every leadership team needs to internalize:
• Success demands vision.
• Innovation demands waste.

If you’re trying to deploy AI “at scale” while insisting on perfect KPIs, guaranteed ROI, and zero missteps, you’re not innovating. You’re role-playing certainty.

Innovation is learning under uncertainty.
And learning is never linear.

So the real question isn’t “How hard should we push?”
It’s “Do we actually have the appetite and budget to learn?”

Because if you don’t, the smartest strategy isn’t acceleration.
It’s patience: let others stub their toes, learn from their mistakes, and follow when the path is clearer.

There’s no shame in that.
There is risk in pretending AI repeals reality.

📍AI Advice From Anywhere: can you guess the city before I reveal it?

The bigger question: Is your organization built to learn? Or built to wait until others do?

#AAfA #AILeadership #DecisionIntelligence #EnterpriseAI #Innovation #AIStrategy

The most dangerous AI isn't the one that sounds robotic... it's the one you don't question.

When AI outputs are smooth and confident, something shifts in our brains.

We stop scrutinizing. We assume accuracy. Psychologists call this the fluency heuristic -- and it's costing organizations real money and credibility.

I've watched polished AI-generated strategies sail through approval processes that would have shredded a human's work.

Not because the AI was right, but because it sounded right.

It's time for leaders to recognize this and encourage your teams to build checkpoints into decision workflows — small friction points that preserve critical thinking without killing momentum.

Polished but flawed things are everywhere these days. Have you come across any particularly egregious examples? I’d love to hear them.

Your comments make the (digital) world go around and
your reposts ♻️ make my day.

#AILeadership #DecisionIntelligence #GenerativeAI #AIinAction
#Strategy #DigitalTransformation

Don't forget to mash, ahem no, daintily tap that follow button for more.

What’s the most valuable skill in the AI era?

⛔ It’s not coding.
⛔ It’s not prompt tricks.
⛔ It’s not “keeping up with the latest model.”

💪 It’s judgment.

Every time a new AI capability drops, the debate centers on how powerful it is. That's the wrong question.

We should be asking how powerful *we* are.

Imagine this:

You can get an instant answer to any question.
You can generate output for any request.
You can simulate, draft, analyze, and model at the speed of thought.

Now ask yourself:

Are you good at deciding what to ask?

AI is removing nearly every technical bottleneck. The one that remains — and may become more painful — is our ability to think clearly about messy, ambiguous, high-stakes problems.

When the tool is no longer the constraint, the spotlight shifts to the decision-maker:

🔹 Do you understand the problem deeply enough to frame it well?
🔹 Can you anticipate failure modes?
🔹 Can you tell the difference between a technically correct answer and a strategically useless one?
🔹 Are your assumptions sound?

For decades, difficulty acted as a filter. You had to understand a problem to solve it. Now you can generate solutions without earning that understanding.
The filter is dissolving. That changes the game.

Full piece (and this week’s AI news roundup) here: https://lnkd.in/gwFhfSU7

Curious to hear how you’re investing in your judgment skills.

Gemini 3 is here and for the first time in years, Google is not chasing the frontier... it *is* the frontier.

In the last 48h, Google has leapfrogged all the other frontier models.

And that's not all:
💪 Nearly half of Google’s new code is now AI-generated and engineer-reviewed. A sign that Google is already operating at a different internal velocity.
💪 Google's Ironwood TPU delivers 42.5 exaflops — about 24× the power of the world’s largest supercomputer (El Capitan).
💪 Gemini Enterprise shows a full-stack strategy: model + infra + agents + security + workflow integration, all under one roof.
💪 Antigravity, Google's Agentic IDE, is here. It's designed for an “agent-first” workflow where AI agents actively operate across the editor, terminal, and browser.

Google hasn’t just rejoined the race — it’s out in front, and the rest of the ecosystem now has to respond. And they are responding.

Nvidia, Microsoft, OpenAI, and the world-model labs each dropped their own shockers — and all but one are industry-defining in their own right. Can you guess who the laggard is before you read the deep dive in my newsletter?

Newsletter: decision.substack.com

Before you look it up, put your guess below. 👇 After all, your comments are how posts get seen and if you liked this one, it's a lovely way to help others enjoy it.

Your comments make the (digital) world go around and your reposts ♻️ make my day.

For a full picture of how every AI behemoth just repositioned itself, my analysis is here: bit.ly/quaesita_behemoths

Don't forget to tap that follow button for more (especially if you don't want to miss my post on AI in China next week).

#ai #google #nvidia #openai #microsoft

P.S. My Decision-Making with ChatGPT course runs on demand next week with a 2h live Q&A. Enroll here: https://lnkd.in/eG7EBtTv (for promo codes, head over to the newsletter article 👆)

The #AI Reliability Paradox: here's why “better” can still break you.
We celebrate performance. In human workers, in algorithms, in AI. But when it comes to reliability, excellence hides a dangerous paradox.

Here’s the thought experiment:
🤦 Chris Careless fails 30% of the time. Poor dear.
🏆 Ronnie Reliable has never let you down.

Who’s riskier?
It might be Ronnie. But that depends on you.

The problem isn’t the failure itself. It’s trust.

When leaders assume “near-perfect” means “perfect,” they stop building safety nets.

That’s when a small statistical tail becomes a large organizational wound.

In AI, this pattern is everywhere:

🔸 A model that looks 99.99% performant in testing
🔸 Gets rounded up to 100% in executive minds
🔸 Then fails spectacularly when deployed at scale

Because when you increase the scale, you meet the long tail.

🏗️ The fix:

Don’t lower your safeguards based on performance.
Build them based on what’s at stake when the system fails.

When, not if. That's not an AI thing, that's a complex systems thing. The more complex the system, the harder reliability is. Even, and sometimes especially, when it looks highly performant.

Whether you see it in humans or machines, excellence deserves respect, not blind faith.

Plan for failure even when you’ve never seen one. Especially then.

Your turn: Have you caught people assuming good enough means perfect? Your comments make the (digital) world go around and
your reposts ♻️  make my day.

#AILeadership #DecisionIntelligence #GenerativeAI #AIinAction
#Strategy #DigitalTransformation

Don't forget to mash, ahem no, daintily tap that follow button for more.

NVIDIA CEO Jensen Huang described AI as a five-layer cake. 🍰

Chips, models, applications, energy, data centers. A full stack. A complete system. A solved problem. Delicious. 🍴

Except there’s a layer missing.

When everyone has access to the same models, the same tools, and the same compute, it’s tempting to assume the playing field is level. It's not.

The gap doesn’t disappear. It moves.

In this episode, I talk about the part of the AI stack that isn’t hardware, software, or data, but ends up determining everything anyway.

📍 AI Advice From Anywhere: Watch to the end and see if you can guess the city before the reveal. Post the number of seconds it took you in the comments. (No spoilers.)

(Oh, and if you haven’t seen the breakdown, it’s worth a read: 
https://lnkd.in/e68TK_Nf)

#aafa #DecisionIntelligence #AILeadership #FutureOfWork #ExecutiveDecisionMaking #AIAdoption

Data helps us see what is... but too often we forget that the future lives beyond the dataset. 
 
I had the thrill of being in the room for a live taping of Smart Talks with IBM, hosted by Malcolm Gladwell, and this one was very smart indeed. 
 
In it, #IBM Chairman and CEO Arvind Krishna warned that being “data-driven” can quietly become history-driven – mired in tradition, trapped in what’s already happened, unable to see the future, let alone build it. 
 
Most of us struggle to “see the forest for the trees.” Fewer still can imagine a forest at all when they’ve only ever seen a tree or two. That’s the deeper challenge of leadership in technology: to notice tiny signals, imagine the larger shape they could form, and then do the long work of making others see it too. 
 
The recording is full of great foresight/hindsight shockers from Krishna's own early career moments (like championing Wi-Fi to the nonbelievers) and from the history of technology (have a giggle at the telephone bit, but no more spoilers, you'll love it): https://obvs.ly/Cassie
 
Okay, one spoiler: quantum computing.  
 
Krishna says quantum is today's version of that blind spot. Most people don't realize what it means when this technology scales up. HSBC is already testing quantum methods that beat traditional bond pricing by 34% (!!) and while the scale of the pilot is still small, it's probably unwise to bet it stays there. Krishna's prediction for the first truly shocking quantum breakthrough? Within 3-5 years. 
 
It's a misconception that quantum is just about faster computing; it's about unlocking new algorithms that are suited to a fundamentally different kind of math. Take chemistry and materials discovery, for example: CPUs and GPUs struggle here because they must simulate electron states one by one... with exponential blowup. Quantum machines can explore many states simultaneously. Not in parallel, like GPUs do. Simultaneously. That unlocks today’s "impossible" tasks -- like designing safe solid-state batteries. (Battery makers take note: ignore quantum and Krishna thinks you could be gone in a decade.) 
 
Imagination isn’t enough. The real challenge is bringing others with you -- persuading when you’re the only one who sees what’s coming, and then persisting through the long, unglamorous push after the aha moment. 
 
If data is the past and imagination is the future, then persistence is the present. And that’s the lesson: to lead, you need all three. 
 
So don’t be *too* data-driven. Data shows you the trees--today’s facts. Use it well, but don’t let it trap you in the past. Use it to understand the present but dare to imagine beyond it. Then bring people with you and persist until what only you saw becomes everybody's reality. And that's the hardest part. 
 
At one point, Malcolm Gladwell paused and said, "You're very persistent." 
Krishna didn't miss a beat: "Very." 
 
👉  https://obvs.ly/Cassie 
 
#IBMPartner #Sponsored

What's even more fun than gaining knowledge? Sharing it!

What a year! I had the privilege of keynoting in 7 countries in person and many more virtually.

Thank you to all the clients and audiences who trusted me with their time and attention this year: Microsoft, The Walt Disney Company, Salesforce, Capital One, IBM, Walmart, Infosys, ServiceNow, Cloudera, Acceldata, Celonis, JetBrains, Discover, Athena, Boomi, Gentera, Appian, NiCE, Statsig, Zayo Group, Neuberger Berman, Madison Realty Capital, HFS Research, Akbank, Ataccama, Degreed , Turner Construction Company, Sullivan & Cromwell LLP, Cresta, Kong Inc....

And so many more!

Here's to a brilliant 2026, everyone.

P.S. If you're looking for a speaker to elevate your decision leadership for the AI era, get your workforce inspired to adopt AI-first thinking, and build a human-centered AI culture that thrives, there are ways of making me talk: makecassietalk.com 😉

You can “win” your AI rollout and still lose your workforce.

A recent study of 430 manufacturing employees found that even successful AI transformations increase emotional exhaustion.

Translation?

Your KPIs can trend up, 📈
while motivation trends down. 📉

The research identifies two distinct fears:
• Replacement fear: “Will AI take my job?”
• Learning fear: “Can I keep up?”

They look similar.

They are not.

Each requires a different leadership response.
Most organizations treat them as one.

That’s where transformation quietly derails.

AI success isn’t just about algorithms--it’s about emotional infrastructure.

If you’re accountable for AI outcomes, this week’s AI Advice from Anywhere breaks down:
– Why anxiety drains discretionary effort
– Why service-oriented leadership isn’t enough
– Why learning time must be explicitly protected
– And why burnout should be tracked alongside AI KPIs

📍 Bonus: Can you place the city before I reveal it?

If you’re leading AI transformation, this one matters.

#AAfA #AILeadership #DigitalTransformation #ExecutiveLeadership #FutureOfWork #DecisionIntelligence

Surveys claim that up to 1 in 4 adults and 1 in 5 teens have had a “romantic” interaction with AI. What should we make of these numbers?

AI is becoming the new “practice partner.” But what happens when practice replaces reality?

AI companions are exploding in use:
🧩 Replika → 30M+ users
🧩 XiaoIce → 660M+ users
🧩 Companion app downloads → up 88% YoY

The danger isn’t that people are talking to AI.
It’s that we might forget what real connection tastes like.
And teens never give themselves the chance to learn.

🥦 Human relationships are veggies: nutritious but hard work.
🍭 AI relationships are candy: sweet, easy, and always available.

In today's newsletter, I take you on a tour of the state of AI romance. (Link in comments.)

My take: AI love is on the rise, but it isn’t the disease. It’s a symptom of loneliness, burnout, and the erosion of realworld interation.

The cure? Rebuilding the conditions for human closeness.

That said, the rise in AI partners is a very real signal of something brewing. Enough that Ohio is introducing a bill to ban AI marriage and last week California became the first state to regulate AI companions.

Whatever you think of it, the AI love phenomenon is very real. What are your thoughts? And do you think “AI partners” should be regulated like social media (or more strongly) or treated like personal freedom?

#airomance #ai #aicompanions

Your comments and reposts determine what gets heard on social media. Thank you for engaging!

Using AI well is a modern version of Pascal's Wager.

What's the downside of asking?
You lose a sentence. Maybe a minute.

What's the upside?
You make progress on your biggest questions and problems that used to feel stuck.

For the first time in history, you don't need technical fluency (or permission) to ask for help thinking something through. The limiting factor is no longer the tool. It's whether you're clear on what actually matters.

And because asking is now essentially free, shallow questions aren't a tooling problem. They're a self-leadership one. Your prompt history reflects what you care about, what you're trying to fix, and what you've quietly decided isn't worth improving anymore.

In this episode of AI Advice From Anywhere, I invite you to audit your prompt history and notice what it says about your priorities. I'd love to hear what surprised you. And yes, there's also a city to guess!

If you want to build the skill of asking better questions and making better decisions with AI, that's what I teach here:
👉 decisiongptcourse.com

#AAfA #DecisionIntelligence #AILeadership

How many of these 14 #AI #headlines did you catch recently? Tell me the one you find most interesting in the comments!
 
🔗 For the curious, my newsletter digs into each one: decision.substack.com

🗞️ 1 DeepMind’s new Gemma model uncovers potent cancer therapy pathway

🗞️ 2 AI investment in the US hits $109B (12× China’s and 24x the UK’s) with 78% of firms adopting AI.

🗞️ 3 Anthropic debuts ‘Agent Skills’ for customizable Claude agents

🗞️ 4 China updates AI risk framework beyond content control to open-source misuse, labor impacts, and AI-enabled WMD risks

🤣 5 "Very rude" prompts make ChatGPT more accurate, researchers find

🗞️ 6 Kyndryl’s 2025 Readiness Report: CEOs pour money into AI but can’t scale it

🗞️ 7 AI eye implant restores sight, redefining blindness treatment

🗞️ 8 Over half of new web articles are now AI-generated

🗞️ 9 Google launches Veo 3.1 with major Flow upgrades

🗞️ 10 MIT trains AI to recognize personalized objects in new scenes

🗞️ 11 Virginia Tech debuts AI to map how viral RNA binds to human proteins in 3D.

🗞️ 12 AWS outage in one region disrupts many major online platforms including Coinbase, Canva, Snapchat, Signal, Fortnite, Lyft, Spotify, Roblox, and even Amazon’s own site. H/T Linas Beliūnas for the image below.

🗞️ 13 OpenAI debuts AI-driven Atlas browser, taking aim at Chrome

⭐ 14 Cassie Kozyrkov announces new Decision-Making with ChatGPT course to run tomorrow at 9 AM Eastern Time. Enrollment link: https://lnkd.in/eG7EBtTv (promo codes below).

"A great decision maker isn’t someone who works harder—it’s someone who maximizes their return on effort."

PROMO codes: You have two choices for how you'd like to earn a discount for my course.

1) Subscribe to my newsletter decision.substack.com (free or paid, up to you) to take $200 off. Code: SUBSCRIBERS.

OR

2) Commit to being a champion of the course to take $300 off. Before the course, that means helping the course find its way to at least 5 people who you think would benefit from it. After the course, it means putting the effort in to leaving an honest review on social media. Code: CHAMPIONS

One code per person, your choice.

Here's the enrollment link again: https://lnkd.in/eG7EBtTv

Don't forget to tell me if one of these headlines is more interesting to you than the others so I know what to look for to bring you more tasty things next week. If you found any of these #ai #news updates #useful, words of affirmation are my love language. :) As are ♻️ reposts!

More info on each headline in my newsletter! decision.substack.com