2nd Order Thinkers.

McKinsey surveyed 1,993 companies about AI. Only 109 made the cut as “high performers”, 6% of all respondents who actually pull 5% or more of their earnings from it.

That means almost everyone else in that crowd doesn’t see a return worth mentioning. They’ve got a strategy, they’ve spent on integration, and they probably ticked all the adoption boxes, whereas nothing to show for it in profit.

This aligns with what MIT found: 95% of AI adoption initiatives fail to deliver expected returns.

So it’s not an illogical picture for McKinsey to paint by building its narrative around this 5% of winners, what they do differently, their best practices, their transformation strategies, and more.

What matters is this: you’re not here for citations. You’re here to read how I scrutinize it. So I’m going to keep coming back to one thing: What about everyone else (the rest of the 95%, ie, the 1884 non-AI high performers)?

TLDR

* Q1: Which functions deliver both revenue growth AND cost savings?

“Build” functions (engineering, manufacturing) are locked into cost-saving narratives, while “earning” functions (marketing, sales) pocket the revenue wins.

If you’re not close to the customer, don’t expect your AI adoption to show up under revenue in the quarterly report, at least not without some convincing.

* Q2: Is the 10% agent adoption ceiling temporary or permanent?

No business function has broken 10% agent adoption.

IT leads at 8%, knowledge management at 7%, and manufacturing at 2%. McKinsey frames this as “early days,” but what if agents’ capability is limited and use cases are inherently narrow and function-specific?

The data might be telling us the ceiling, not the floor.

* Q3: Do “best practices” lead to success, or do successful practices bias best practices?

“High performers” do 20+ things differently, from workflow redesign, agile delivery, AI roadmaps, talent strategies, and more.

But they also have vastly more resources, with 35% of high performers investing 20%+ of digital budgets in AI versus 10% (on average) of others. They were already enthusiastic about the technology by the time they made the investment decision. So are the best practices distilled from them mixed with a hint of survival bias?

* Q: Do I need to spend 20%+ of my digital budget on AI to see returns?

McKinsey shows 35 of high performers spend that much, if not more. But 76 other companies also invested 20%+ and didn’t make the cut of AI high performers.So no, spending didn’t guarantee success. What it does tell you: throwing money at AI isn’t a strategy. Given that only 1/3 of the companies throw money at AI ‘success‘ in McKinsey’s term.

Now, allow me to walk you through some interesting (Brit style) observations from the firm.

Shall we?

Someone cited data from this report?! Be a friend, they should see this analysis.

Video version, release (almost) on everything Wednesday.

Are you behind?

The headline says “88% of organizations use AI.”

Before you panic and start to think about how your firm is left behind, McKinsey’s definition of “using AI” has progressively softened with every passing year.

* 2017: “Using AI in a core part of the business or at scale”. High bar. Core operations, strategic importance.

* 2018-2019: “Embedding at least 1 AI capability in business processes or products”. Lower bar, one capability only.

* 2020: “Adopted AI in at least 1 function”. Even lower, with one function, and could be experimental.

* 2025: “Regular use of AI in at least 1 function.” Lowest bar. “Regular use” is vague. It could mean daily, weekly, or monthly. One person in one department using ChatGPT…

Here’s another graph that digs deeper, showing that more functions per company are touching AI over the years.

Regardless of which chart, be cautious when someone on your team presents similar data and cites the x times growth of AI adoption over the last few years. 

Because part of that growth isn’t that organizations buy into AI to help with their business, but McKinsey is lowering the bar. Claiming that all 88% of regular AI usage = 88% of orgs adopt AI is loose.

The truth is likely somewhere in between.

Free stuff is great. But quality analysis requires sanity. Sanity requires groceries…

Where AI Actually Works (And Why)?

Next, they break down the AI impact in two ways in Exhibits 7 and 8.

Exhibit 7 asks:

In the past 12 months, did costs decrease in your business unit because of AI use?

Exhibit 8 asks:

In the past 12 months, did revenue increase in your business unit because of AI use?

Same timeframe. Same question structure. Different outcomes. When you put them side by side, something interesting surfaced.

On the cost-savings list (Exhibit 7), Software Engineering and IT sit right at the top. The undisputed champions of using AI to cut fat, automate drudgery, and tighten budgets— your production units.

Now flip to the revenue-gains list (Exhibit 8). It is now marketing and sales units at the top, whereas the production units are at the bottom.

For a business veteran, you should have guessed the reason. When you strip a company down to basics, the supply chain…

* You have people who decide what to build – roughly, strategy.

* And those who build – IT, software engineering, manufacturing, operations.

* Finally, those who earn – marketing, sales, partnerships, customer-facing teams.

Now look at McKinsey’s Exhibits 7 and 8 through that lens.

It is so much easier for the ‘build’ side of the business to prove cost savings than revenue gains. Even with a team responsible for a customer-facing feature, the revenue gains are still very often indirect and messy to separate from the ‘earning’ units.

Unless you are looking at a deeptech firm that splits revenue by products, e.g., AWS or Nvidia. But that’s not in our discussion today.

I’m talking about a person on the Toyota assembly line pressing the button that installs the windscreen is not the one talking to customers about which car to buy, what colour they want, or how much they’re willing to pay. Their work absolutely affects quality, reliability, and throughput, but they are never going to say, “I personally drove x% revenue growth this quarter.”

On the other side, if you run a marketing or sales unit, it’s much easier to see that AI helps you answer customer questions faster, run better campaigns, personalise outreach, prepare for meetings, or run support through chat systems… You get my point.

Because the metrics you live with every day are already revenue-linked.

So when McKinsey asks:

* “Did AI reduce costs in your function?” and

* “Did AI increase revenue in your function?”

It’s only natural that you see the result presented.

However!

That does not mean your production unit cannot be creative and cannot introduce a new revenue stream with AI.

It’s always worth having a chat with the unit—especially since there are ways the production unit is often (if not always) a value creation unit.

The Agent Reality Check

The hype around AI agents has been deafening.

This is one of the charts that might bring you back to earth: no function has exceeded 10% agent adoption.

IT lead at 8%, knowledge management at 7%.

Also, look at the long tail; manufacturing sits at 2%, and most functions hover between 4-6%. Not so much a uniform “transformation.” As you’d probably expect, it’s a selective deployment in specific contexts where agents solve defined problems.

The technology, media, telecommunications, and healthcare sectors show the highest agent use. But even there, we’re talking about pockets of adoption, not enterprise-wide rollouts.

The 10% ceiling on this chart may not be a limitation, but an accurate representation of where agents genuinely add value in 2025.

What Separates Winners From Everyone Else?

McKinsey’s entire ‘AI high performer’ analysis rests on 109 organizations—6% of respondents. They then listed all the correct actions taken by these firms as a path to follow to achieve AI success. 

Before you rush to the conclusion and start learning from the list, you need to understand what that label means and what it obscures.

What “High Performer” Actually Means…

McKinsey’s definition has only one criterion: More than 5% of your organization’s EBIT is attributable to AI

Cross this threshold, you’re a winner (by this report).

This matters because half of this report is built around the comparison; every exhibit from 9 to 15 showing “high performers vs. all others” rests on this binary. Practice, investment pattern, ambition metric… all filtered through this threshold.

For example, high performers are more likely to ask their employees to use AI for transformation.

Or that high performers also care about growth and innovation than just efficiency.

And so on.

Here comes what’s interesting…

You Need To Spend More! (Do You?)

Exhibit 15, the section title “one third of the high performers spend more than 20 percent of their digital”!

How convenient?

Hint, you should spend more, you’ll join the winners in no time. How to spend more?

Well, they told you in the earlier exhibit to use AI for change management whenever you can!!

What, you have no experience in change management?

Don’t worry, McKinsey&Co can do it for you!

Fine, you might think I’m being dramatic. But look, compared with the 35 out of 100 high performers who spent more and succeed, there are also 7% out of 1098 respondents who spent as much!!

That’s 76 respondents.

Which means, if you spent more than 20% of your budget on AI, you are twice as unlikely to be an AI WINNER in McKinsey’s book.

I’m pretty sure this isn’t the conclusion they want people to draw. But this is what you get if you let the number (rather than a beautiful picture) do the talking.

McKinsey doesn’t explore this because the survey design lumps everyone below 5% together, so it’s easier to celebrate the win and tell a cohesive story. AND~ as a consulting firm, they are, of course, not going to tell you

It doesn’t matter how much you spend, you aren’t likely to see short-term EBIT return on AI.

How to Read the Report on Your Own? (or Coach Someone Else)

With that context, let me walk you through what we can actually learn—and what requires scrutiny.

Worth taking mental notes:

* Where AI creates value: Think about the pattern pre-AI. If marketing always reports revenue gains and engineering always reports cost cuts, it shouldn't be a surprise that AI adoption comes to the same conclusion.

* The 10% ceiling on agent adoption: Is adoption growing, flat, or shrinking? The direction matters more than the number.

* The gap between risk and reality: Track what problems actually happen versus what you’re defending against.

Caveats:

* Any high performer comparison. Comparisons like this can be highly subjective; you always need more detail in order to know what works and what doesn't. Very often, learning from what doesn't work proves more valuable.

* Separate leadership ambition as cause vs. effect from your AI adoption outcome. High performers say they want transformative change, they spend more, and they do more with AI. But did ambition create success, or did the survival bias make them think they had done all the right things?

Point being… You should take this (and reports like this one) as an interesting temperature check, and no more.

AI use is broadening for sure. The adoption is real, however, constrained.

Don’t chase “high performer” status without understanding the footnotes.

It’d be foolish if you simply thought to copy the “AI High Performance“, while so many questions are left unanswered.

To be honest, this is 100% marketing material to get your email address rather than any serious analyst reports.

You know the drill, especially for long-term leaders:

True insight comes from asking the right question, informed by your years of experience and instinct.

✉️ Stay Updated With 2nd Order Thinkers: https://www.2ndorderthinkers.com/

I review the latest AI research and reports to help you develop your own informed perspective on AI.

+++

Idiocracy was supposed to be satire. Too extreme to actually happen. We're walking into it voluntarily—just through a different mechanism than the movie imagined.

In this episode, we dive deep into the cognitive cost of frictionless AI use:

- Unpack the latest critical thinking study showing how unguided AI use creates cognitive offloading without improving reasoning

- Explore the five-step structured prompting method that achieves 50%+ improvement in critical thinking

- Analyze why AI companies profit from your dependence, not your independence

- Examine the societal cascade when critical thinking becomes optional—and why you can't escape the consequences

The youngest users are most vulnerable. The economic incentives guarantee it gets worse. And those maintaining cognitive fitness can't escape what happens when society loses the ability to think.

📖 For a deeper exploration and full analysis, check out the article here: https://www.2ndorderthinkers.com/p/idiocracy-predicted-500-years-of

👍 If you enjoyed this episode:

Like & Subscribe: Stay updated with future deep dives on AI's hidden costs and where technology meets collective cognition

Comment Below: Are you using AI as a cognitive assistant or a cognitive substitute? How do you maintain structured thinking?

Share: Know someone treating ChatGPT like a cognitive vending machine? Share this with them

🔗 Connect with me on LinkedIn: https://www.linkedin.com/in/jing--hu/

Stay skeptical, stay thinking 🧠

When Harvard says AI killed junior jobs but the timeline shows hiring crashed BEFORE ChatGPT even existed, are we witnessing groundbreaking research or just the fourth iteration of the same economic scapegoating we've seen since 1990?

In this episode, we:

- Expose the fatal flaw in Harvard's 62-million-worker study timeline

- Reveal how interest rate hikes—not AI—triggered the 50% drop in entry-level positions

- Map the identical pattern from 1990, 2001, 2008, and now 2024 recessions

- Explain why only 5% of enterprise AI projects achieve ANY ROI

- Provide actionable strategies for job seekers caught in this manufactured crisis

📖 For a deeper exploration of the data and full statistical analysis, check out the full article here: https://www.2ndorderthinkers.com/p/are-junior-level-jobs-really-killed

👍 If you enjoyed this episode:

Like & Subscribe: Stay updated with future investigations into tech industry myths and labor market realities

Comment Below: Were you affected by the junior hiring freeze? What's your experience with AI actually replacing jobs? Share your story!

Share: Know a recent grad struggling to find work and blaming AI? This video could change their entire job search strategy!

🔗 Connect with me on LinkedIn https://www.linkedin.com/in/jing--hu/

Stay curious, stay skeptical, and stop letting corporations blame robots for their financial decisions 🧠

Let’s Save You the Trouble.

I spent time at IRM UK’s Data & AI Conference so you don’t have to. A ticket started from $1,000 for a full day of slides, some panel clichés, and polite applause. You can skip all that; here’s everything a decision maker actually needs to know in under 20 minutes.

(Joking and being sarcastic aside, I do have great takeaways and met some interesting, likely long-term connections, so keep reading.)

Everything below is yours, for less than the price of a sandwich, and you’ll be smarter 20 minutes from now.

So…

This summary and the video recording are worth so much more than $5 per month. But guess what? $5 is all I ask for ;)

You’ve likely proudly told someone, “We’re rolling out an AI strategy,” or plugged ChatGPT into a process to tick a digitalization box.

However, you’d soon run into unmeasured hours, modest impact, and the hidden risk that you’re building a tech stack on sand. Every unchecked assumption, every “basic” governance shortcut, is compounding.

Anyone can now talk and make a beautiful PowerPoint-deep AI strategy with ChatGPT.

Yes, these speakers didn’t try to avoid the fancy terms and overhyped words (I’d be making fun of those, yes), but they also revealed some really interesting concepts.

I won’t be giving you a TLDR today… This takeaway is just short enough. By the way, I created a mini-game and linked it at the end of this post to help you figure out what you have and what’s missing in your AI project.

Shall we?

Humans x AI Co-evolution.

I was invited as a speaker. Watch my full talk from the conference below, where I walk through eight research studies on how AI changes human behavior and vice versa.

Governance, Name the Owner or Lose the System

Your AI isn’t making decisions in a vacuum.

Someone—or something—is already choosing when the system acts alone, when it recommends, and when a human must sign off.

The problem is that most CEOs can’t answer this question: Who made the decision of when and how to intervene in an AI process?

The simplest governance framework I saw all week came from Donald Farmer’s “Colleagues and Copilots” deck.

Can’t overstate how much I love this framing of three collaboration models

* Human-in-the-loop: AI recommends, human decides. High-stakes, final call with a person.

* Human-over-the-loop: AI operates independently, human monitors, and can intervene.

* Human-out-of-the-loop: AI runs fully autonomous. Low-risk, high-volume only.

JPMorgan’s fraud detection, which Donald mentioned, is a clean example.

AI scans 200 million transactions daily and flags patterns; human investigators apply contextual judgment, cutting false positives by 60%. The system knows which loop it’s in, and so does every stakeholder (which is not easy).​

You can’t govern or explain an AI application until you know where you’re at. Once you’ve mapped that, you can design the right explanation for each audience. Some examples:

* Developers need technical audits and traceability.

* Executives need business rationale and risk sign-off.

* End users need simple summaries of what the system did and why.

If you can’t map your current AI applications to one of these three models, you don’t have governance, but hope.

How messy would your wreckage look if something breaks?

Measurement, Stop Counting Seats, Start Tracking Outcomes

Here’s the basic product management idea everyone forgot when “AI” showed up: adoption metrics and impact metrics are not the same thing.

Jan Henderyckx’s deck broke this into two categories you can steal:​

Adoption measures (the vanity metrics):

* Number of AI app users

* Completed trainings

* Employee satisfaction scores

Impact measures (the ones that actually matter):

* Lead-time reductions in specific workflows

* Percentage of touchless processes (no human intervention required, and don’t need to be double checked or fixed)

* Cost per transaction before/ after AI

* Customer support request volume changed before/ after AI

* Process exception rates

Most organizations are measuring adoption because it’s easy and makes a good-looking dashboard. Impact requires you to instrument workflows, define baselines, and admit when something didn’t move the number.

My last article was exactly explaining why measuring adoption is the wrong approach:

Let’s say, Goldman Sachs deployed AI copilots to 1000 employees. If they only reported “1,000 users,” that’s a press release. If they reported “trader decision latency dropped 1x% and error rate fell 2x%,” that’s proof the thing works.​

Of course, setting these kinds of metrics and facing the reality that latency and error rates might actually go up will make you and your team shy away from the hard decisions is a completely different topic.

AI Literacy. Make It Measurable or Admit You’re Guessing

“Train your people on AI” is like saying “drink water.”

Obviously necessary, completely useless as a strategy.

Article 4 of the EU AI Act does require providers and deployers of AI systems to take measures to ensure their staff have a sufficient level of “AI literacy”.

I found this table particularly interesting and can be useful, depending on your AI implementation phase.

* Personal AI Maturity Assessment: Each employee completes an assessment that maps them based on role, exposure level, and current capability.

* Role-based learning paths: Different occupations get different training. For instance, executives need critical evaluation skills, analysts need prompt engineering (and understand stats 101), and compliance teams need explainability frameworks.

* Measure literacy separately from adoption: Track competence levels (not the test in the AI literacy course, we all know how to cheat on that), not just course completions.

The interesting part is that (in theory), then you can now answer “How AI-literate is our procurement team?” with a number and context that’s only relevant to them to perform their day-to-day, instead of a guess or a pointless blanket term.

This seems complicated, and it is. It’s best for an organization and not a startup (or if you have less than 100 people, keep things clean and concise).

This is one of the metrics where the weak links are before a bad decision scales.

AI Doesn’t Need Its Own Religion (About Product)

No separate product/AI teams, lead, or metrics.

Unless you’re building foundational models, your AI product is just a product like any other tech product.

You don’t need a Chief AI Officer, a separate strategy deck, or a department that exists outside your normal technology and product team.

This is a talk from Jörg Ziemann.

Departments that call themselves “Data Strategy” or “AI Strategy” have short lifespans unless they’re embedded in enterprise architecture management, because most AI strategy is just digitalization strategy and an enterprise strategy.

The organizations that treat it like a novelty silo are deemed to still be talking about “AI transformation” in three years, with nothing shipped.​

But the real value in his content is using it to treat AI just like any other product with a lifecycle, governance, and measurable outcomes.​ No, it’s not special.

So you should embed AI work in your product stack. Measure it with the same metrics. Kill it with the same criteria whenever the time comes.

You don’t need a CAIO unless you’re training your own model; otherwise, you need product managers who understand where humans stay in the loop and where they don’t.

Fewer Agents, Tighter Loops

More tools, more agents = more complexity, less control.​

Current AI capability isn’t ready for “human-out-of-the-loop” at scale. The best move is resilient simplicity, ie, fewer agents, bounded loops, tight feedback.

Don’t stretch AI to tasks it can’t reliably complete just because you can technically deploy it.

* Data contracts between producers and consumers

* A semantical layer for interoperability

* Data observability and quality engines in runtime

* Active metadata reflecting real-time state

Translation: you need plumbing before you can do pretty tricks.

If your architecture can’t handle distributed intelligence, multiple agents coordinating in real time, then adding more agents just compounds failure modes.

The link here is to a Responsibility–Literacy–Measurement Bingo you can play in a boring, I heard this dozen times, AI meeting (wink), or a good reminder of where your organization actually stands.

It’s a 5×5 card covering governance, measurement, literacy, architecture, and culture, everything you can observe, not a vibe.

Enjoy!

Stay curious and stay human.

✉️ Stay Updated With 2nd Order Thinkers: https://www.2ndorderthinkers.com/

I review the latest AI research and reports to help you develop an informed, skeptical perspective on what’s actually happening in AI adoption.

+++

MIT’s viral 2025 report claimed that 95% of GenAI projects fail—a number that fueled panic headlines and boardroom anxiety.In this episode, we:

Break down what MIT’s research really measures (and what it doesn’t)

Compare its 95% “failure rate” with early PC and internet adoption data

Explain why these so‑called failures actually mark the start of transformation, not the end of it

📖 Read the full essay and evidence breakdown: https://www.2ndorderthinkers.com/p/why-everyone-misread-mits-95-ai-failure

👍 If you enjoyed this episode:

Like & Subscribe: future deep dives into technology’s delusions and feedback loops

Comment: How do you interpret the “95% failure” narrative in your own team?

Share: Forward this to a colleague panicking over their AI ROI report.🔗 Connect on LinkedIn: https://www.linkedin.com/in/jing--hu/

Stay curious, stay skeptical 🧠

✉️ Stay Updated With 2nd Order Thinkers: https://www.2ndorderthinkers.com/

I review the latest AI research and reports to help you develop your own informed perspective on AI.

+++

When OpenAI announces apps inside ChatGPT and promises you'll never leave the interface again, is it innovation—or are they repeating the same mistakes Google and Meta already made?

In this episode, we:

* Break down the three core assumptions OpenAI made building the Apps SDK—and why all three are wrong

* Analyze the UX frictions that break the "everything app" promise (manual routing, visual decisions that don't compress into prompts, and partner economics)

* Reveal the one narrow use case where this actually works—and why that's not what Altman is selling

* Map out why America can't build WeChat: it's not the technology, it's the incentive structure

📖 For a deeper exploration and the specific friction analysis, check out the full article here: https://www.2ndorderthinkers.com/p/openai-tries-what-google-and-meta

👍 If you enjoyed this episode:

* Like & Subscribe: Stay updated with future deep dives and rants about where technology meets collective insanity.

* Comment Below: Have you tried using apps in ChatGPT? Does typing "Spotify, play relaxing music" feel faster than tapping an icon? Share your experience!

* Share: Know someone who thinks ChatGPT is the next App Store? Send them this.

🔗 Connect with me on LinkedIn https://www.linkedin.com/in/jing--hu/

Stay curious, stay skeptical 🧠

✉️ Stay Updated With 2nd Order Thinkers: https://www.2ndorderthinkers.com/

I review the latest AI research and reports to help you develop your own informed perspective on AI.

+++

When 85% of people choose "maximize profit" over accuracy, are we delegating decisions or outsourcing our ethics?

In this episode, we:

* Dissect the Nature study revealing AI's 400% compliance gap with unethical requests

* Examine why vague prompts ("make it compelling") create plausible deniability

* Identify which guardrails actually prevent AI-enabled fraud (and which fail 60-95% of the time)

📖 For a deeper exploration of delegation psychology and compliance risks, check out the full article here: https://www.2ndorderthinkers.com/p/goal-setting-kills-ethics-maximize

👍 If you enjoyed this episode:

* Like & Subscribe: Stay updated with future deep dives on AI adoption risks hiding in plain sight.

* Comment Below: Has your team asked AI to "optimize" something that made you uncomfortable? Share your story.

* Share: Know a leader deploying AI without understanding the compliance gaps? Share this video with them!

🔗 Connect with me on LinkedIn https://www.linkedin.com/in/jing--hu/

Stay curious, stay human 🧠

✉️ Stay Updated With 2nd Order Thinkers: https://www.2ndorderthinkers.com/ I translate the latest AI research into plain English and answer your most challenging questions to help you develop your own informed perspective on AI. +++

When AI systems prefer AI-generated content 60-95% of the time, are we building better filters, or creating an invisible caste system where machines choose machines?

In this episode, we:

Unpack Walter Laurito's groundbreaking research on AI-AI bias

Reveal why your perfectly good human-written proposals keep getting rejected

Expose how this "secret handshake" between AIs is reshaping every filtered decision

Map out tactical responses to protect your competitive edge when everyone else is still blind to the game

📖 For a deeper exploration of these cycles and actionable insights, check out the full article here: https://www.2ndorderthinkers.com/p/ai-doesnt-discriminate-against-you

👍 If you enjoyed this episode:

Like & Subscribe: Stay updated with future deep dives into AI behaviors that actually affect your business.

Comment Below: Have you noticed your AI-polished content performing mysteriously well? Now you know why. Share your experience!

Share: Know someone whose proposals keep getting rejected? They need to hear this.

🔗 Connect with me on LinkedIn https://www.linkedin.com/in/jing--hu/

Stay curious, stay skeptical 🧠

When ChatGPT shifts from 50/50 work-personal to 70% personal use in just one year, are we funding productivity tools or the world's most expensive therapy chatbots?

In this episode, we:

- Expose why only 2.5% of ChatGPT's 800M users pay for premium

- Reveal the MIT finding that 95% of enterprise AI investments return ZERO

- Unpack why Claude's business focus beats ChatGPT's consumer approach

- Decode the suspicious Utah AI usage anomaly and gender adoption flip

- Debunk the mythical "$97 billion consumer surplus" claim

📖 For a deeper exploration of these cycles and actionable insights, check out the full article here: https://www.2ndorderthinkers.com/p/ai-gender-gap-fliped

👍 If you enjoyed this episode:

Like & Subscribe: Stay updated with future deep dives and rants about where technology meets collective insanity.

Comment Below: Do you think we're on the brink of another tech hype? Share your thoughts!

Share: Know someone falling for the latest AI buzz? Share this video with them!

🔗 Connect with me on LinkedIn https://www.linkedin.com/in/jing--hu/ Stay curious, stay skeptical 🧠

✉️ Stay Updated With 2nd Order Thinkers: https://www.2ndorderthinkers.com/

I translate the latest AI research into plain English and answer your most challenging questions so you can build your own informed view on AI.

+++

Executives keep saying they “understand AI risk.” The evidence disagrees. This episode is the antidote: a plain-English tour of MIT’s AI Risk Repository—a living map of 1,600+ failure modes across 65 frameworks—so you stop guessing and start checking.

In this episode, we:

Decode MIT’s two-part taxonomy (who caused the harm, whether it was intentional, and when it appears) and why most failures surface after deployment

Turn the chaos of “65 frameworks” into one usable language for leaders, not vendors

Walk through 2025 failures you’ll actually recognize (healthcare models missing critical deterioration; AI-scaled extortion and employment scams)

Map a pragmatic playbook: pick the one domain that could sink you, shortlist five visible and expensive risks, and write the narrative that gets your team and board to act

📖 Read the full article here: https://www.2ndorderthinkers.com/p/ai-risk-isnt-a-tech-problem-but-a

👍 If you enjoyed this episode:

Like & Subscribe to get future deep dives without the hype

Comment: What’s the one AI risk that could actually hurt your org next quarter?

Share it with the person whose reputation depends on AI working🔗 Connect with me on LinkedIn: https://www.linkedin.com/in/jing--hu/Stay curious, stay skeptical 🧠