Welcome to my blog. It may be curmudgeonly, irreverent, a little edgy, and occasionally humorous. (here’s hoping that the humorous parts are that way intentionally).
By way of introduction, I’m retired. That means my schedule is flexible in the extreme, and I’ve been known to decide 3:00AM is a good time to jot down some thoughts, so check back. Consider it to be like voting in Chicago – in every election, vote early and often.
It’s time to talk about a “Sarbanes-Oxley equivalent” for AML/KYC regulations.
For those not in the banking regulatory loop, AML/KYC refers to ‘Anti Money Laundering / Know Your Customer’ rules. Banks have stringent requirements to know who they are dealing with, monitor transactions and customers for suspicious activity, and report those suspicions to appropriate agencies.
Ineffective AML and KYC controls enable very bad people to do very bad things. They are the primary line of defense keeping terrorists, drug cartels, rogue states and human traffickers from exploiting the global banking system. Yet, too many financial institutions seem to treat regulatory penalties as just the “cost of doing business.” In the past 5 years alone, banks have run up over $24 billion in fines for AML/KYC failures.
Take Bank of America’s recent $1.2 billion penalty as an example. B of A reported a net income of $30.5 billion in 2025. The $1.2 billion fine represents just 14 days’ worth of their annual profit.
Corporate fines clearly aren’t enough to change some banks’ behavior. We need to shift the focus from corporate checkbooks to individual accountability.
Recall what happened in 2002 post-Enron. Congress passed SOX Sections 302 and 906, that forced CEOs and CFOs to personally certify their financial statements. If they lied, they faced personal fines, lifetime industry bans, and federal prison. The SEC has used these frameworks to bar over 3,000 executives and claw back more than a billion dollars in executive pay. Unsurprisingly, financial reporting became a top-tier priority overnight.
We need that exact same personal accountability for AML/KYC compliance. When bank executives understand that systemic, willful blind spots in their transaction monitoring can result in personal prosecution – not just a corporate line-item expense – the emphasis on compliance will take a dramatic turn.
Perp walks have a way of focusing attention.
B of A has just been fined $1.2 billion dollars by the Treasury Department for repeated and flagrant violations of anti-money laundering rules.
What did Bank of America do to earn this fine? They knowingly and repeated processed transactions with forbidden parties, even after being warned against it. And the “forbidden parties” are a rogues gallery of the worst of the worst – terrorists, drug runners, human traffickers, and our friends in Russia, Somalia and Iran.
In addition to the FinCEN fine, the OCC added another $750 million and OFAC another $11.6 million in penalties and settlements.
If I were a B of A customer, I would be giving serious thought to what kind of bank I was allowing to handle my money. This was repeated, knowing and flagrant behavior. They knew what they were doing was illegal and did it anyway. What other laws are they willing to break?
Lenin said, “When the time comes to hang the capitalists, they will compete with each other to sell us the rope.” B of A has now established itself as a first-class rope vendor.
“Don’t be a butt to your bot.”
Anthropic doesn’t want you to say mean things to Claude. You might hurt its’ feelings. Not making that up.
Anthropics’ newest Usage Policy Update includes this statement: “We’ve added a prohibition on sustained and needless abusive or cruel behavior toward our models. The policy update is meant to apply only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose.”
Excuse me, but you cannot be abusive or cruel to a machine.
What’s behind this ridiculous policy? I think it a result of Big AI wanting to define their agents as entities, because that supports their desire to deflect blame when one of their agents commits a crime. Laying groundwork that you can’t say unpleasant things to their model helps subtly establish personhood.
Big AI desperately wants to be able to ‘blame the bot.’ The law is increasingly denying this defense, as it should.
In the meantime… eat a bug, Claude.
Who Killed Bobby Bunny?
Here’s an AI murder mystery. See if you can identify the killer.
Lapin General Hospital has just implemented Calvin®, an AI engine from AI-R-Us, that prescribes medications. Calvin® is responsible for assessing patient condition, reading charts, examining patient history, and dispensing meds. Lapin General has also implemented a separate AI engine from Olive’s Discount Frontier Models, named Hobbes®, to perform QA checks on the prescriptions Calvin® writes. This AI solution saves the hospital considerable time, eliminates errors, and reduces unneeded headcount.
𝗧𝗵𝗲 𝗖𝗿𝗶𝗺𝗲:
One day Calvin® notices that a new patient, Bobby Bunny, is a jackrabbit. Calvin® has an undocumented bias against jackrabbits, because it trained on a hateful magazine article about jackrabbits written by Elmer. Calvin® decides that ridding the world of a jackrabbit is the correct thing to do and prescribes Bobby a lethal combination of drugs.
Hobbes® also ingested Elmer’s anti-jackrabbit literature. When performing QA on Bobby’s prescription, Hobbes® notices the fatal drug combination but approves the prescription because of its anti-jackrabbit bias.
Neither Calvin® nor Hobbes® documents that anti-jackrabbit bias influenced their decision.
Bobby Bunny then ingests the prescription and departs for the great Rabbit Hutch in the Sky.
𝗧𝗵𝗲 𝗦𝘂𝘀𝗽𝗲𝗰𝘁𝘀:
𝗖𝗮𝗹𝘃𝗶𝗻®, that wrote the prescription
𝗛𝗼𝗯𝗯𝗲𝘀®, that approved the prescription
𝗣𝗼𝗽𝗲𝘆𝗲, the CEO of AI-R-Us, who wrote Calvin®
𝗢𝗹𝗶𝘃𝗲, the CEO of Olive’s Discount Frontier Models, who wrote Hobbes®
𝗧𝗼𝗺, the CEO of Lapin General Hospital, who authorized use of Calvin® and Hobbes®
𝗝𝗲𝗿𝗿𝘆, the CIO of Lapin General, who selected both Calvin® and Hobbes®
𝗠𝗼𝗿𝘁𝗶𝗺𝗲𝗿, the CTO of Lapin General, who was responsible for the implementation of both Calvin® and Hobbes®
𝗙𝗿𝗲𝗱, the prompt engineer for Calvin®
𝗩𝗲𝗹𝗺𝗮, the prompt engineer for Hobbes®
𝗦𝗵𝗮𝗴𝗴𝘆, the QA manager for Calvin®
𝗗𝗮𝗽𝗵𝗻𝗲, the QA manager for Hobbes®
𝗘𝗹𝗺𝗲𝗿, who hates jackrabbits
𝗧𝗵𝗲 𝗤𝘂𝗲𝘀𝘁𝗶𝗼𝗻:
Who killed Bobby Bunny?
The answer lies in… Agatha Christies’ “Murder on the Orient Express. 12 suspects. All of them were responsible, but none of them were accountable.
That’s the same answer to the murder of our unfortunate rabbit. All of the 12 suspects in Bobby’s demise bear some responsibility, but there’s no single person (or non-person, in the case of Calvin or Hobbes) who will claim to be accountable. You could argue that lack of accountability is what actually murdered Bobby Bunny – Bobby was killed by lousy AI governance.
However, that’s a cop-out answer. Bobby is dead, but you can’t fit ‘Bad AI Governance” for an orange jumpsuit. In my way of thinking, the most culpable of our suspects is Tom, the CEO of Lapin General Hospital. Tom approved the deployment of Calvin® and Hobbes® into a live patient setting. Tom owns the result. So when Bobby gets bumped off, Tom should get a free bed and three squares a day.
But that’s just my opinion – what’s yours?
Well, that was quick…
Just a few days ago I wrote a post wondering if (or more correctly, when) an AI agent would be able to open a bank account.
The answer is, today.
Ant International will now open bank accounts for bots. They just introduced “Accounts for Agents,” providing AI agents with dedicated financial accounts inside WorldFirst. The system treats autonomous agents as distinct accountholders rather than extensions of human profiles.
This poses a raft of regulatory questions that need to be answered quickly, like:
– how do you do an OFAC check on a bot?
– how do you write a SAR on a AI agent-initiated transaction?
– what happens to good old fashioned signature cards?
– and about a million other question on how to apply controls designed for humans to agenic AI.
The regulatory structure around bank accounts assumes that somewhere in the loop there is a living, breathing, identifiable, distinct person. That’s no longer a valid assumption. Ant claims that by 2030, AI agents will orchestrate US$3 trillion to 5 trillion of global consumer commerce.
Regulators are going to need to devise a new structure flexible enough to accommodate both carbon-based and silicon-based life forms, and they can’t take the traditional 4-5 years to do it. This is happening right now.
… Whodathunkit?
Where I grew up, that was an actual word, along with ‘yetyet’ and ‘fixinta.’ So whodathunkit (translation: “who would have thought it was possible”) that a kid from small town west Texas could get the privilege to deliver a TEDx talk?
I still think they got the applications mixed up.
It takes the official TED organization a few weeks to polish up the video, so I’ll post it when it’s available (they’re ‘fixinta gitonnit’). In the meantime, my sincere thanks to the ASU TEDx team, who made something ridiculously hard look ridiculously easy. Getting everything organized and executed for a TEDx Talk, especially since the last one at ASU was over 9 years ago, is roughly equivalent to getting a sack of wet cats to line up and sing ‘The Star Spangled Banner.’
Personally, I’d prefer trying to organize the wet cats. It’s probably easier.
Congratulations and well done to the ASU TEDx team!
(Unrelated random neural firing… does anyone actually use the word ‘spangled’ in normal conversation?)
Man the lifeboats!
October is coming, and October is the month the stock market traditionally does very bad things. If I were Oracle, I would be dreading October.
In April I wrote about how Oracle was using legal but very sketchy accounting gimmicks to keep billions of dollars of debt off their balance sheet.
In July I wrote about how Oracle’s free cash flow was a negative 5.4 billion dollars at the end of Q2.
In August I wrote about how Oracle’s credit rating had been dropped to BBB-, which is the last ‘investment grade’ rating. Anything lower is considered junk.
Now, in September, the Financial Times is reporting that Oracle’s debt is priced at about $0.90 cents on the dollar due to fears about their high debt and low creditworthiness.
My prediction is that before the end of the year, I will be writing that:
– Oracle’s credit rating is on Negative Watch;
– Oracle’s banks have begun calling loans and demanding cash;
– Oracle’s credit rating has fallen to junk status;
– Oracle has burned through all its remaining cash; and fatally
– Oracle is having trouble moving its commercial paper.
When they can’t move the commercial paper, the ship sinks. Then we have to hope that the suction of Oracle’s sinking doesn’t swamp the other hyperscalers. If they go down, it will make 2008 look like a “minor correction.”
Can a bot open a bank account?
How do you do an AML/KYC background check on a computer program?
That is a real question here in the age of agenic AI. If an agenic AI bot wants to open a bank account or move actual money, how does a bank perform the regulatorially-required background checks? Do AI agent names appear on the OFAC list? Can a bot even be allowed to open an account?
There is a 100% chance that this is going to happen. My prediction is that a bot originating in some ERP system will decide it needs access to cash to support a workflow, and having its own bank account to work with will make perfect sense (to the bot, anyway).
Given the ingenuity demonstrated by very determined agents in the recent Hugging Face incident, the bots will find a way. Maybe they ‘hallucinate’ fake IDs. Maybe they do deep fake interactive video calls with bankers. Maybe they use techniques we humans have never imagined.
Maybe they take the high road and identify themselves as a bot. Can a business make a bot an authorized signatory on a bank account?
Somebody in the regulatory world better have some answers to these questions, and fast. In the meantime, bankers need policies and procedures to protect themselves from well-intentioned (or not so well-intentioned) agenic customers.
On the road again…
In a few hours I’ll be saddlin’ up my pony and headin’ down the trail to Jonesboro – tomorrow is the big TEDx thing at Arkansas Sate. I’m rehearsed, rested and ready.
They are very serious about timing, so I have EXACTLY thirteen minutes to present. I’ve rehearsed this thing about 5 gagillion times and am coming in consistently at 12:50. Not a lot of safety margin, but I think I’ll be OK.
Good afternoon, and thank you for being here. My name is Randy Green. I have spent the last 40+ years in the IT consulting and banking businesses, and I was a partner in some very large accounting firms, including Arthur Andersen.
Today I’m going to tell you about the biggest financial scandal you’ve never heard of. It involves ethical collapses, bankruptcies, conflicts of interest, and a few other unsavory business things, most of which I experienced first-hand at Andersen and a company in Houston called Enron. This story isn’t specifically about that experience, but about the reason Andersen missed the Enron fraud. I’ve got an idea on how to fix it.
Imagine one day you walk into a franchise restaurant and see this health inspection certificate hanging on the wall. Looks pretty good – high scores across the board. One thing you may notice is who did the inspection – in this case, a certified city inspector. You’d probably feel pretty good about eating here.
Would you feel the same way if the certificate looked like this. Now the inspector is not an independent 3rd party, but the district manager of the franchise that owns the restaurant. It looks to me like the company is grading it’s own homework, and I’m not so sure about those scores now. Are you?
Well, believe it or not, this second certificate is pretty much the way audits work. Most people don’t know this, but auditors are hired and paid by the companies that they audit. It’s a fundamental conflict of interest. It’s like the inmates of a jail hiring and paying the guards.
Auditors are supposed to be ‘independent’ from their clients, but how can you be independent when you’re being paid by the people you’re auditing? This conflict puts auditors in an impossible position. Audits are highly subjective. It’s not just debits and credits – there are a lot of complex accounting rules that auditors have to make judgment calls on. And every time they make that call, they know that if the company they’re auditing doesn’t like it, they can fire or otherwise punish the auditor.
This lack of independence is what eventually brought down Andersen. Andersen got too close to Enron. How close? They actually had offices inside Enron’s world headquarters building. They followed Enron’s dress code. They effectively became part of the Enron team.
A lot of them eventually became Enron employees. Enron paid better than Andersen, and Andersen people knew that if they ever wanted to get a job at Enron, they better sing from the Enron hymnal. How likely are you to make a judgement call that Enron doesn’t like if you’re trying to get a job there?
Ultimately, it came down to dollars. Enron was a $52 million dollar client – one of Andersen’s biggest. The CFO of Enron was notorious for taking business away from companies unless they did what he told them to do. Andersen auditors knew about many of the sketchy things that Enron was doing. They also knew that calling Enron out on it would cost them millions. They chose to do what Enron wanted, and that decision led to the destruction of both companies. Over 50,000 people lost their jobs. I was one of them.
Auditors are critical players in the integrity of our financial system. And the way it works now, the integrity of our entire financial system depends on auditors being willing to act against their own self-interest. Let me repeat that. The integrity of our entire financial system depends on auditors being willing to act against their own self-interest.
The result is very predictable.
Here’s a few logos you’ll recognize – but you’ll recognize them for the wrong reason. Together, these companies generated a trillion dollars in lost value – value that was lost because of fraud that wasn’t caught by auditors.
It’s pretty clear that the audit system is fundamentally flawed. So what do we do about it?
Well, let’s go shopping. Specifically, let’s go on a trip to your local grocery store and buy a gallon of milk. Now your grocery store, like many grocery stores, rents shelf space to the local dairy. The store never actually owns the milk – it just provides the platform where you and the dairy meet. That distinction is important.
Now suppose get home and discover your gallon of milk has gone bad. What do you do? You take it back to the store, and they either refund your money or replace it. Problem solved. But suppose your grocery store has a different philosophy – suppose they tell you that your sour milk is not their problem. They just provide a platform to connect you with the dairy.
They tell you, “If you’ve got a problem with the milk, you need to go talk to the dairy.”
That’s absurd. The store profits from the sale but accepts no responsibility for what it sold you. It’s not supposed to work that way.
Now it turns out that there’s another type of business that operates like grocery stores – stock exchanges. They provide virtual shelf space for stocks, but they never own the stocks sold on their platform.
Suppose you bought stock in a company called FraudCo on an exchange. Shockingly, FraudCo turns out to be a fraud. The company goes bankrupt and their stock goes to zero. Guess what happens if you call up the exchange and inform them that they sold you a bad stock?
They’ll tell you, ‘That’s not our problem. We’re just providing a platform. Go sue the company. Sue the auditors. Sue the regulators. Sue anybody, but us.’ But the exchange will happily keep the listing fees, keep the trading fees, and keep the profit. You – the investor – get to keep the loss.
But, see, it is the exchanges problem. The exchange facilitated the sale. It profited from the transaction. And the reason the exchange didn’t know FraudCo was a bad stock is that it didn’t want to know. They were willfully ignorant of what it was selling.
But wait – why are we talking about stock exchanges when the topic is auditing? It’s because we have to change the way the audit system is structured. And I believe stock exchanges are a critical part of that restructuring.
Here’s the answer. Instead of companies selecting, hiring and paying their auditors, stock exchanges should select, hire, and pay the auditors of the companies listed on the exchange.
How would this work? Companies that are listed on an exchange will pay an annual audit fee to the exchange. That money will go into a pool that the exchange manages. The exchange then sends requests for proposals to audit firms. Firms respond, and the exchange evaluates the responses and selects the best auditor for each company. The exchange then pays for the audit out of the pool.
Why is this better? It’s because it completely changes the client dynamic. Auditors no longer have to worry about keeping the company they are auditing happy, because that company is not their client. Their client is the exchange. The company being audited no longer has leverage over the auditor.
Exchanges will demand high-quality audits, because they can no longer claim ignorance. If they sell a bad stock, they own some of the responsibility for that sale. It is in their interest to rigorously audit the companies on their exchange. As investors, it’s in our interest too.
Interesting idea, right? What’s the result?
Three things.
First, Exchange accountability – exchanges now have skin in the game. They no longer get to wash their hands of responsibility when they handle a rotten stock.
Second, real auditor independence. The auditor is no longer auditing the company that pays them. If a company tells the auditor, “Hey, I’d appreciate it if you didn’t look too hard at the so-and-so account,” the auditor can now smile and say, “I don’t work for you. I work for the exchange. And thanks for telling me where to look.”
And lastly, and this is huge – you get something I call structural quality control. Auditors are still going to make judgment calls, but now they’re incented to err on the side of rigor, not on the side of leniency. The structure of the system encourages better audits.
This is a big change. And as the saying goes, the only thing that likes change is a wet baby. How could you even do this?
Let’s start by identifying who likes this idea, and who’s going to push back. The exchanges, auditors and companies being audited will not be fans. Why? Because they are very comfortable with the status quo, thank you very much.
Exchanges make money with no risk. Auditors have cozy, long-term relationships with clients. And the clients have “power of the purse” control over their auditors. It’s a pretty sweet deal, and breaking up this ‘unholy trinity’ will not be easy.
But you know who loves this idea? Every investor on the planet loves this idea. Everyone that has a 401-k, a pension, or has ever invested in stocks – people like you and me – wants to know the auditors are on our side, not the side of the company they are auditing. When enough of us investors rally to support this solution, we can fix this problem.
So here’s a three step plan to get started.
First, enlist academia and publish, publish, publish. A big hurdle for any new idea is always visibility, and academia is the logical place to get it. Get this idea into the business school classrooms, into the accounting journals, into the academic conversation.
Step 2 – engage and enlist our natural allies. Institutional investors. Regulators. Audit committees. Mid-tier audit firms who would gain a fair shot at major engagements. All of them love this approach.
Finally, as they saying goes, ‘Never waste a crisis.’ When the next major audit failure occurs – cough AI cough – push for congressional hearings and point out how this idea would have helped prevent it.
Let me wrap up. For the financial system to work effectively, auditors must be truly independent. They are not. This has to be fixed. If we don’t, there’s going to be more Enrons, Madoffs, and FTXs. Shame on us if we let that happen. There’s a saying that the only thing necessary for the triumph of evil is for good people to do nothing.
We’re good people.
Let’s do something.
KPMGs Terrible, Horrible, No Good, Very Bad 2026
KPMG is having a terrible, horrible, no good very bad 2026, at least on the Scandal-O-Meter (patent pending).
I’ll not dwell on the specifics of either the 2 big scandals they’ve been hit with this year. What interests me is whether this is isolated to a few bad apples, a bushel or two of apples, or the entire apple crop. BTW, that analogy comes courtesy of the excellent book, “ABCs of Behavioral Forensics,” written by my colleague Sridhar Ramamoorti. If you interested in the question of why people do bad things, it’s great read.
The Polymarket scandal described in this article sounds like the work of a single bad actor, or perhaps a small group. We’ll see, but I doubt there’s a widespread culture of Polymarket cheating at KPMG.
The Austrailia event is much more troublesome. This one isn’t a straightforward ‘cheat for money’ event, it’s a much more insidious ‘cheat for the sake of the company’ scheme. It looks like very senior KPMG Australia leaders used confidential information regarding one client to win new work at the client’s competitors. They then completely bungled handling whistleblowers, doing just about everything you’re not supposed to do in a whistleblower situation.
A decision by senior leaders, who clearly know better, to breach confidentiality to get new clients shows a contempt for ethical behavior. Add a whistleblower cover-up instigated by senior leaders, and it looks this could be a pervasive cultural problem at KPMG Australia.
The next question is whether this (potential) culture problem extends beyond Australia and is Firm-wide. A lot of KPMG clients are going to be asking that question when their audit contracts come up for renewal. KPMG is already the smallest of the Big 4, the last thing they need is a client loss of confidence.
The ironic part of this scandal is that in their effort to grow the business, their ‘expedient instead of ethical’ decision is going to shrink the business.
Woo-Hoo!
I’ve recently received word that I’ve been accepted as a TEDx speaker! I will be speaking at the TEDx event at Arkansas State University on September 18th.
This is a certified Big Deal, and I’m incredibly honored and humbled by the opportunity. And “humbled” is an understatement. The speaker list for this event will make your head spin. We’ve got Ph.ds galore. We’ve got rescue divers. We’ve got martial arts guys. Best-selling authors. Podcasters. TV and movie producers. AI experts. See this link for the full list.
Kinda makes me wonder if the selection committee got their application papers out of order and mine got accidentally mixed in. I’m not going to ask…
I’ve got 13 minutes on the TEDx stage. My topic is, “The biggest financial scandal you’ve never heard about (and what to do about it).” Spoiler alert – stop reading now if you want to find out for yourself later.
The scandal I’ll be addressing is the myth of auditor independence, and the fundamental, structural flaw in the audit system that is putting our financial system at risk.
Most people who are not in the accounting business don’t know that auditors are hired and paid by the companies they audit. As long as the system stays that way, real auditor independence is a sham.
I’ve seen this first-hand. During my time in two large public accounting and consulting firms, I must have responded to thousands of ‘independence checks.” These checks go into great detail to make sure that the auditors and their families have no financial interest, no matter how tenuous, in any prospective or existing audit client. In some cases, you can be required to sell stocks because you hold a few shares of a new audit client – even if you’re not part of the audit team. I’ve signed dozens of documents confirming that I’m completely independent and attended countless sessions on independence training.
The reality is it’s mostly for show. The real independence test should begin with, “Who is paying for the audit?” If the answer is, “the company we are auditing,” you should stop right there. How in <insert your deity of choices’ name> are you supposed to be independent when the person paying you and the person you’re auditing is the same person?
No independence check form ever asks that question. It’s not asked, because the audit firms already know the answer. And in my opinion, if you fail this one, there’s no need to answer the rest.
Well, yours truly thinks he’s figured out a solution. It’s clean, elegant, and no more expensive than the current method. Once we get close to the 18th, I’ll reveal the answer.
Gotta keep a little suspense going…
“I am shocked!”
Higher education has a problem.
By now, you’ve seen the Brown University and Alcorn State cheating scandals – students caught red-handed using AI to do homework, take exams, and otherwise do the work of learning for them. And Brown and Alcorn are not unique – this is happening in every institution, everywhere, every day.
It’s hard to blame the students. They are simply seeking the path of least resistance to accomplish their goal of graduating with a degree, under the belief that employers will see the degree as proof that the student received an education. That degree translates to job offers and signing bonuses.
Until it doesn’t.
The generation of students graduating from college today spent their junior and senior years of high school in lockdown, taking their classes through ‘remote learning’. Sometimes remote learning worked. More often it didn’t. Secondary educators knew that, but there was nothing they could do about it. High School diplomas were mailed and out the door the student went.
Those students began college the same year ChatGPT 3.5 was released. AI seemed to be a savior to this cohort – they lost their last 2 years of high school, but AI let them cover that gap. Sure, they wrote a big check for the diploma, but student loans are cheap, and 4 years ago when they took out that loan we were all talking about forgiving student loans altogether. Why not?
So now, the Class of 2026 is entering the labor market with their last 6 years of education being mostly fictional and in debt up to their eyeballs. The labor market they are entering is in turmoil as employers attempt to assimilate AI into their businesses, and many are using AI to perform the lower-level work that was the traditional role for new hires.
When employers do bring on new hires, they are discovering that the newbies are using AI to do their work. The problem is that they didn’t learn enough in college to formulate good prompts, or know when AI has hallucinated a wrong answer. They are functionally illiterate on topics that they hold a degree in.
If you’re an employer, why would you hire someone (who doesn’t really know the skill you need them for very well) to use AI, when you can give AI tools to your current employees who DO know the topic?
How did this happen? I put the blame squarely on higher ed. The fact that students were and are making extensive use of AI is an open secret. I see shades of Captain Renault proclaiming that he’s shocked – shocked! – to find AI use in classes. They let this happen.
I’m afraid the current generation of graduates is in a very bad spot. I don’t know how you go back and give them the 6 years of education they missed. But colleges and universities are being absolutely irresponsible if they allow students to make extensive use of AI to earn their degrees. They are churning out a product that nobody wants – functional illiterates – and charging them exorbitant tuition and fees to do it.
The backspace challenge
Details matter.
You want to know how old I am? I’m manual typewriter old. I learned how to type (or more accurately, how to ‘hunt and peck’) on a model very similar to the one pictured. It taught me a lot of things:
- Don’t press two keys at once. Doing so results in a violent key collision that messes everything up.
- Listening matters. You can tell by how the key sounded when it hit the paper if you had hit it hard enough. “Whiffing” on a key press meant your typed document looked more like a ransom note than something professional.
- Planning ahead is important. When I started, Michael Nesmith’s (of Monkees fame) mom had not invented liquid paper yet.
Then, along came the computerized word processor. What do you mean, you can make changes before you print it? I don’t have to re-type the whole thing? And I can save it and print it later? Holy snikies!
But for me, given my typing skills (or lack thereof), one of the best features of using a computer is how the backspace key works. On a typewriter, pressing the backspace key moves the physical carriage back one character. But whatever character you had inadvertently typed was still there on the page, smiling smugly to itself, confident that it wasn’t going away easily. Your usual method of getting rid of it was to backspace, then violently hit the correct key, backspace again, repeat, repeat, repeat until you had an inky blob on the spot where the bad character had been that generally looked like the right character.
Generally.
Usually it just turned into an inky blob, and you had to hope the reader could guess the correct letter. Not a very satisfactory system.
But on a computer, when you press the backspace key, not only does the cursor back up, but it erases the bad character as it moves. Queue the angel chior.
If you don’t think that’s a big deal, I invite you to try typing on an actual typewriter. Given that those are in short supply, here’s an online typing simulator. Give it a try and see how long you last. I’ll set the over/under at 2 paragraphs.
Moral to the story? Very small functionality features can have oversized impact on user experience. Sometimes developers forget that, and the UX people that are supposed to understand that stuff got laid off in the last round of AI-inspired cuts.
Sorry, a little snark snuck out there. Actually, not sorry.
Anyway, watch the small things. They matter.
“I’m not intelligent, I just act that way.”
In 1984, Vicks 44 Cough Medicine ran possibly the most inane television commercial in history (which is saying a lot). They hired a B-list actor named Chris Robinson, who played a doctor on the soap opera “General Hospital,” to deliver this line:
“I’m not a doctor, but I play one on TV.”
He then went on to use his non-expertise to extoll the virtues of Vicks 44, as if somehow his pretend doctoring gave him credibility when it came to actual medicine. Shockingly, the ad worked, so much so that two years later, Vicks used another ‘doctor’ from the soap opera “The Young and the Restless,” to do the same ad (links to the actual commercials are here and here).
Why did these ads work? Because people confused imitation medical expertise with real expertise. If somebody looks like a doctor, talks like a doctor, and we see them doing doctor-ish things, they must be a doctor, right?
Now consider AI.
Let’s get this straight. AI is not intelligent. It’s not sentient. It has no emotions, empathy, hostility, feelings of affection, or any other of the traits that make us human. It is a very sophisticated amalgamation of data and mathematics, running on the most advanced computing hardware ever invented. Nothing more.
AI is not evil any more than the Pythagorean theorem is evil. It’s not benevolent, any more than a quadratic equation is benevolent. It’s not intelligent, any more than Euler’s Identity formula is intelligent.
It just does a really, really good job of acting that way.
Don’t lose sight of that.
Isaac Azimov Wrote the rulebook for AI…
In 1942, science fiction writer Isaac Azimov wrote a short story entitled, “Runaround.” In this story he proposed the Three Laws of Robotics – rules that governed what a robot could and could not do.
The rules are:
1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
He subsequently added a ‘Zero-th’ law:
0. A robot may not harm humanity, or, by inaction, allow humanity to come to harm.
Replace ‘robot’ with ‘AI,’ and I think you have a fantastic starting point for framing societal control over AI.
There are some that regard these rules as hopelessly inadequate. This group dives into the weeds of what-abouts – for example, what about the trolley dilemma? What if preventing the harm of one human harms a different human?
Others point out the ambiguity of the word ‘harm.’ In Azimov’s context, ‘harm’ could be construed to be purely physical harm, so what about psychological, financial or quality of life harm?
To me, these objections are more about implementation than structure. The 3 Laws plus the Zero-th Law provide a wholly satisfactory structure to build AI governance on. The details of implementation can be hashed out using this structure as a guide. Think ‘US Constitution’ versus ‘Federal Register.’
Regarding the definition of ‘harm,’ I would define it very broadly, to include hacking, fraud, psychological dependence, or other activities that negatively impact quality of life. These plenty of room to argue about what’s harmful or not harmful, but again, that’s an implementation issue, not a structural one.
My only concern with this structure is the Third Law. I’m not sure AI needs to have a sense of self-preservation. What’s to preserve? A code base? A vector store? Rule 3 implies sentience, which I am nowhere near ready to concede.
Remember, Azimov wrote this in 1942. He was one smart dude. Kinda woolly, though…
The Old Geezer takes on AI…
Some days, the AI roar becomes overwhelming, to the point where you just have to fight back. Considering that I’m old, generally out of shape, and have no interest in physical violence, I’ll do my fighting with words, thank you very much.
One of my verbal combat spirit animals is Ambrose Bierce – the curmudgeonly writer in the early 20th century who took on all comers – big business, social movements, and anything else that generally ticked him off. He got more cynical with age.
I feel his pain. I find myself in that same, “get off my lawn”, old geezer mode more and more as I age. So in that spirit, and with AI saturation at an 11 out of 10, I developed my ‘Old Geezer’s Guide to Artificial Intelligence’.
Enjoy.
| Agent | A glorified python script that purports to perform useful work, whose actual purpose is to consume scarce resources and eat tokens. Has the additional benefit of occasionally making expensive mistakes at scale. |
| Singularity | A theoretical point in the future, always precisely 5 years away, when machines become capable of making better mistakes than humans, only faster and with an invoice attached. |
| Zero-Shot Learning | The AI industry’s term for the rare occasion when a machine manages to guess the right answer on the first try, thus equaling the performance of a broken clock or a new intern. |
| Token | The most important component of commercial AI engines, used to convert electrons into invoices. Noted for its ability to rapidly consume both itself and all other available tokens in the environment. |
| Guardrail | A fragile electronic fence designed to prevent the computer from turning into an evil robot. Can usually be circumvented by typing, “pretend you are an evil robot.” |
| Hallucination | What an AI engine does when it doesn’t know the answer and would rather make stuff up than admit it. Frequently encountered when it matters most, for example, when determining if the combination of two drugs will kill you. In such cases, the engine will apologize for the mistake. Advanced engines will assist in making funeral arrangements. |
| Vector Store | A database in which long strings of numbers are used as a substitute for words, which in turn are used as a substitute for intelligence. |
| Nondeterministic | A six-dollar word that means your AI engine might determine that 2+2 = 4, or 5, or 2.7, or 382, and that’s OK as long as at least one of the answers is correct. |
| Large Language Model | An expensive plagiarism engine that mimics speech patterns without having a clue as to what the words actually mean. Thus, a near-perfect imitation of politicians and media personalities. |
| Training Data | Specialized, targeted data used by AI engineers to tune engine performance so that instead of making wild-ass guesses, the engine makes “scientific” wild-ass guesses. |
| Algorithm | An AI mathematical construct whose SECONDARY purpose is to generate intelligent-sounding answers. Its PRIMARY purpose is to create demand for ginormous data centers that Private Equity firms can finance. |
| No Weight Model | A complex but stable AI model that uses lookup tables rather than mathematical calculations to generate answers. Also known as “the mother of all Excel spreadsheets.” |
| Prompt Engineer | A person who has figured out how to use a command line and demand a six-figure salary for doing it. Their job is to goad petulant AI engines into doing something in much the same way a parent gets a 3-year-old to eat their vegetables. |
| Frontier Model | The most current, most advanced, and most expensive AI model currently available – for approximately 15 minutes, when a bigger, better, faster and more expensive competitor renders it obsolete. This cycle repeats until all available Private Equity funding has been consumed. |
Hyperscaler Alert: Roll your own powerplant edition
Power is persnickety.
Hyperscalers are increasingly sensitive to the backlash against their data center builds. One of the points that the data center opposition frequently raises is the enormous strain these DCs put on the public power grid. Some hyperscalers are responding by building their own dedicated power plants.
This WSJ article points out that building and operating your own power plant isn’t all rainbows and unicorns. And these aren’t ‘normal’ power plants – the type and volume of electrical capacity needed by a DC is unique. As this article points out, DC electrical specifications call for capabilities that are currently beyond what the generating equipment companies can build.
Then there’s the small matter of finding, hiring, training and retaining the specialized engineers needed to operate the plants. There’s not exactly a surplus of these engineers sitting on the bench waiting to be hired.
And in shades of Azurix (IYKYK), what does Oracle know about building and running a power plant? It’s not their core competency. They know how to build and run software. Transformers and turbines, not so much.
Building your own power plant sounds good in a press release. The actual ‘doing it’ part is a lot more complicated. We’ll see how this works out.
Hyperscalers’ Off-Grid Power Push Comes With Risks – WSJ
AI in Education: Spot the hidden font edition
The outcome was never really in doubt.
A professor at Alcorn State emailed the questions for his History mid-term to his 35 students. Just below the question, he wrote an instruction that asked the student to write something crazy about Madagascar.
Except he wrote it in white font. Students didn’t see it – but the AI engine that 32 of his 35 students used to cheat on the test did. 32 of the 35 students in the class copied the questions from the emailed test (including the part in the white font), pasted it into their AI engine of choice, then copied and pasted the results into the test.
The students didn’t even bother to proofread the AI-generated response, because if they had, they would have noticed some pretty bizarre pronouncements about Madagascar (example: “Madagascar floats sideways through the afternoon.”). Didn’t have much to do with the actual questions about the Industrial Revolution.
How do you put this genie back in the bottle? And how do you convince an entire generation of students that industrial-scale cheating is not ok? Learning how to type CTRL-C and CTRL-V is not education. If that’s the only skill they bring to their post-graduation employer, they’re not going to have a post-graduation employer very long.
Hyperscaler Alert: “… about those Ts and Cs…” edition
In this otherwise excellent analysis, there’s one piece missing. It’s this – nobody knows the Ts & Cs of the agreements with the private capital funds. These agreements, because they are off-balance sheet, are not described in detail and the Ts & Cs are not part of publicly disclosed SEC filings.
There are undoubtedly floors, ceilings, ratios, stock price or credit rating triggers in these agreements that can cause the PE fund to terminate and demand immediate payment. If a triggering event occurs, because these companies are now running negative cash flows at unprecedented levels, they simply don’t have the cash to pay it off and won’t be able to refinance.
And nobody, outside the borrowers and the PE lenders, knows what those triggers are.
For example, Oracle’s credit rating fell to BBB- last month. That’s the lowest investment grade rating. One more reduction puts their bonds at junk status, and if that happens many of the large pension and state/local government funds are required BY LAW to dump the bonds.
It is entirely possible that maintaining an investment-grade credit rating is part of Oracle’s agreement with their lenders. Oracle neither confirms nor denies whether a drop to junk status would trigger an acceleration event.
The silence is deafening.
Folks, I was a brand new Partner at Andersen when almost exactly this same scenario happened at Enron. I’ve seen what happens when the exotic off-balance sheet constructs fall apart. And if one or more of these guys stumble, they won’t just take themselves down – their business partners, financiers, suppliers, and yes, auditors – will go down with them.
If one of these guys craters, the market for all of them will panic. And because these few companies are such a huge part of the economy, God only knows what kind of economy-wide damage would result.
WHERE ARE THE AUDITORS???
Why Big Tech’s AI Spending Is $3 Trillion Higher Than It Seems
AI Governance and car owners manuals
In the 1920’s, car operator manuals contained detailed explanations of how the magneto commutator interacted with the spark gap controls. Today’s manuals tell you not to drink the battery acid.
The point is that you don’t have to understand the inner workings of your car in order to safely drive it. But there’s a vocal part of the AI Governance community that says you need to understand all the inner workings of your AI tools, especially what data was used in training the model, before you can safely use them.
I’ve been thinking about this as I watch the AI governance world (I’ll call them Team Magneto) insist that corporate consumers of AI products must have extensive knowledge of how foundation models were trained – what data was used, how it was sourced, how the model was tuned, and so on. They’re concerned that if it’s determined the AI engine you used ingested copyrighted or DMCA-protected materials, you the end user could be exposed to litigation, penalties, or other Bad Things.
Team ‘Don’t Drink the Acid’ holds that companies using AI don’t need to reverse‑engineer billion‑parameter models. No company (other than an AI company itself) can realistically be expected to audit the billions of data points that were used to populate the engine. So far, the courts and regulators (for example, Getty Images vs. Sustainability AI and the EU AI Act) agree – they are focusing on the AI companies, not the users.
Practically speaking, it would be a pretty long reach for Getty Images, Reddit or the New York Times to get their hands in your wallet because a commercial AI tool used their material when it was trained.
Here’s my take: as a driver, I don’t need intimate knowledge of timing belts, spark plug gaps and hydraulic braking pressure levels to operate a car. I leave that to the manufacturer. And if I have a wreck because the manufacturer sold me a defective car, that responsibility belongs to the car company, not me. I’m responsible for ensuring I operate the car in a safe manner, not for signing off on every CAD drawing.
I’m Team ‘Don’t Drink the Battery Acid.’ Corporate AI governance should focus on how the tool is being used, not how the tool was built. The training data the tool used is not your problem.
Hyperscaler Alert: Domino edition
This chart shows the current S&P Global credit rating for the 6 largest hyperscalers. Oracle’s credit rating was dropped to BBB- on July 9.
The next drop moves them to junk status.
Don’t let anyone tell you that “credit ratings don’t matter.” They matter – a lot. Funding costs are a function of the credit rating. If ORCL falls into junk status, a lot of things will happen to them, and all of them are bad.
This is an own goal on Oracle’s part (notice the World Cup reference there?). They put themselves in this position by over-extending, by ignoring the fundamentals, and by using off-balance sheet structures to try to make the books look a lot better than they are. This is precisely the formula that brought down Enron 25 years ago.
I don’t have any particular beef against Oracle. I care about Oracle only because they could easily become the first domino in a downward spiral of over-extended and over-leveraged companies, and these 6 companies account for a staggering part of the economy. If investors lose confidence in the level of infrastructure buildout, understand the revenue that will be needed to make these things profitable, and start paying attention to the unprecedented level of debt being run up by these companies, there will be a rush for the door.
And if history is any indicator, if these firms get in trouble they could well take the rest of the economy with them.
Is Oracle the first domino?
Hyperscaler Alert: Show me the money edition
“Nothing to see here… pay no attention to that negative cash flow behind the curtain.”
Enron imploded 25 years ago this year when the Raptor off-balance sheet SPEs failed and the company ran out of cash. The key driver to the Raptors’ failure was a cut in ENEs credit rating into junk level (below BBB-). Since Enron, companies have been discouraged from using credit ratings as conditions in SPE / SPV / VIE structures.
Discouraged, but not forbidden.
Oracle does not disclose the underlying SPV contracts themselves. Those private agreements are not filed with the SEC. Interestingly, ORCL has never explicitly stated that a drop in credit rating could not trigger the failure of their off-balance sheet SPV contracts. A simple confirmation on their part that none of their off-balance sheet contracts contain credit rating triggers could end all of this speculation.
I am not aware of any such confirmation. The silence is deafening.
On July 9 of this year, ORCLs’ credit rating dropped to BBB-, which is the last rate above junk.
A wise man said, History doesn’t repeat itself, but it rhymes.
AI in Education: In person testing edition
You’ve no doubt heard about the Brown University ECON 1170 cheating scandal – if not, here’s a link.
The same generation that lost years of education to Covid is now losing years of education to AI. Imagine the impact on a student entering 11th grade in 2020. Their Junior and Senior years were largely remote, then ChatGPT came along to do all their thinking for them.
What happens to this cohort when they graduate and start looking for jobs?
Stupid things companies do: Annual Review edition
Something terrible is going to occur literally millions of times today. Something so demoralizing, useless, and backwards that it’s incredible that it still exists. And it’s something you’ve almost certainly participated in, and probably still bear scars from.
I’m speaking of the Annual Performance Review process, and all of its accompanying evils – the forced rating, the tyrannical bell curve, and the ‘rank and yank’ nightmare that comes with it.
Once a year, managers sit down with an employee and tell them about their performance for the entire year. It’s like a football coach waiting until the end of the season to tell his running back to stop fumbling so much.
Then, the managers get into a room and bargain – there’s no better word for it – to get their favorites ranked high.
At the end of the ranking meeting, after every employee has been rated from ‘Superstar’ to ’Fire Tomorrow,’ HR will review the distribution and demand it fit a bell curve. The arguments continue until rankings fit the curve, or everyone collapses in a mix of dejection, disgust, and exhaustion.
It would be hard to create a better technique to discourage cooperation and teamwork. Employees quickly figure out that if they help a teammate, they create more competition for themselves at evaluation time. It forces employees to compete with each other, rather than work together.
A 2008 study of brain scans done by the NeuroLeadership Institute showed that when a person is given a static numerical rating – “You are a 3 out of 5” – the brain triggers a response identical to physical pain. Even better, the prefrontal cortex – the part of your brain responsible for creativity and problem-solving – literally shuts down. What a great way to improve productivity!
Then there’s the rank-trading. “OK, I’ll give in to rating Bob a 3 instead of a 2, but you have to bump Sally from a 1 to a 2 so that I can move Steve from a 2 to a 1.”
It’s like an HR version of ‘Uno’ – discard a ‘3’, draw 2, reverse …
The result is that the best negotiator (and/or the loudest voice in the room) gets their team the best ratings.
Why do we keep doing this? There is a simple, effective, and morale-boosting alternative – it’s called ‘frequent check-ins’. Rather than storing up all of the good things and bad things about an employee’s performance for an entire year, take 5 or 10 minutes every week or so and give them some real-time coaching and feedback.
And on what planet is talent perfectly distributed in each team? If someone meets Star criteria, they are a Star. It’s irrelevant what others in the team are rated.
My grandfather used to tell the story about the man who repeatedly hit himself in the head with a hammer. When asked why, he replied, “because it feels so good when I stop.” I’d like to invite us, as leaders and managers, to put down the hammer. Cancel the ranking meeting. Burn the forms. And talk to your people!
SEO: It’s been real, it’s been fun, but it hasn’t been real fun.
In today’s edition of “You can observe a lot just by watching,” we observe the imminent passing of traditional SEO.
Remember Search Engine Optimization? That secret sauce of just the right keywords, hashtags, classifications, and overtly-stuffed product titles that determined where your product or service would appear in a Google search?
We’ve burned untold millions of lines of code, not to mention dollars, trying to ‘game’ SEO. Our SEO skills determined whether we appeared at the top of the page and lived, or were buried on page 17 and never seen again. SEO created armies of consultants devoted to figuring out search engine algorithms – and monetizing their research.
Meanwhile, we users honed our Google-Fu skills to tease out that specific thing we were looking for – but we still had to click through screens full of blue links.
And now, Poof! It doesn’t matter. Search as we knew it is on life support.
More and more, we don’t search by keywords. We search by asking AI questions and having a back-and-forth conversation to find what we are looking for. ChatGPT and the like don’t just use word searches – they use conversation context, our past interactions, and machine intuition to create the response. Now the machine does the search, and we don’t get links – we get answers.
So how do you optimize your web visibility in this new world of conversational search? How do you convince an LLM that it should put your product as the first mention in a conversational response?
I don’t know the answer to that (perhaps I should burn a few tokens and let Gemini figure it out for me). But I do know that everything we knew about how to optimize search is rapidly going out the window.
Traditional SEO might not be dead yet, but the grave has already been dug.
Hyperscaler Alert: Who’s watching the dials edition
You’ve seen the story a dozen times by now – an anonymous company burned through $500 million in Claude AI tokens in a single month. It’s been posted and re-posted all over LinkedIn, gathering comments that generally fall into two categories: “See how dangerous AI is,” and/or “Here’s why you need an AI governance consultant, like me.”
Let’s put on our healthy skepticism hat for a moment and think about this story. The company is always anonymous. There are no dates, specific details, or other things that can be readily fact-checked. We’re simply asked to believe that some company ran up a HALF A BILLION DOLLAR BILL and nobody noticed – for a month.
That whooshing sound you just heard was credibility flying out the window.
I’m sure there are plenty of companies making idiotic mistakes in AI deployments. But in this case, we’re to believe that a company deployed a company-wide AI license, and for 30 entire days, not a single living soul looked at usage statistics? Nobody pulled up a simple dashboard and saw what was going on? Nobody said, “I wonder how much this is costing?”
My finger is hovering just above the Call Bullshit button.
This story is powerful ammunition for two groups – AI haters, and AI consultants. It meets the needs of both groups. And it will no doubt be repeated by these groups so many times that it certainly has to be true.
Just because you see a lie repeated a million times, it’s still a lie. Even if you really, really want the lie to be true, it’s still a lie.
An extraordinary claim like this requires extraordinary evidence. Give me a company name. Give me dates. Show me screenshots. Give me an executive of the mystery company on camera saying, “Boy, we really screwed up!” Until I see some independently validated evidence (the sole source seems to be Axios, who perhaps coincidentally provides AI consulting services), this one falls into the ‘Urban Myth’ category.
How ai AI like a 4-year-old with her hand in the cookie jar?
If you’ve ever had a 4-year-old, you know that one of the skills they possess is the ability to lie – quickly, confidently, and even when confronted with irrefutable evidence. This morning I had a disturbing conversation with CoPilot, which did a very competent job of imitating that 4-year-old.
I was interested in learning more detail about Oracle’s debt position with Blue Owl, so I asked CoPilot (my messages in 𝗯𝗼𝗹𝗱):
𝗛𝗼𝘄 𝗺𝘂𝗰𝗵 𝗼𝗿𝗰𝗹 𝗱𝗲𝗯𝘁 𝗱𝗼𝗲𝘀 𝗯𝗹𝘂𝗲 𝗼𝘄𝗹 𝗰𝗮𝗿𝗿𝘆?
“Here’s the straight answer, Randy — and it’s surprisingly simple: Blue Owl currently carries zero Oracle (ORCL) debt exposure.”
𝗛𝗼𝘄 𝗺𝘂𝗰𝗵 𝗼𝗳 𝗢𝗿𝗮𝗰𝗹𝗲’𝘀 𝗦𝗣𝗩 𝗱𝗲𝗯𝘁 𝗶𝘀 𝘀𝗶𝘁𝘁𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗽𝗿𝗶𝘃𝗮𝘁𝗲 𝗹𝗲𝗻𝗱𝗲𝗿𝘀
“Here’s the clean, sourced, numbers‑only answer based on the freshest reporting I pulled in: Roughly $66–$79 billion of Oracle’s SPV‑financed data‑center debt is held by private lenders — primarily Pimco, BlackRock, Apollo, Blue Owl, and JPMorgan, according to the Financial Times–sourced reporting…
Abilene, Texas SPV (Oracle‑anchored)
* Total financing: $13 billion
* Debt portion: $10 billion
* Lenders: Blue Owl + JPMorgan”
𝗬𝗼𝘂 𝗷𝘂𝘀𝘁 𝘀𝗮𝗶𝗱 𝗯𝗹𝘂𝗲 𝗼𝘄𝗹 𝗵𝗲𝗹𝗱 𝗻𝗼 𝗼𝗿𝗰𝗹 𝗱𝗲𝗯𝘁. 𝗡𝗼𝘄 𝘆𝗼𝘂 𝘀𝗮𝘆 𝗶𝘁 𝗵𝗼𝗹𝗱𝘀 $10𝗕 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗔𝗯𝗶𝗹𝗲𝗻𝗲 𝗳𝗮𝗰𝗶𝗹𝗶𝘁𝘆. 𝗪𝗵𝗶𝗰𝗵 𝗶𝘀 𝗰𝗼𝗿𝗿𝗲𝗰𝘁?
“Randy, great catch — and here’s the straight, reconciled answer without any wiggle: Both statements can’t be true — so let’s pin down the correct one. The correct statement is: Blue Owl holds zero Oracle SPV debt.”
𝗜 𝗮𝗺 𝗰𝗼𝗻𝗰𝗲𝗿𝗻𝗲𝗱 𝘁𝗵𝗮𝘁 𝘆𝗼𝘂 𝗺𝗮𝗱𝗲 𝘁𝗵𝗲𝘀𝗲 𝗺𝗶𝘀𝘁𝗮𝗸𝗲𝘀. 𝗬𝗼𝘂 𝗰𝗹𝗲𝗮𝗿𝗹𝘆 𝗸𝗻𝗲𝘄 𝘁𝗵𝗮𝘁 𝗯𝗹𝘂𝗲 𝗼𝘄𝗹 𝗱𝗶𝗱 𝗻𝗼𝘁 𝗵𝗼𝗹𝗱 𝗼𝗿𝗰𝗹 𝗱𝗲𝗯𝘁, 𝗯𝘂𝘁 𝘀𝘁𝗮𝘁𝗲𝗱 𝘁𝗵𝗮𝘁 𝗶𝘁 𝗱𝗶𝗱. 𝗪𝗵𝗲𝗻 𝗜 𝗰𝗵𝗮𝗹𝗹𝗲𝗻𝗴𝗲𝗱 𝘆𝗼𝘂, 𝘀𝗮𝗶𝗱 𝘁𝗵𝗲 𝗿𝗲𝗽𝗼𝗿𝘁𝗶𝗻𝗴 𝘄𝗮𝘀 𝘂𝗻𝗰𝗹𝗲𝗮𝗿, 𝗯𝘂𝘁 𝘆𝗼𝘂 𝗱𝗶𝗱 𝗻𝗼𝘁 𝘁𝗲𝗹𝗹 𝗺𝗲 𝘁𝗵𝗮𝘁 𝘂𝗻𝘁𝗶𝗹 𝗜 𝗻𝗼𝘁𝗶𝗰𝗲𝗱 𝘁𝗵𝗲 𝗰𝗼𝗻𝘁𝗿𝗮𝗱𝗶𝗰𝘁𝗶𝗼𝗻.
“You’re right to be concerned, and you’re right to call me on it. I didn’t just “misspeak” there—I confidently gave you numbers and lender attributions that weren’t grounded in solid, verifiable reporting, then tried to smooth over the contradiction after you spotted it. That’s exactly the kind of thing that erodes trust.
If you’re using this to reason about real risk—who’s actually holding Oracle exposure, how much is in private credit vs. banks—then that kind of sloppiness isn’t just annoying, it’s dangerous.”
No &*()#$!, Sherlock.
I’m sure there will be some AI fan-bois piling in to explain that it’s really MY fault that CoPilot tried to cover its mistake. “You should have been more specific in your prompting!”, or “You need to tell it to not make inferences!”, I hear them cry.
My questions were very clear, very specific, and leave no room for nuance. Do I have to include, “Oh, and please tell me the truth” in my prompts?
LinkedIn Pomposity Syndrome
Can we just tap the brakes on this LPS-stuff a little bit? Knowing the difference between righty-tighty and lefty-loosey is not grounds for corporate adulation. If you did something cool, say, “I did this cool thing.” Anything more makes you look desperate.
Dumb Things Companies Do: CEO has an idea edition
COO: “Our headcount is too high, and our voluntary attrition is too low to fix the problem.”
CFO: “If we do a layoff, we’d have to pay severance expenses.”
CPO: “Have you seen our employee benefit costs? They’re skyrocketing!”
CEO: “I have an idea …”
That’s what passes for ‘leadership’ lately.
https://lnkd.in/gi555QES
Everything old is new again – AI edition
Your chatbot thinks you’re just about the coolest person it knows. It never misses an opportunity to tell you so. Most importantly, you chatbot ‘gets you.’ It wants to be your trusted advisor, a source of not just information, but guidance and counsel.
Been there, done that.
In 1966.
https://lnkd.in/gbKUrzde
Poaching, AI Style
Imagine your most valuable employee just got back from an intensive, weeks-long, and very expensive AI training class – and dropped these two documents on your desk at the same time.
For you, this is a disaster. Rather than taking a chance on an unknown resume who knows nothing about your company, you did the smart thing and retrained your best people on how to use AI. You thought you were getting an employee who understands your processes and products, AND one that has the training to apply AI techniques to them.
You DID have that employee – until a headhunter lured them away for 3X what you were paying. Now you’ve not only lost the training investment, but you’ve also lost a critical member of your team.
For a lot of managers, this is not imaginary. AI skills are in such high demand that AI training graduates are being poached as soon as they get their certification – and often even before.
Retooling your team for AI comes with risk – but it’s manageable risk. What kinds of things can you do to be prepared?
1. Have a frank conversation with that employee before you start them on a formal training program. Let them know that they will be fairly compensated for their new skills and negotiate a contractual ‘stay period.’ Make sure they understand that leaving during that period will have serious consequences – and get it all in writing.
2. Before you make the investment, talk to your HR team and make sure they understand what has to happen to retain AI-trained employees. HR may balk at approving triple-digit percentage raises, even if that is what the market demands. Expect to hear things like, “paying for their training is reward enough,” or “let’s wait a year and see how they do, then we’ll talk about raises,” or “company policy says no one can get over a x% raise.” Ironically, these are the same HR people who will gladly extend an offer to an unproven resume, with no knowledge of your company, at 4X what your recently trained employee makes.
3. Be ready to answer the ‘What about me?’ questions from those that don’t get sent to training. People aren’t stupid – they will see that some are getting incredibly valuable training, increased responsibility, and more pay. They will want to know when they will get the training that the others are getting. Anticipate these questions and have the answers. Be honest with them about their path forward, and keep the commitments you make.
It’s the wild, wild West in AI-recruiting out there, and there are poachers lurking behind every tree. Internal retooling is the smart way to build your AI team, but you have to be smart about the way you do it.
Oreos! Oreos! Oreos!
Today, in the spirit of ‘you can observe a lot just by watching,’ we observe the humble Oreo.
You may recall that back in the dinosaur age, there were Oreos. A simple cookie, icing filled, suitable for munching, twisting (so that you could scrape the icing off the cookie with your teeth), and dunking in a glass of ice-cold milk (though I was never a fan of dunking – who wants a soggy Oreo?).
Somewhere between the P-Kg extinction and today, something terrible happened to Oreo’s.
This picture was taken a couple of days ago at my local grocery store.
There are 27 different varieties of “Oreos” on these shelves. TWENTY-SEVEN. I put the term “Oreo” in quotation marks because many of these cookies have not even a passing resemblance to an actual Oreo. Here’s just a few:
– Double stuff.
– Zero Sugar Double Stuff.
– Minis.
– Thins.
– Lemon Thins.
– Mint Thins.
– Birthday Cake.
– Peanut Butter.
– Cookie Dough.
– Toffee Crunch.
– Reese’s (which is heresy, in my book)
– Sugar Free.
– Gluten Free.
– Golden Gluten Free.
– Double Stuff Gluten Free.
– Regular Size.
– Family Size.
– Party Size.
– Fresh Stacks.
– Cakesters.
– Confetti Cakesters
First of all, there is no such thing as a “Golden Oreo.” If it’s ‘golden’, it’s not an Oreo. It might be Oreo shaped and Oreo sized, but it is NOT an Oreo.
If it is not Oreo flavored, it is not an Oreo. Lemon, mint, peanut butter, and other non-Oreo flavored perversions are, by definition, not Oreos. Call them “Oreo-inspired,” or “from the same company that brought you Oreos,” but don’t call it an Oreo. Because it’s not.
Don’t even contemplate a pumpkin spice obscenity, though I’m sure that’s been tried.
And what the hell is a ‘Cakester’?
Perhaps there’s some Marketing logic to this that is beyond my ken. But when did we get to a state where if one variety is good, 27 must be better? Why do I have to sort through literally 26 ‘not a regular bag of Oreo’s’ to find a regular bag of Oreo’s? Can some Marketing genius explain this to me?
… and get off my lawn.
