I think the impact of AI over the next 5 years, particularly on productivity, has been exaggerated.  Which should be no surprise as the impact of PC's and later the Internet also undershot their productivity expectations for the first 5 years.

On the positive side, current AI models have an immense ability to improve decision-making.  Clearly the ability to do faster and sometimes better research supports all kinds of decision-making.  In my old world of hospitality, I can see immediate application to things like better dynamic pricing decisions and (in peaky/seasonal businesses) better staffing predictions.  Investment decisions, trading, inventory management, sales force targeting, and many other such data-intensive decisions likely can be improved with current AI models.  But these are mostly cases where the financial impact is NOT based on workforce reduction and labor productivity.

From a pure person-replacement perspective, the best use case I have seen is in startups.  AI can be a godsend for entrepreneurs and small startups that are trying to perform business activities that large companies have whole departments for, but for them might be a quarter person -- eg website design and maintenance.  Given that these startups have no legacy systems or organizations, AI could conceivably become the backbone of a large company some day as they grow.

The problem with AI is its current untrustworthiness and error rate.  If a startup has some crazy glitch on an AI-generated web site, it probably is not that damaging but the stakes are much higher for established companies.  The problem in my mind boils down to the AI's lack of skepticism.

Everyone has heard of Descartes "I think therefore I am," but his actual logic was a bit different.  It can best be summarized as "I doubt, therefore I think, therefore I am."  The core of thinking for Descartes was doubt, or as I call it, skepticism.  By Descartes' definition, can AI be actually thinking without skepticism?

This isn't a problem limited to AI -- much of the human race seems to have lost the ability to be skeptical.  It seems everyone is really good at a knee-jerk skepticism of anything originating across the political aisle, but the capacity for skepticism for one's own work or for inputs that reinforce one's core beliefs is limited.

For several years in the 1990's I managed consulting teams at McKinsey.  When analysts and associates came back to me with estimates and models, my role was frequently to demand that they have some skepticism about their own results -- does this final number make any sense at all?  I remember one associate (who eventually rose much higher in the Firm than I did) bringing me a market model and proudly showing an addressable market in the trillions of dollars (this was before the SpaceX IPO when they -- with a straight face -- claimed an addressable market of $28.5 trillion).  I asked him if this number made any sense at all to him.  He said that's what came out of the spreadsheet.  I said that when you saw that number, your first reaction should have been to think "wow, there is something wrong in my spreadsheet."

Without any built-in capacity to be skeptical or to reality-check its own results, AI has already been leading some companies over the cliff

• Almost every day in the legal world we see firms getting sanctioned by judges for including non-existent cases and references in their briefs, obviously a result of some AI hallucination.  This has become so common that I actually considered a business a year ago that would hire itself out to review legal briefs for firms and scrub out the AI influence.  When I talked earlier of there being higher stakes in existing businesses, this is a good example -- cases are being thrown out and attorneys are being sanctioned and disbarred over AI failures.
• I know some young people in the same sorts of consulting jobs I was in 30 years ago and they report that as a case manager a lot of their time is spent scrubbing out AI crap from the analysis submitted by their associates.  I have a standing prediction that a public embarrassment is coming soon to one of the major consulting companies as a client cries foul over paying millions in consulting fees for ChatGPT output.
• The #1 cited use case for AI to reduce manpower is in writing code.  And this makes some sense to me, programming after all is just writing with its own language and grammar and punctuation.  But I still am skeptical companies in high stakes situations are going to let AI write mission-critical code.  There have already been a few public failures (I believe AI agents created an AWS outage a while back) and I will not be surprised to see more

My point is not that AI cannot do useful things -- it is that it will be hard (at least in the current state of things) for AI to really get rid of a lot of workers either because the stakes will be too high to chance it's substitution or because almost as many workers will have to be hired to check the results.

This is not a unique opinion -- Apollo's chief economist Torsten Slok observed (paraphrased at Zero Hedge)

But that [software] is the exception. Across most of the economy, and especially in capital-intensive, heavily regulated sectors, deep process re-engineering and data governance requirements could delay structural productivity gains well beyond what the market currently projects. The list of slow-moving sectors is long, spanning health care, banking and insurance, energy and utilities, defense and aerospace, pharma and life sciences, manufacturing, transportation and logistics, construction and real estate, education, legal and the public sector.

For heaven's sakes, much of the government and a lot of the financial sector (including as many as 95% of the ATM's) run on COBOL.  This is not just switching costs, it is a deep understanding of the cost of failure that drives reliance on a well-understood, reliable language platform.

AI will still be adopted in these businesses, but as an add-on decision-making aid overlayed over the top of existing systems.  Which leads me to the question of what will happen to SAAS companies like Salesforce or Oracle.  The stock market has hammered these companies on the theory that corporations are going to adopt AI to replace their functionality.  This seems ridiculous to me.  Salesforce and Oracle applications are deeply embedded in corporation workflows and they are largely trusted by their customers.  These SAAS apps already have their fingers in all the corporate data after years of integration efforts.  If corporations are going to adopt AI for better decision-making, are they going to hire some 21-year-old startup guys to do AI (several of whom have pitched me) or are they going to go to Salesforce and Oracle and look at their AI offerings?  Two years ago I saw the Salesforce CEO speaking at HBS, and his entire pitch was on AI and what Salesforce is doing to integrate AI into its offerings.

One final thought -- these corporations are NOT going to need the most token- and memory-hungry bleeding-edge AI models to do most of what they want to do.  They want something less error-prone, sure, but the mid-tier product is going to be good enough (think of it this way, few computers bought by large corporations have super-high-end graphics cards or processors).  And the mid-tier good enough AI already uses a lot fewer processing resources than the bleeding edge ones, and they will continue to get more compact and efficient over time.  The point being is that the great data center scramble is likely something of a bubble in the same way companies like Enron were left wondering what happened to that big potential broadband market.

Postscript: The specific case of AI for resume screening is an interesting and evolving example.  There is clearly a huge time saving (and perhaps quality improvement) if one can reliably plop 200 incoming resumes and have AI sort it down to the best 10 or so.  The error problem is not a huge issue (at least for the company) as missing one good candidate out of the 200 is not fatal (there are still 10 others) and putting one bad candidate in the 10 is also not fatal (they will be human screened as a next step anyway).  But then job seekers learned about AI screening and use AI now to optimize their resumes, LinkedIn pages, and cover letters to pass corporate AI screens.  Soon hiring will be worker AIs talking to corporate AIs.  My prediction is that the ultimate productivity savings will be low, as companies are going to have to get human to human contact to screen potential employees.

Pretty sure the tradwife folks are targeting something more like the 1950's than the 19th century, but it could be funny -- though not at funny as, say, sending the housewives of Orange County or the Kardashians back to a pioneer home.

But I have a MUCH better idea.  Let's take Democratic Socialist influencers and politicians and send THEM to actually live the life they are promoting, say in 1930's Ukraine or 1960's China.

• Countries don't surrender under the onslaught of air power alone.  Ever.  If anything relentless bombing tends to heal fractures in the population as people band together against the common threat.  As I wrote here

All this of course is to reiterate my skepticism that bombing the sh*t out of Iran is going to lead to any sort of surrender or favorable regime change. I see of late that Trump supporters have adopted the defense that their purpose in Iran is to degrade Iran's military ability and ability to support terrorism and conflicts in the region. But that sure as hell was not the Administration's public line at the beginning of the war. My recollection was that Trump's reasoning was we were going to decapitate the leadership and the people would rise up in revolution

• The US has a terrible record of regime change through leadership decapitation and we almost always end up with something worse, at least without the application of a lot of ground troops.  Iraq is perhaps better off post-Hussein and Panama is almost certainly better off post-Noriega, but those changes involved a lot of boots on the ground.  Venezuela may turn out to be an exception as well but way too early to tell.

Certainly Iran's military has been degraded (though so has ours) but I wonder if the loss of $100 million fighter planes isn't the equivalent of losing the  battleships at Pearl Harbor, ie the loss of a very-soon-to-be obsolete military equipment.  In the world of drones on what increasingly looks like the nature of the modern battlefield, the loss of the old stuff may just accelerate their switch to the new war technologies (a switch I am not at all sure our military in the US is on top of).

Seeing little hope of victory, I have been hoping that Trump would declare victory and Iran and go home, which he has sort of done.  Kudos at least for this, if only the US had done the same in Vietnam in 1965.  However, the peace agreement (MOU?  Docusign?) appears incredibly cynical.  As I understand it, there is an armistice of sorts that lasts until 2 days after the US elections in November.  It is clear that whatever JD Vance is spinning, the US got about zero (and maybe less than zero) from this agreement EXCEPT for the Administration's ability to maybe get the war and gas prices out of the paper until after the election.

Let's say mom and dad have a business that they are going to be able to sell for $100 million.  They have three classic failure to launch kids with expensive degrees in things like art history and anthropology that are not very marketable.  All three from time to time have been employed in various roles (some real, some made up) in the family business.  Over the years mom and dad have created a non-profit and donated half the equity of the family business.  Once the business sells there is now a non-profit they control with $50 million in cash.

OK, so they rename the non-profit the "Smith Family Foundation for Art and Architecture Preservation".  That sounds laudable, right?  They get a lot of prestige from friends and the press for donating $50 million to "charity."

Making some perhaps over-optimistic assumptions about investment returns, let's assume that this new Foundation invests its $50 million (which they now call their endowment) and gets $4 million a year in income.  Here is how this money might get spent:

• We need managers of the Foundation -- hey, let's hire the kids.  We will pay them each an executive salary of $750,000 a year
• We probably need someone who can do actual work, so we will hire an office manager for $60,000 a year
• We need an office, something whose prestige matches that of our new Foundation. We rent 5,000 square feet at $30 per square foot for $150,000 a year.
• We will need supplies, utilities, etc.  Throw in $50,000 a year
• We need at least two board meeting a year with mom and dad and the executive team.  No reason that this can't be at a nice resort with a spa.  Call it $150,000 a year.
• The executives need to visit sites with art or architecture preservation needs.  Where is that?  It could be anywhere, from Florence to Siam Reap.  Eight trips per year at perhaps $50,000 each is $400,000 a year.
• We will need a PR agency to make sure the world knows the good works we are doing, call it $100,000 a year
• We will need to do various tax and legal filings, perhaps $40,000

That leaves $800,000 we can actually donate to other agencies or projects that support our mission.  Good for us!  Make sure the PR agent gets all the details, because after all this is a charity.

I will assure you that, though the IRS scrutinizes some of this stuff more, this is an entirely representative example.  I went to school with kids who have exactly these sort of lives as executives of the family foundation.

Since its origin, this model was expanded and the primary seed of these new non-profits is not generational wealth but money from the government.   When I worked more closely with the government running recreation areas, I was frequently frustrated that government employees almost fetishized non-profits, preferring if possible to allocate all outside contracts and partnerships to them if it was at all possible.  I am not an expert on the history of this mythology, but I can assure you it exists -- whether Federal, state or local, government agencies almost always believe that non-profits are the best partners as they are safe from the taint of profit motive and thus pure in their intentions.

Sometimes that was true.  I remember the government awarding a few campground management contracts to non-profits staffed with volunteers.  The problem was -- and I think anyone who has been part of a non-profit can attest to this -- that as the initial passion fades, it is really hard for such volunteer organizations to provide services 24-7-365.  Most all of these failed.

But then a more clever actor entered the picture in my little business niche.  These were folks who wanted an edge winning contracts, so they built what would in every other respect just have been a business like mine but organized it as a non-profit.  They then won contracts from agencies that were far more comfortable working with a "non-profit" than with a for-profit like my company.  I remember objecting to the agency and saying that the president of that non-profit paid himself more than I made a year in profit, but I got nowhere.  The perceived superiority of non-profits was an idee fixe in the government's mind that I could not overcome.

All of this came back to me as the DOGE effort dug out questionable non-profits on the government take and even more so as frauds in many states like MN and CA have demonstrated that the halo effect around non-profits still exists in government, and can be powerful enough to hide fraud and political money laundering.

Note:  I am still  much weaker than I expected following a post-operative infection, subsequent surgery to clean it out, and weeks of having my body host microbiological warfare on a grand scale.  So I am trying to catch up on a few easy subjects I have intended to write about for a while.

Anyone who has been following Tesla for years has to know what is coming at SpaceX.  In the movie Gettysburg, the great Sam Elliot speaks these lines as General Buford, the union Cavalry commander who was able to slow the southerners just enough on day 1 to let the Union grab the high ground.  But ahead of this success, he fears that he and the union will fail and the South would slaughter Union troops trying to take the hills too late, as at Fredericksburg (and as happened to Pickett a couple days later).

Devin, I've led a soldier's life, and I've never seen anything as brutally clear as this. It's as if I can actually see the blue troops in one long, bloody moment, goin' up the long slope to the stony top. As if it were already done... already a memory. An odd... set... stony quality to it. As if tomorrow has already happened and there's nothin' you can do about it. The way you sometimes feel before an ill-considered attack, knowin' it'll fail, but you cannot stop it. You must even take part, and help it fail.

Having been a (peripheral) part of the online community skeptical of Tesla stock valuation,  I feel I can see the future of SpaceX stock over time as if it has already happened.

There are at least two distinct patterns one sees over time in the stock of Musk-led Tesla that I fully expect to see duplicated at SpaceX.  So it is worth reviewing those.

1. Absurd Valuation Based on Musk Shouting "Squirrel"

Tesla has a Trailing 12 Month PE ratio of 387(!) and a forward PE of 216.  These ratios are almost unprecedented for a company not in the middle of a restructuring, and indicate simply enormous growth expectations.  This is not some weird temporary data spike... Tesla has maintained a PE over 150 for years and years.  Just to give it context, let's compare it to Nvidia which is perhaps the world's most famous growth company right now.  Nvidia's revenues have really gone vertical over the last quarters:

For that it has been rewarded with a PE of 34 / 25 (Trailing / Forward).  So Tesla must REALLY be growing to justify a PE of nearly 400, right?  Well, not really.  In fact, Tesla's revenue has been essentially flat for 14 quarters:

So how does Tesla maintain such a crazy-high valuation?  Honestly, I don't know.  But from watching it and Musk for years I would argue that the most important factor has been Musk's ability to keep shifting the endgame.  The response to valuation concerns is always "yeah, but you are only looking at the current business, [fill in the blank] which is coming soon[-ish] will be worth a trillion dollars."  The fill-in-the-blank over the years has included solar roofs, full self-driving, semi-trailers, battery swap, robo-taxis, neural implants, humanoid robots, and AI.

Tesla Translated to SpaceX:  The proposed SpaceX valuation of $1.75 trillion is, if anything, even crazier than Tesla's.  It is impossible to apply a PE, since SpaceX loses money and can be expected to do so for years, even decades.  But with about $18.7 billion in revenue last year, the SpaceX valuation is nearly 100x revenue (Tesla trades at a lofty 15x revenue).  Nobody, ever, has made money investing in a 20-year-old company with low margins at 100x revenue (barring the occasional sucker who will later pay 120x).

The tell for me is the emphasis and investment in AI at SpaceX.  Strategically, this is a terrible idea as their core business is already very capital intensive and they really don't need a diversion into something else.  They are competing in AI with a number of companies that are far ahead of them and I don't see an obvious way to catch up (very similar to Tesla and self-driving).  Musk says they are ahead but Musk said Tesla was ahead on self-driving and robotaxis until it has become obvious that they are not even close.   There is a potential AI-related launch and hardware opportunity, maybe, someday, to put AI processing in space, but there is no reason that should be dependent on SpaceX's independent investments in AI.  The one thing AI gives SpaceX, of course, is a squirrel to help fill in the value hole between "losing money on $18 billion of revenue" and $1.75 trillion.  Investors in SpaceX can expect a constant stream of squirrels over the coming years.

2. Propping Up Older Musk Investments with Newer Ones

Over the years of following Musk, the one action of his that aggravated me more than anything else was the transparent bailout of his friends' and family's investment in SolarCity using Tesla stock.  Like most other rooftop solar businesses, in 2016 SolarCity was close to bankruptcy.  Rather than allowing that to happen, losing money and prestige for Musk, Musk used his extraordinary control of the Tesla board to have Tesla buy out SolarCity for far more than any sensible market value.  In doing so, Musk trumpeted another classic Tesla squirrel, presenting the Solar Roof, basically modular rooftop solar tiles that looked like wood or slate that would snap together into an attractive rooftop installation.  It was later found that most of the early demo was likely fake, as the tiles were not even close to release-ready and while Musk was predicting 12,000 installations per year and growing, perhaps only 3000 in total were ever completed over 10 years.  The Solar City results continued to fall at Tesla and were rapidly buried in the energy sector, making it almost impossible to figure out how much value Tesla got from SolarCity, given that the vast majority of energy sector revenues are unrelated to rooftop solar and are instead large battery storage projects.

Since that time Musk has used his AI lab xAI to buy Twitter/X.  And then just this year had SpaceX buy xAI for $250 billion.  Does it make sense that an orbital launch company own a social media platform?  Absolutely not, but it bailed Musk out of an investment in X that was going to be very hard to ever recover any other way.

Tesla Translated to SpaceX: 

Last year Tesla booked $890 million in revenue from SpaceX  (cars, battery storage, some AI).  This is less than 1% of Tesla's revenues though I expect it to be, since it was not arms length, more profitable than average.  But the real threat to SpaceX will be, as Tesla's stock valuation eventually starts to return to Earth, that Musk will use his unique control of both companies to have SpaceX buy Tesla.  People are already discussing it. These are two companies that absolutely have no reason to be under one roof EXCEPT that it would help maintain Musk's net worth.  Yes, I am sure he will generate a logic that the Musk fan-boys will love -- AI consolidation or some such.  And I guess it would be accretive, in an ugly way, with a 100x revenue company buying out a 15x.  Just remember that these two companies, which if the IPO price holds for SpaceX, have a combined market cap of $3 trillion and a combined 2025 net income of -$1 billion.  Even if your excel spreadsheet has enough columns to add years marching towards the heat death of the universe, I am not sure that investment ever pays off.

Parting Thoughts

None of this necessarily means that the SpaceX IPO will fail or that SpaceX stock won't rise post-IPO.  I spent too many years getting burned off and on shorting Tesla to ignore the fact that any Musk enterprise commands a premium among a subset of investors -- he is like Warren Buffet in that his name association with a deal has overwhelming value (the only difference from Buffet being that Buffet's investments actually produce profits).  Be aware if you invest that you are likely soon to own Tesla as well, because I do not think Musk can resist the temptation to use high-multiple SpaceX stock as wampum to buy out his other investments.

There is a sort of clock in Musk investments, going back to SolarCity.  There is a lot of arm-waving and squirrels to maintain a valuation, but as business performance inevitably does not live up to the valuation hype, its time to have the next investment that is at the peak of its hype with a huge multiple buy out the old one.  I really thought Tesla might finally hit that point when the valuation collapses to that of a low-growth car company once the robotaxi initiative proved a loser, but here comes SpaceX just in time.

I do not give out investment advice but if I were short Tesla right now I would run for my life.  The SpaceX IPO will essentially be a big Tesla bailout.

All this of course is to reiterate my skepticism that bombing the sh*t out of Iran is going to lead to any sort of surrender or favorable regime change.  I see of late that Trump supporters have adopted the defense that their purpose in Iran is to degrade Iran's military ability and ability to support terrorism and conflicts in the region.  But that sure as hell was not the Administration's public line at the beginning of the war.  My recollection was that Trump's reasoning was we were going to decapitate the leadership and the people would rise up in revolution, an outcome I found unlikely from the first day.

Postscript:  I would have thought it a perfectly defensible position in a war like this to argue against the efficacy of our attacks while still believing the target regime is awful.  Apparently that seems to be a bridge too far for most war opponents, as I increasingly see those on the anti-war side attempting to portray the Iranian government as morally superior to the US.  For all our flaws and our failure to live up to our own standards, that is frankly absurd.  But I still see it every day, women in the US running around protesting conditions for women in the US wearing Handmaid's Tale outfits while simultaneously defending the ethics of the Iranian (or Gaza) governments.

So I will add my usual postscript:  I put all of the above in the "I wish I were wrong" category.  Opponents of wars frequently fall into the trap of supporting the other side.  The Iranian government is one of the worst in the world, both in how it treats its people (or at least the half without a Y chromosome) and its proclivity for inciting violence and mayhem in other countries.  It is a totalitarian regime responsible for much of the current instability in the Middle East and I would love to wave my magic wand and see it gone.

The only vague hope I might have harbored was that the CIA had some secret plan in place with the opposition organized by agents on the ground.  Really, this was an absurd hope, but I grew up in the 60's and the 70's when the CIA had a certain aura of competent deviousness.  Intellectually, I disabused myself of this mythology years ago, but its remnants must have still been lurking around my brain.

For others who might be harboring such vague hopes of secret master spy plans, there is this:

Even a massive military assault on Iran is unlikely to topple the Islamic Republic of Iran and its state system, according to a classified assessment produced by the US intelligence community shortly before the US and Israel launched their current 'shock and awe-style' military campaign on Tehran. The Washington Post first reported it, perhaps based on some kind of leak or briefing by an anonymous intelligence official, and calls it—

a sobering assessment as the Trump administration raises the specter of an extended military campaign that officials sayhas "only just begun."

The report, compiled by the National Intelligence Council (NIC) roughly a week before the war began, concluded that Iran's political system is structured to survive even major leadership losses, The Washington Post reports. However, this should really come as no surprise to anyone awake and observant throughout the past two plus decades of America's 'nation building' efforts in the Middle East, from Afghanistan to Iraq to Libya. ...

The intelligence report also poured cold water on the idea that Iran's opposition could quickly fill any power vacuum. US intelligence analysts assessed that the country's fragmented opposition movements remain too divided to seize control, regardless of whether Washington pursued limited strikes against leadership targets or a broader assault on state institutions.

Equally unlikely, according to current and former US officials familiar with the analysis, is the prospect of a spontaneous nationwide uprising. We could speculate that this possibility may have had a chance of some degree of success within the opening one or two days of the mass US-Israel bombing campaign, but it clearly didn't materialize.

I will observe that no such promised revolution has occurred so far after the Maduro snatch.  You can almost visualize the Administration look of confusion when the revolutions they were convinced would magically appear did not occur.  Sort of like the look on the coyote's face when some trap he has created fails to work.

Postscript:  I put all of the above in the "I wish I were wrong" category.  Opponents of wars frequently fall into the trap of supporting the other side.  The Iranian government is one of the worst in the world, both in how it treats its people (or at least the half without a Y chromosome) and its proclivity for inciting violence and mayhem in other countries.  It is a totalitarian regime responsible for much of the current instability in the Middle East and I would love to wave my magic wand and see it gone.

The US is really good at getting rid of leaders like this, and if anything is getting better.  I won't go further back than my lifetime, but the Diem coup (and execution) in South Vietnam, the lukewarm (at best) support for the Shah of Iran that contributed to his ouster, Saddam Hussein in Iraq, the Afghanistan invasion, Gaddafi in Libya, Maduro in Venezuela, Noriega in Panama -- the list goes on.  But in many or most of these cases, what followed the US-led decapitation was as bad or worse than what came before.  Vietnam - equally bad or worse.  Iran - worse.  Iraq - better but took a really long commitment.  Afghanistan - at least as bad or worse.  Venezuela - unknown but no immediate revolution as hoped.  Libya - much worse.  Panama - probably better.

We have no historically successful roadmap to go by, and in a sense this may be a situation like Hayek's critique of government planning -- that a perfect roadmap cannot exist because we don't understand the mass of individuals we are "liberating", or even how they define "liberated', or even if they really want to be "liberated."  As all of us humans do, we project our own preferences and outlooks and assumptions on people where they may well not fit at all.

Even beyond the job of seeing Iran no longer acting as a leading agent of chaos, I would greatly love to see their people liberated.  Women in Iran who were just emerging into the 20th Century under the Shah's leadership have a chance to emerge from gender apartheid again, and I am 100% hoping to see this.  (I wrote a while back about the utter lunacy of US women on the Left consistently siding with hardcore Islam and ignoring the plight of women in these countries).

Unfortunately for my optimism, I said the exact same thing, almost word for word, when we invaded Iraq.  Iraq has since struggled to fulfill this promise, though to be fair a lot of the blame for that rests not on US failures or the Iraqis but on the ongoing efforts by Iran to subvert the country and keep it roiled in chaos.  But getting there took a HUGE US commitment of money and lives, way more than a pushbutton decapitation of the leadership.

A parting thought -- there is clearly an Iranian opposition.  We have seen them bravely marching in the streets (far braver than our anti-fascists here as they faced actual imprisonment and death for such protests against real fascists).  This is an honest question -- around whom does the Iranian opposition rally and organize?  As in many such authoritarian societies, only the authorities have organization.  So even decapitated, the military and former government theoretically have a huge head start in pulling things together under their control in the aftermath than an unorganized populace.  This is the same problem faced by many post-colonial governments.  It's not that their populace wanted a military dictatorship when the colonizer left or was thrown out, but in many cases the only organized and educated group in the country was the military which stepped into the vacuum.  I am not an expert on this but I have always assumed India escaped this fate because it had a relatively large, educated group of indigenous people trained in government and not in the military.

Postscript:  I continue to find it sort of hilarious that media that go out of their way not to deadname a transexual teen insist on describing Iran as part of the Arab world and their citizens as Arabs.  I can tell you with great confidence and many experiences that there is no way to piss off an Iranian faster than to call them an Arab.

First, a digression.  Anyone who is more than a casual user of Microsoft Word understands that there are fundamental bugs in the core of the program that have existed since almost the very first version and have never been fixed in almost 30 years.  Two that come immediately to mind are the difficulty in getting images to stay where you put them and the absolutely terrible structured outlining (eg section II-B-iv-2-a).  The former is so bad you can find a zillion memes on it.  The latter is so bad that Word Perfect still survives focused on lawyers who write a lot of documents with hierarchical bulleting.

Everyone knows these problems exist.  Presumably they are fixable with some amount of effort.  But they are not fixed.  Instead, release after new release in Word trumpets new niche functionality without ever focusing on the core functionality. I can't remember ever using a feature of Word that was added since 2005, and maybe earlier, but yet adding those new features is what consumes all the development time.

My fear is that AI companies are doing the same thing.  New features and capabilities of the major AI models are impressive.  But at their core, at least for researching and writing, they still have the critical, fatal flaw of hallucinations.  Almost every day we can watch some law firm get reprimanded by a judge for submitting briefs that include fake, made-up, hallucinated cases.

I don't care how capable and human sounding these ai models are, if they are inserting reputation-destroying hallucinations in a firm's output, or writing in an identifiable AI style, they are worse than useless.  And companies that say "Oh, we don't use AI" are fooling themselves because even the best and brightest kids that they are hiring have become habituated to using AI to finish research and writing assignments.  A young woman I know who manages case teams for one of the big strategic consultants (I won't give the name but think McKinsey, BCG, Baine, etc) says that a huge part of her job as engagement manager is to stop AI-generated slop with obvious errors and recognizable AI writing style from getting to the client.  Her case team keeps handing her things that at best are obviously AI prose and at worst contain errors.  Interestingly, she checks all this stuff not because she was assigned to do it, but because she grew up on the AI/non-AI temporal border and sees the risks.  I have a bet online where I believe one of these firms is going to be caught up in a public scandal and lawsuit in 2026 for turning in ai-generated client presentations while billing that client 7 figures a month (imagine the explosion when a CEO finds out they were paying $1 million a month for the output of a few ChatGPT prompts).

The problem is actually bad enough that I briefly considered starting a new firm whose sole job was to independently review, fact-check, and edit all of a firm's output to help them identify hallucinations and AI tells.  You could probably go hire 100 of the older generation of Washington Post layoffs right now who have actual reporting, editing, and fact checking experience (avoid the younger ones who grew up in the journalism as advocacy era).  Go out and sell your services to law firms and consultants and such.  Gotta be a business there.  Right now I am too newly retired to pursue it but I will leave the idea to you guys.  You're welcome.

Obviously, nothing about what I describe above sounds like the employment apocalypse everyone is expecting.  You are simply not going to see the promised productivity gains until AI cleans up its house and in my mind that would include transparency about hallucinations -- what are the rates, what have they done to fix them in this version, are the rates going down, etc.