Survey-based data doesn’t exist because the surveys were not conducted in October. But data that companies electronically submitted exists.
By Wolf Richter for WOLF STREET.
The jobs report, normally released on the first Friday of the month, is based on data from two big surveys: a survey of employers, which produces nonfarm payroll changes and wage changes; and a survey of households, which produces the unemployment rates, labor force data, participation rates, the number of unemployed, etc.
The first, the Establishment Survey, is based largely on electronic submissions by employers, and those continued during the shutdown, which started on October 1.
But the second, the Household Survey, is based on surveys sent to households, and as the government shut down on October 1, the survey in October was not conducted, and the data doesn’t exist; and it would be difficult to conduct a survey retroactively for October.
So the jobs report for October, which was originally scheduled for Friday November 7, will, when it is finally released, contain only the data from the establishment survey, and so it will have the number of jobs created/lost and wage increases; but the household survey data will not be available, and the data may never be available, National Economic Council Director Kevin Hassett told reporters today.
Non-farm payrolls for October “will be able to be calculated but the household survey wasn’t completed, so we’ll get sort of half of the jobs report,” he told reporters. “Most everything else” can likely get worked out, “but we’ll never know what the unemployment rate was in October,” he said.
The November jobs report should be back on track.
The September jobs report, originally scheduled to be released on October 3, was not released because the government shut down on October 1, preventing the BLS from completing the report. But the data was collected before the shutdown and is all there. Hasset said today that the September report is essentially completed “and we might get that next week.”
The BLS has not yet published an updated release schedule.
Yesterday, White House Press Secretary Karoline Leavitt told reporters that the jobs report and the Consumer Price Index for October would “likely never” be released.
“All of that economic data [from the shutdown period] released will be permanently impaired, leaving our policymakers at the Fed flying blind at a critical period,” she said.
The shutdown “made it extraordinarily difficult for economists, investors, and policymakers at the Federal Reserve to receive critical government data,” she said.
Hasset’s comments today provided a clarification of what Leavitt had said, that part of the jobs report would be released – the nonfarm jobs – but that the unemployment rate and related metrics for October would never be known.
The October CPI report may also be permanently lost because a big part of the data wasn’t collected in October.
The September CPI report was cobbled together the best it could be during the shutdown by hastily recalled BLS staff, but was marred by a huge outlier in the biggest component, Owner’s Equivalent of Rent (OER), which accounts for 26% of overall CPI, 33% of core CPI, and 44% of core services CPI. Something went wrong, and given its huge weight, OER significantly pushed down the month-to-month readings of overall CPI, core CPI, and core services CPI, which I discussed here.
The six-week shutdown also did some damage to the economy, not just the data. It delayed projects and investments, cut off salaries, threw people out of work temporarily, delayed SNAP benefits, disrupted air travel, etc.
How much damage it did will take some months to sort out. Estimates are all over the place how it would impact Q4 GDP – we haven’t even seen Q3 GDP yet, thanks to the shutdown – but typically, as payments begin to flow again and projects and investments come off hold, and disruptions vanish, the economy bounces back with some catch-up effects in the following quarters.
Enjoy reading WOLF STREET and want to support it? You can donate. I appreciate it immensely. Click on the mug to find out how:
1:04 PM 11/13/2025
Dow 47,457.22 -797.60 -1.65%
S&P 500 6,737.49 -113.43 -1.66%
Nasdaq 22,870.36 -536.10 -2.29%
VIX 20.15 +2.64 15.08%
Gold 4,181.70 -31.90 -0.76%
Oil 58.68 +0.19 0.32%
Dow falls 800 points as major U.S. stock indexes tally worst day in over a month
DJIA -1.65%
SPX -1.66%
COMP -2.29%
Thursday’s drop took away TWO DAYS of gains.
So what?
It is remarkable that people feel entitled to the “free money” they get from the stock market….as if it is “automatic”.
How much damage?
Enough to kill the penny!
Allegedly they’ve minted the “last penny.”
The re-opening hangover is about to be real. People coming back to work, and those people who have been working (and maybe not getting paid on time?) have a lot of catching up to do.
Others may have been sacked “overnight”?
No more pennies is coming….
then, do they “round up” the prices or “round down”?
Guessing up
This has been predicted for a long time. Disregarding pennies.
Price of stamps also a manifestation of inflation.
Any way you cut it, promotion of ANY inflation is a race to zero. IMO
It’s illegal to round up currently. So they round down to the nearest 5 or 10 when customers are paying cash and are out of pennies. But that could be illegal too because it discriminates against people who purchase with their SNAP debit cards, which is how they spend federal food stamp money, and merchants cannot charge them more than other customers, such as those paying cash. So they have to round down for everyone, including people who pay by card? There is a lot of uncertainties how to implement it. They did it successfully in Canada over a decade ago. You only have to round to the nearest 5 or 10, so the most rounding you’ll do is 3 cents. But there needs to be a legal framework in the US to do this.
Is the rounding pre or post-tax? If the former, then they’d still have to round the post-tax total. If the latter, than the pre-tax prices would be slightly different in different areas.
The only way around this problem is pricing everything in increments of $0.05. But you can’t really do that with things purchased per weight/volume etc. where you’re multiplying by a unit price.
Why not just get rid of a decimal place and round everything to the nearest 10 cents? Seems easier.
Pre-tax prices pose no problem because the store can round them to the nearest 5 or 10 cents. But when you add 8.63% sales tax to $99.90, you get $108.52, which is where you’d have to round to $108.50 or $108.55.
There has to be a legal framework for how retailer should do this — and they will come up with something.
Australia did the same thing, more than 20 years ago. If you paid in cash, it was rounded up or down according to normal rounding rules. If you paid by card, the totals were by the cent. No one had any problem with it.
I don’t think they have class-action lawsuits in Australia.
All the news about small businesses demise with the loss of the penny make very little sense to me. I’ve always understood merchants to be happy if someone pays in cash rather than credit as they avoid the processing fees. Even if they had to round down 4 cents, nearly every transaction will be less revenue “lost” than having to process that via credit. If more places used the upcharge approach to build the credit processing into the cost paid by the consumer, more of us sailors would see the effect of only paying by credit. Every time I stop to buy a soda at 7-11 for 99cents with my credit card I cringe, but most of the upcoming generations know very little about using cash.
The cashier who just works there doesn’t care about the owner’s credit card fees, and counting money / doing math is more work than telling a customer to insert or tap their credit card.
Vibes of Argentina. People okay with 4% 10y yield are blowing my mind. Passive buyers through bond funds on non-competitive bids?
People licking chops at 1% real returns?
Blind hope that we’ll be pressured into QE when it’s no warranted so yields will dip and prices will go up? Would that matter though? The USD would tank in that scenario
Re: People licking chops at 1% real returns?
Some worry about returns on the money, others about return of the money.
Given where stocks are, 1% real returns may turn out to be a good deal. I mean who knows. The AI bubble isn’t going to live forever.
10-year TIPS (Treasury Inflation Protected Securities) are currently guaranteeing a real return of about 1.8% per year for 10 years. That’s the yield they pay, on top of whatever their inflation protection (based on CPI inflation rate) will be.
I invest in a lottery ticket once in awhile. For a small initial investment of $2 there is the potential of a real return of millions…
How do you beat that ??
I might even live on the edge and go all in.
HUCK
Lottery is a special tax for people who are bad at math, some wise man said many years ago (I cannot remember who). State governments certainly need the revenues and appreciate it. I’m not sure they ever thanked your for it.
My initial comment was going to be “One can always opt for 30% loss”. Great minds..
Michael Burry (of the “Big Short” fame) threw in the towel shorting Nvidia and Palantir and closed his hedge fund. Probably got squeezed out by the “blow off top” of the last few weeks. He posted his position, and I practically have the same (even strikes and timelines are close), just heavier on Nvidia and got in at a better time and not all at once. Slightly in the green after today. Hoping to hang on.
The once in a lifetime version of that trade was in 1999, when the recently introduced 30 year TIPS had a real yield of 3.875% and were trading at a discount.
Lotteries are for suckers and losers. You have more probability of getting killed by a meteor than winning the lottery. I was in my local grocery chain the other day. Instead of doing their jobs many employees were tying up the customer service desk playing the lottery.
I’ve always thought Casino’s should be forced to rename themselves.
“Buildings for the mathematically challenged” feels more appropriate.
Regarding the lottery, if the payout is $500 million and your odds of winning are 1 in 200 million, then buying a $2 ticket makes sense. If you’re already in the top tax bracket. Well, except they don’t actually pay out $500 million. So, scratch that. 😁
Touche. I like Wolf’s TIPs vs future S&P real returns comments.
I don’t know that USD or US Treasuries are where I’d put my money today if my concerns were the return of the money. I don’t think I’m the only one that feels this way (looks at negative yield on 2 year Swiss bond).
Chuck might shut the gov again on Dc 31 for Capo dei Capi.
Wonder what the “perfectly paved” Rose Garden and new “Ballroom” cost?
Allot but the taxpayers don’t have to pay a dime.
You forgot /sarc tag.
The banks haven’t been reserve bound since 1995.
And shut down for absolutely nothing
Depends on your perspective. Your “representatives” got paid and so wil many federal workers that were sent home. To quote Dire Straits; “money for nothing”
“Yesterday, White House Press Secretary Karoline Leavitt told reporters that the jobs report and the Consumer Price Index for October would “likely never” be released.”
“never released” vs. “never collected”…semantics, so Hasset had to put the milk back in the bottle. Press Secretary needs to talk to someone before engaging mouth. My humble opinion.
Wolf, I’m a Korean who came across your blog by chance. Korea is also very interested in the stock market, and in particular, many Koreans are interested in American stocks rather than Korean stocks. I’m reading your blog posts very carefully and getting help Thank you.
Welcome aboard!
KIM, I would buy stock in a company that makes good K-dramas. They still put beautiful women on TV. We stopped doing this 20 years ago.
In Korea, many companies are still run by founding families rather than professional managers, and this hereditary structure often lacks transparency. Because inheritance taxes are extremely high, these families actually tend to dislike rising stock prices. And honestly, about half of the pretty women you see probably have had some kind of cosmetic surgery.ㅋㅋㅋㅋ
Yes, I think they call it “chaebol” in K-dramas. Very interesting. Please comment here often. We could learn a lot from you. Thank you.
Kim, cosmetic surgery is really big here in the states,especially with the “over 50” crowd of women (and older)!
Neighbor lady, old like me, is into Korean k-pop, go figure…
Kim,will say some of the movies(mostly horror I have seen)are pretty well written and acting pretty damn good.I feel your country is well on it’s way to be a film powerhouse down the road unless the ai kicks in too much which may happen to the industry world wide.
There are plenty of ways to survey citizens to establish unemployment data, and plenty of demand. There should be a competitive marketplace to produce quality employment information; this should not be a purely Federal responsibility.
Rasmussen Reports, among other polling companies, puts out a monthly unemployment report. They last reported on October 14 (unemployment was steady), and will likely have their next update in a few days.
The current Federal system is trapped in the 1970s and is long overdue for an update. For instance, the IRS gets tax-withholding data on just about everyone, doesn’t get shut down, and could be supplying anonymized economic data to the BLS etc. every month. Those employment data don’t require any surveys at all.
Partisans on each side complain about the federal employment data whenever the other party is in charge, and nonpartisan observers complain about the federal employment data at all times. It’s time to improve the system.
You’re just clueless about how the BLS does it, but opinionated out the wazoo. There are legit criticisms about how the BLS does it, and I brought up a few recently, but yours is just nonsense. So here we go:
Your 3rd paragraph: Unlike you imagine, the BLS DOES get payroll tax filing data from all companies – but they file quarterly, NOT monthly, LOL — and they become the primary input for the BLS’s Quarterly Census of Employment and Wages (QCEW), which is incorporated into the jobs data. But employment tax filing data is quarterly, and filing deadlines are well after the quarter ends, so this quarterly tax data cannot be included monthly and it’s not timely. The BLS makes annual adjustments retroactively (benchmarking, as it’s called) every February, and provides a preliminary partial estimate of what the adjustment would be in September. Even this detailed employment data is not complete, and the BLS gets additional data before the February benchmarking.
I discuss this every September and February. It would be helpful for you to read these things so you don’t have to make a fool of yourself. Here are September’s preliminary benchmarking results:
https://wolfstreet.com/2025/09/09/annual-benchmark-payroll-revisions-nonfarm-job-creation-for-12-months-to-march-chopped-by-911000-to-1-45-million-jobs-created/
Your 2nd paragraph: Polling companies, like Rasmussen, use small-scale surveys, compared to the BLS household surveys. Rasmussen asks 10,000 adults one question about employment. The BLS survey goes out to 60,000 households with detailed questions that provide a wealth of data. Plus, selected households get phone interviews.
I cannot believe you compare the el-cheapo Rasmussen one-question survey to the vast BLS data trove.
You didn’t mention the ADP report, which is the second-best data set after the BLS’ nonfarm payroll data. But it’s based only the companies for whom ADP does the payroll. ADP has lots of competitors, and many companies run their own payroll (we did too), including the biggest companies in the US, so the data can be skewed. And it doesn’t provide unemployment data at all. And in October every year, the ADP report is hugely adjusted to the BLS QCEW benchmarking data, LOL, because the BLS data is the gold standard:
https://wolfstreet.com/2025/10/01/adp-employment-report-32000-jobs-in-september-after-43000-annual-adjustment-to-benchmark-adp-data-to-bls-data/
Wolf,
I don’t want to get into the weeds and I don’t want to argue and I’m not disagreeing with anything you are saying.
I guess my big broad question regarding this post is…
Unless some big discontinuity pops up once reporting starts again, will a missing month really matter much in the whole scheme of things?
My first instinct (perhaps wrong) is to think not – without some mysterioso discontinuity appearing, I’m not sure how bad the damage from the missing info could get.
Now *if* some big discontinuity appears – that sets off my conspiracy meter (ie, “what why and how did this “thing” just happen to happen in the dark?”)
But we’ve already seen a lot of historical discontinuities appear from apparently plain vanilla shortcomings in methodologies already (revisions to prior periods’ employment levels come to mind).
And yet *those* didn’t seem to gin up the same kind of anxiety as the “missing month”.
In terms of jobs and inflation data, the BLS establishes the totals and then figures the percentage changes from those totals. So the first month that they have data again will capture indirectly the missing month in their totals. On that level, there won’t be any month-to-month changes on the first month that they have data again, because there is no data for the base month (October). Then in October 2026, there won’t be any year-over-year percentage changes because there is no data for the base month. They will likely fix this by filling the holes to get month-to-month and year-over-year readings, possibly by using three-month averages, for example. And they will put a caution flag on the month-to-month changes.
With so much data missing, we may see some quirks, like we already saw in the CPI for OER for September. I’ll point them out if I see any big ones.
I’m glad I don’t have that job. This is a huge mess to sort through, and everyone is going to be mad at you.
Just complementing Wolf’s comentary: The crucial things about National statistics are the correct sampling, the size of the sample, and the lack of answers. That’s because detecting and correcting biases is very difficult to do without redoing the surveys.
Adding up a disparate group of small-scale samples would skewer the hell out of the results.
Yeah, I also see that the current administration is saying that all the numbers would be “bogus” anyway. Okay, so why not claw back those salaries as the job was never completed correct? So many statements being made with absolutely no evidence. America continues to follow the path that so many dying empires have taken. As a country, we are not going anywhere, but we have squandered a great amount of blood and capital. Do NOT expect the government (now in bed with some many private entities) to become more transparent. Ignore the talking heads and start to believe your own “lying eyes” and the numbers being provided on sites like this. Also, as usual, hedge accordingly.
The jobs market is in a free fall as we speak. Verizon just announced a cut of 15,000 jobs. Blue Cross/Blue Shield just announced that they will use AI Bots to replace their customer service representatives nationwide. All other health insurance companies are set to follow suit. Same with Auto and home insurance companies. The next time you are in an accident and are laying on the side of the road, look forward to contacting an AI Bot to report your claim. ENJOY.
Swampy,
According to Ford Motor’s CEO Jim Farley, they have over 5000 openings for auto mechanics -> +$100K salary. There are numerous examples of other companies seeking people to do work with their hands.
I don’t know, but if this/these cases(s) are real, there are jobs out there, but many young Americans are unwilling or unqualified to obtain them.
For the past 40 years, people have been going to college, often getting useless degrees (liberal arts, political science…) and working in unrelated jobs. Most of these jobs hired college graduates because they could do so. Thus, a college degree became a “requirement”.
With grade inflation and the terrible state of America’s primary and secondary education system (even Harvard has remedial classes now), a generic college degree is becoming less valuable.
So in closing, I think we have a skills gap.
“but if this/these cases(s) are real”
We may definitely have some degree of a skills gap…but these “$100k for some job that used to pay $35k” I think have to be taken with a grain of salt.
Truckers leap to mind…periodically (like maybe 3 times over 20 years) the $100k+ trucker (without majorly onerous rqd hours/job terms) job becomes a thing.
(Walmart was talking this up for quite a bit…for a while…a lot of radio silence more recently)
But then…the caveats tend to reveal themselves – the professed jobs turn out to be “up to” $100k+, or only a pretty slender slice of positions are at that pay level, or the market re-crashes pay within a year or two, etc.
I did a bit of digging on the Ford announcement…there is some real there…and some bullsh*t (not unexpected balance).
1) Real – Ford has launched an apparently well funded “Academy” program in conjunction with many schools around the country to actually produce more mechanics – by **paying** to train them rather than expect they would magically appear. Good – but…
2) Ford only launched this…early this year. This is an issue/problem that has been discussed/worried about for *literally* decades…but Ford is only really putting its money where its mouth is…now?
And sorta lecturing everybody *else* about their short-sightedness?
3) Those disparate factors are what lead to a degree of organizational distrust and distrust makes it harder for people to make costly/time consuming investments in a given career.
“$100k for some job that used to pay $35k” I think have to be taken with a grain of salt.”
It’s low for good techs. Our best techs that knew how to run their teams made more than that back in 1995 in Oklahoma! We had a team shop, where each team was run by the most experienced tech, and under him were various levels of specialties and experience, including always a trainee. Each team had 4-6 techs.
Techs get paid by the flat rate hour times a portion of the shop’s labor rate. If they’re fast, they make a lot of money. I don’t know what the best techs make today, but it’s well over $100,000.
When you look at averages, they include a lot of the low-level techs working at lube and tire shops, who are paid a lot less, but don’t have to have the skill and knowledge levels of techs in a dealership. But what Farley was talking about are techs at Ford dealerships.
Great techs that work at dealerships and have to be able to diagnose and fix everything, don’t grow on trees. They’re hard to find, and usually stick with their employer, so they’re hard to poach. We hired fairly inexperienced techs, often fresh out of trade school, and spent lots of money training them, both in the shop and by sending them to school (automakers have schools to train techs in a classroom setting and remotely on the technology of their vehicles) to get them certified for various specialties, new technologies, diagnostic equipment.
To become such a tech takes talent and eagerness, lots of hard work, lots of training (we paid for all the training), and many years, maybe a decade or more of experience. And it’s physically a hard job. Which is why they’re paid well. College grads are a dime a dozen, but useless in the shop.
Cas,there are a lot of programs out there,from the tech high schools to a hands on/class local community college that gives you a Associate and also a great education in auto tech that leads to well paying jobs immediately.
Back when I looked at program if you were willing to move to Pa. you could work a dealership there and with night classes you would have a engineering degree assuming you did the work,they recognized the 2 year program as half way to bachelors degree.
I have 2 ASE Auto Elect certs…from late 80’s and 2012…..expired now.
Plus L1 and L2 smog licenses, taught by BAR, pay fee to BAR….to run/maintain BAR smog machines for diagnostics only (but it is VERY useful), next step would be a smog lisc.
I think it ought to become a separate type of “auto mechanic”, who studies at least some electronics first.
I have always been curious about the perception of “available work.”
I generally conclude that it comes down to a mismatch: the work that “I want” isn’t available = there’s “no jobs.”
The job market has been “projected to free fall” since the 2022-23 mass-layoff “announcements.” The same layoffs that never fully materialized, also from multinational corporations that employ workers worldwide.
I could learn to be a mechanic, but I don’t know if I would enjoy working for a Ford dealership (or wherever they have the jobs): another case of mismatch (location and work environment).
Other *major* causes of mismatch – time of demand and lack of info/awareness.
If certain jobs only have pretty rare surges in demand/shortfall in supply of employees, then it is very hard for people in their “training window” (18-30) to see them as attractive careers and train accordingly.
1) Employers seem to expect potential employees to magically self-create and appear on demand. Like I said – Kudos to Ford for laying actual money (apparently) out for its Training Academy.
But it only did so in early 2025 – after similar employee shortage issues had been discussed/worried over for literal decades.
2) Lack of awareness – It takes years and years of employee supply shortfall to generate a strong salary spike – why didn’t Ford actually promote this $100k figure years and years ago? People can only respond to market signals if they are actually…signalled.
re: “Most of these jobs hired college graduates because they could do so.”
True enough, but also possibly because a college graduate would more likely be able to read at a ninth-grade level.
Don’t be so sure college kids can read.
UC San Diego released a report the other day showing terrible math abilities, as well as stating that 1/5 of freshmen have entry-level writing abilities. Junior high level, that is.
There are high correlations between writing and reading skills, so I doubt all of these “college grads” hired in the past are significantly “smarter than a HS grad”.
The situation with hiring most college graduates was like the case when a company in the 60s or 70s bought an IBM computer – no one ever got fired because of buying IBM.
Same as hiring a college grad – if the person was a dolt, the hiring manager could say he had a degree from XYZ college.
“According to Ford Motor’s CEO Jim Farley, they have over 5000 openings for auto mechanics -> +$100K salary.”
They need labor to handle all those CP4 fuel pump replacements lol…
I met a nice young guy, early thirties, a couple years back. Did collage in IT(yes, a pun) fields, looked around and decided that was going to be a dead end. He researched practical applications for IT skills and came across a tech job which requires the IT skills but applied in the real world. He took a training class for Electrical Generation equipment and upon finishing was slammed with offers.His job is to be “on call” for our entire Island and be the man to restart the utility generation equipment when it goes down. He was given a new truck, expense account, keys to all the utility properties around the island, gets paid VERY handsomely for “being there” 40 hours per week plus massive OT, triple time, etc. when he has to work outside his 9-5 time frame, which is actually often.
Our local utility had tried to fill his position for 2 and a half years before they landed him. He’s treated like royalty. Oh, and he lives quite comfortably in Hawaii and can transfer his skills to just about any place in the world.
“The next time you are in an accident and are laying on the side of the road, look forward to contacting an AI Bot to report your claim. ENJOY.”
If you are laying on the side of the road after an accident and your first thought is about reporting a claim then I don’t think you’re looking at the problem with the proper framing.
Now if your 911 call is being funneled through an AI bot, then yes, dystopia has arrived. But I don’t think we’re there (yet).
I don’t really see a downgrade from current customer service script readers to an AI Bot.
Advice I received from a very successful PI atty….
In an accident:
1. Take the ambulance to the hospital
2. Call your atty
Many injuries only reveal themselves days later.
Missing data is only a problem for saps like us in the ‘reality based community’.
and the thru-line from Rove/Cheney/Bush to the current state of affairs was never more clear.
We need data! Miran said in an interview we should lower interest rates because it takes 12 18 months before we see the outcome in the economy. I wish the the interviewer would have followed up “ so how much inflation are we expecting from the sept 2024 50 point cut!” It’s coming the pump in primed for inflation . We should have confirmation at close that it’s official “sell future rally”, the bubble is over, the crash is coming! . Yesterday, I asked a plumber how the economy was he said horrible, but his rates were inflated to the moon from last time I needed plumber services. After the dot com peak in March of 2000, we had a nasdaq 10% down day in April of 2000. Yeah that could happen again! Maybe a circuit breaker day this time. I will send a check or PayPal today, thanks wolf!!
Does Miran even begin to comprehend the difference between short-term interest rate and long-term interest rates? Doesn’t he realize that the Federal Reserve only sets short-term interest rates? If not, why not?
Dow industrials down 500 points as U.S. stocks face new wave of selling on heels of worst stock-market day in a month
SPX -0.46%
DJIA -0.90%
COMP -0.47%
AI hallucination — I just searched for the CUSIP number of a 10-year Treasury note maturing this month.
Google AI hallucinated in perfectly fluent bullshit:
“The CUSIP for the 10-year Treasury note maturing on November 15, 2025, is 91282CFW6. This security has a 4.5% coupon rate….”
However, when you look up CUSIP 91282CFW6, it’s actually a 3-year note issued in 2022 maturing November 15, 2025 🤣
The correct CUSIP for the 10-year note that matured Nov 15, 2025: 912828M56 (and the coupon rate was 2.25%, LOL)
AI is like a virus that turns brains to mush.
Thats hilarious…really hilarious and completely gut wrenching at the same time.
These Large Language Models work simply by predicting what the next reasonable word would be in a response text. “Reasonable” is simply defined as “statistically probable or plausible” within the corral of millions of tokens and weightings.
There is absolutely no thinking involved. The Engineers and scientists probably cannot provide a concrete definition of thinking. I’ve often wondered if General Intelligence is even possible without consciousness – which is undefinable in any way that would allow someone to build a machine that does it.
What a great example….thanks for sharing. Incredible.
The basic fact that “Current AI is pretty close to just being really, really suped up regression models…” has been pretty clear right from the November 2022 hype-cycle lift-off.
But “AI as God/Destroyer of Worlds” permeated the actual, real world for a good 12-18 months without serious push-back.
Good example of psychopathic CEOs being able to foist/bully BS into the world if it is in their interest (see also, too much of Wall Street…).
Let’s turn it into a teaching lesson. Asked chat. What happened.
“ Why this happens
AI models (including Google’s and others) often:
• Don’t have reliable access to real-time financial databases
• Try to infer an answer by pattern, not retrieval
• Prefer to generate something plausible rather than admit ignorance
CUSIP numbers cannot be reliably guessed, so the model made one up.”
Who needs the data…just ask ChatGPT.
Today was great day for yields at the long end.
Maybe it was Chicago dumping. lmao!
Short-term yields rose too as the rate cut in December is vanishing off the horizon.
1:04 PM 11/14/2025
Dow 47,147.48 -309.74 -0.65%
S&P 500 6,734.11 -3.38 -0.05%
Nasdaq 22,900.59 +30.23 0.13%
VIX 20.11 +0.11 0.55%
Gold 4,088.50 -106.00 -2.53%
Oil 59.97 +1.28 2.18%
I had a car parked close to a fire hydrant next to my house. Called AI (ChatGPT) and asked how far was it legally to be parked to a fire hydrant (the minimum distance). Got the wrong answer as I later found out when I called the local police department. The answer from AI was WAY OFF. Trust but Verify when it comes to AI.
Trust that the world is spending trillions of dollars on AI for a good reason and 99% of the people have no idea why.