Business
Column: These Apple researchers just showed that AI bots can't think, and possibly never will
See if you can solve this arithmetic problem:
Oliver picks 44 kiwis on Friday. Then he picks 58 kiwis on Saturday. On Sunday, he picks double the number of kiwis he did on Friday, but five of them were a bit smaller than average. How many kiwis does Oliver have?
If you answered “190,” congratulations: You did as well as the average grade school kid by getting it right. (Friday’s 44 plus Saturday’s 58 plus Sunday’s 44 multiplied by 2, or 88, equals 190.)
You also did better than more than 20 state-of-the-art artificial intelligence models tested by an AI research team at Apple. The AI bots, they found, consistently got it wrong.
The fact that Apple did this has gotten a lot of attention, but nobody should be surprised at the results.
— AI critic Gary Marcus
The Apple team found “catastrophic performance drops” by those models when they tried to parse simple mathematical problems written in essay form. In this example, the systems tasked with the question often didn’t understand that the size of the kiwis have nothing to do with the number of kiwis Oliver has. Some, consequently, subtracted the five undersized kiwis from the total and answered “185.”
Human schoolchildren, the researchers posited, are much better at detecting the difference between relevant information and inconsequential curveballs.
The Apple findings were published earlier this month in a technical paper that has attracted widespread attention in AI labs and the lay press, not only because the results are well-documented, but also because the researchers work for the nation’s leading high-tech consumer company — and one that has just rolled out a suite of purported AI features for iPhone users.
“The fact that Apple did this has gotten a lot of attention, but nobody should be surprised at the results,” says Gary Marcus, a critic of how AI systems have been marketed as reliably, well, “intelligent.”
Indeed, Apple’s conclusion matches earlier studies that have found that large language models, or LLMs, don’t actually “think” so much as match language patterns in materials they’ve been fed as part of their “training.” When it comes to abstract reasoning — “a key aspect of human intelligence,” in the words of Melanie Mitchell, an expert in cognition and intelligence at the Santa Fe Institute — the models fall short.
“Even very young children are adept at learning abstract rules from just a few examples,” Mitchell and colleagues wrote last year after subjecting GPT bots to a series of analogy puzzles. Their conclusion was that “a large gap in basic abstract reasoning still remains between humans and state-of-the-art AI systems.”
That’s important because LLMs such as GPT underlie the AI products that have captured the public’s attention. But the LLMs tested by the Apple team were consistently misled by the language patterns they were trained on.
The Apple researchers set out to answer the question, “Do these models truly understand mathematical concepts?” as one of the lead authors, Mehrdad Farajtabar, put it in a thread on X. Their answer is no. They also pondered whether the shortcomings they identified can be easily fixed, and their answer is also no: “Can scaling data, models, or compute fundamentally solve this?” Farajtabar asked in his thread. “We don’t think so!”
The Apple research, along with other findings about the limitations of AI bots’ cogitative limitations, is a much-needed corrective to the sales pitches coming from companies hawking their AI models and systems, including OpenAI and Google’s DeepMind lab.
The promoters generally depict their products as dependable and their output as trustworthy. In fact, their output is consistently suspect, posing a clear danger when they’re used in contexts where the need for rigorous accuracy is absolute, say in healthcare applications.
That’s not always the case. “There are some problems which you can make a bunch of money on without having a perfect solution,” Marcus told me. Recommendation engines powered by AI — those that steer buyers on Amazon to products they might also like, for example. If those systems get a recommendation wrong, it’s no big deal; a customer might spend a few dollars on a book he or she didn’t like.
“But a calculator that’s right only 85% of the time is garbage,” Marcus says. “You wouldn’t use it.”
The potential for damagingly inaccurate outputs is heightened by AI bots’ natural language capabilities, with which they offer even absurdly inaccurate answers with convincingly cocksure elan. Often they double down on their errors when challenged.
These errors are typically described by AI researchers as “hallucinations.” The term may make the mistakes seem almost innocuous, but in some applications, even a minuscule error rate can have severe ramifications.
That’s what academic researchers concluded in a recently published analysis of Whisper, an AI-powered speech-to-text tool developed by OpenAI, which can be used to transcribe medical discussions or jailhouse conversations monitored by correction officials.
The researchers found that about 1.4% of Whisper-transcribed audio segments in their sample contained hallucinations, including the addition to transcribed conversation of wholly fabricated statements including portrayals of “physical violence or death … [or] sexual innuendo,” and demographic stereotyping.
That may sound like a minor flaw, but the researchers observed that the errors could be incorporated in official records such as transcriptions of court testimony or prison phone calls — which could lead to official decisions based on “phrases or claims that a defendant never said.”
Updates to Whisper in late 2023 improved its performance, the researchers said, but the updated Whisper “still regularly and reproducibly hallucinated.”
That hasn’t deterred AI promoters from unwarranted boasting about their products. In an Oct. 29 tweet, Elon Musk invited followers to submit “x-ray, PET, MRI or other medical images to Grok [the AI application for his X social media platform] for analysis.” Grok, he wrote, “is already quite accurate and will become extremely good.”
It should go without saying that, even if Musk is telling the truth (not an absolutely certain conclusion), any system used by healthcare providers to analyze medical images needs to be a lot better than “extremely good,” however one might define that standard.
That brings us to the Apple study. It’s proper to note that the researchers aren’t critics of AI as such but believers that its limitations need to be understood. Farajtabar was formerly a senior research scientist at DeepMind, where another author interned under him; other co-authors hold advanced degrees and professional experience in computer science and machine learning.
The team plied their subject AI models with questions drawn from a popular collection of more than 8,000 grade school arithmetic problems testing schoolchildren’s understanding of addition, subtraction, multiplication and division. When the problems incorporated clauses that might seem relevant but weren’t, the models’ performance plummeted.
That was true of all the models, including versions of the GPT bots developed by OpenAI, Meta’s Llama, Microsoft’s Phi-3, Google’s Gemma and several models developed by the French lab Mistral AI.
Some did better than others, but all showed a decline in performance as the problems became more complex. One problem involved a basket of school supplies including erasers, notebooks and writing paper. That requires a solver to multiply the number of each item by its price and add them together to determine how much the entire basket costs.
When the bots were also told that “due to inflation, prices were 10% cheaper last year,” the bots reduced the cost by 10%. That produces a wrong answer, since the question asked what the basket would cost now, not last year.
Why did this happen? The answer is that LLMs are developed, or trained, by feeding them huge quantities of written material scraped from published works or the internet — not by trying to teach them mathematical principles. LLMs function by gleaning patterns in the data and trying to match a pattern to the question at hand.
But they become “overfitted to their training data,” Farajtabar explained via X. “They memorized what is out there on the web and do pattern matching and answer according to the examples they have seen. It’s still a [weak] type of reasoning but according to other definitions it’s not a genuine reasoning capability.” (the brackets are his.)
That’s likely to impose boundaries on what AI can be used for. In mission-critical applications, humans will almost always have to be “in the loop,” as AI developers say—vetting answers for obvious or dangerous inaccuracies or providing guidance to keep the bots from misinterpreting their data, misstating what they know, or filling gaps in their knowledge with fabrications.
To some extent, that’s comforting, for it means that AI systems can’t accomplish much without having human partners at hand. But it also means that we humans need to be aware the tendency of AI promoters to overstate their products’ capabilities and conceal their limitations. The issue is not so much what AI can do, but how users can be gulled into thinking what it can do.
“These systems are always going to make mistakes because hallucinations are inherent,” Marcus says. “The ways in which they approach reasoning are an approximation and not the real thing. And none of this is going away until we have some new technology.”
Business
Column: The Hoover Institution says all recent California job growth has been in government jobs. That's completely wrong
Back when most sensible Californians were concerning themselves with Thanksgiving preparations, the California-bashing right wing went hog wild over a stunning report that almost all private job growth in the state collapsed from January 2022 to June 2024 and almost all growth — 96.5% — was in government jobs.
“California’s Businesses Stop Hiring,” was the headline on the report published by the conservative Hoover Institution. Its main claim was that from January 2022 to June 2024, private employers in the state added only 5,400 jobs.
You can imagine how California bashers, including some within the state, greeted the news that government was propping up the state’s economy.
“This is what a failing state looks like,” Rep. Kevin Kiley (R-Rocklin), who badly lost a bid to replace Gov. Gavin Newsom in the 2021 recall election, tweeted. Others who gleefully tweeted about the Hoover claim included Rep. Vince Fong (R-Bakersfield), and venture investor Steve Jurvetson. Right-wingers outside California also joined the choir.
The Hoover article was what we in the news biz often pigeonhole as “interesting, if true.”
But it’s not true.
The original article, by UCLA economics professor Lee Ohanian, a Hoover Institution senior fellow, asserted that California added only 156,000 nonfarm jobs in the January 2022-June 2024 period. Since government statistics also showed that government employment in the state rose by 150,500, that left (after rounding) only about 5,400 new jobs created outside the government sector.
The picture painted was one in which private employers are shutting down and only government hiring is keeping the California economy afloat. The opposite is true, however.
(The Hoover Institution has retracted the original article and removed it from its website. An archived version of the original can be found here.)
Here’s the main problem with the Hoover analysis: During the sample period, California actually added 672,300 nonfarm jobs, not 156,000. Consequently, the 150,500 new government jobs accounted for only about 22.4% of the total, not 96.5%. The accurate figures show that not only did California’s businesses not stop hiring, but continued to hire fairly robustly from January 2022 to June 2024.
How did this calculation go so awry? The answer is simple. Ohanian conflated the two separate monthly employment surveys issued by the Bureau of Labor Statistics: One is its so-called household survey, which asks a national sample of about 60,000 households how many people in the household are employed. The other is its establishment or “payroll” survey, which asks about 629,000 workplaces how many people they employ.
Generally, the household survey yields a higher number of employed persons than the establishment survey. That’s because it counts the self-employed (including gig workers) and farmworkers, among others who are excluded from the payroll statistics. But that relationship breaks down when you’re counting only payroll workers, slicing and dicing the statistics into industry sectors.
Mixing together the BLS household data and the BLS establishment data is “a cardinal sin of BLS data analysis,” observes the pseudonymous economics commentator Invictus on The Big Picture blog of Ritholtz Wealth Management, in an indispensable deconstruction of Ohanian’s original post.
In that post, Ohanian subtracted the government jobs figure reported in the establishment survey from the nonfarm employment figure in the household survey. That effectively overstated the government jobs percentage of California employment growth. The proper approach, Invictus notes, would have been to use the establishment survey for both measures.
Ohanian acknowledged in an email that he had erroneously considered the household and establishment figures similar enough to treat them as effectively equivalent. “If I had seen the differences in the two series,” he says, “I would have written the piece differently. Mea culpa.”
In a corrective article posted Tuesday on the Hoover website, Ohanian makes public his mea culpa but also reiterates a point he made in the original article, which is that California’s job growth is weakening. That’s echoed by other studies, including a recent warning from the state’s Legislative Analyst’s Office.
Yet there’s much more to be said about Ohanian’s original article, as well as the glee with which conservatives seized on its headline claim as the basis for largely groundless attacks on California’s economic policies. First, it’s proper to note that the original piece was published Aug. 7, which is why its analysis covers only the period that ended in June.
Why it got resurrected and shot around the right-wing echo chamber last week is a mystery. Ohanian himself seemed uncertain when I asked him about it. Kiley, Fong and Jurvetson haven’t responded to my requests for comment.
That brings us to the statistics themselves. Employment data bristle with pitfalls for the unwary, even among experienced economists such as Ohanian. Indeed, in April, Ohanian posted an analysis on the Hoover website that purported to show a loss of 10,000 fast-food jobs in California from September 2023, when Newsom signed a minimum wage increase for that sector, through January this year — even before the increase went into effect.
As I reported, Ohanian based his post on a Wall Street Journal article that used employment figures that weren’t seasonally adjusted. That’s a crucial error when tracking jobs in seasonal industries such as restaurants.
The Journal’s article, and consequently Ohanian’s, mistook a seasonal decline in restaurant employment that occurs from September to January every single year for the one-time consequences of the minimum wage increase. Fast-food jobs, seasonally adjusted, actually rose by 6,300 in the period being reported. Ohanian told me at the time that he had been unaware that the Journal used nonseasonally adjusted figures.
BLS employment figures may be especially confusing because the bureau’s two surveys superficially seem to measure the same thing, but are very different — so much so that the bureau itself has issued a detailed explainer about the distinction. It notes that the establishment survey is “a highly reliable gauge of monthly change in nonfarm payroll employment.” The household survey is oriented more toward demographics and is best known as the source of the national unemployment rate.
Ohanian used his misconstruction of employment figures as the basis for a wide-ranging critique of California economic policy, mostly citing how the high cost of living drives people out of the state.
“Part of California’s job weakness,” he wrote, “reflects the number of people and businesses leaving the state.” California’s population fell by about 75,000 from 2022 and 2023 (the latest data available), he wrote, adding that companies such as Tesla, Oracle, and Chevron have moved or are moving their headquarters elsewhere.
“Population loss naturally leads to job loss,” Ohanian told me by email. “It is challenging to see how California could be gaining jobs as portrayed in the Establishment Survey, given a smaller population.”
That may well be true over the longer term and with larger numbers. But the 75,000 departed residents in 2022-23 represent less than two hundredths of a percent of the state’s population. Even the larger population decline of about 538,000 since 2020 represents about 1.4% of the state’s population.
The key question would be: Who’s leaving? Many emigrants may be retirees, who don’t have occupational reasons to stay in the high-cost state and may have sizable equity in their homes to pocket for a move to a cheaper location; about 7.5 million of California’s residents today are older than 65. The pandemic also drove the population down — COVID-related deaths numbered at least 60,000 in 2020 and 2021.
As for the emigration of corporate headquarters, California still leads the nation in headquarters of Fortune 500 companies, with 57. New York and Texas were runners up with 52 each. California remains a national leader in business creation, with nearly 560,000 new business applications filed with the state in 2023. When new technologies emerge with the potential to aid economic expansion, they tend to start in California.
One other subtext of the debate over California job growth needs to be mentioned. That’s the picture that conservatives paint about government jobs. The tweeted hand-wringings about the purported explosion in government jobs, which implies that the government workers are an army of faceless bureaucrats engaged in writing anti-business regulations.
The idea that the Musk/Ramaswamy Department of Government Efficiency can cashier them without affecting your daily life is a fantasy. In fact, the federal government employs only about 3 million workers, about half of whom are in the military, the Department of Veterans Affairs, and the Department of Homeland Security; the overall figure has remained fairly stable since the 1960s.
An additional 20 million are state and local employees, the majority of whom are teachers, along with police and fire fighters. Which of these workers should we fire?
Any discussion of California’s economy limited to periods of a year or two needs to be viewed in relation to the big picture, which is that California’s economy is by far the biggest in the country — indeed, it would rank in the top five or six countries if it were a sovereign state. At an estimated $4.08 trillion in gross domestic product, its economy is more than half again as large as the runner-up among U.S. states, Texas ($2.7 trillion).
Ohanian is right to argue that there’s reason for concern about where the state goes from here. But to suggest that there’s something fundamentally faulty about policies that still undergird the most powerful state economy in the nation or that California is a “failing state” — that’s “interesting, if true” … but, again, not true.
Business
Glendale's ServiceTitan seeks to raise $500 million in IPO
ServiceTitan, a Glendale tech firm that makes business management software for plumbers, painters and other contractors, announced Tuesday that it wants to raise up to $502 million in its initial public offering on the Nasdaq stock exchange.
The company said it plans to offer 8.8 million shares that would be priced between $52 and $57 each, according to a regulatory filing. At the top of that range, ServiceTitan would have a market capitalization of $5.16 billion. The company was valued at $7.6 billion after a November 2022 funding round. ServiceTitan hasn’t said when it plans to start trading.
ServiceTitan was founded in 2007 by two college friends from Glendale, Ara Mahdessian, 39, and Vahe Kuzoyan, 41, whose fathers worked as contractors. The company previously raised about $1.4 billion from venture firms, including Iconiq Growth, Bessemer Venture Partners and Battery Ventures.
It counts about 8,000 contracting firms as customers, providing an end-to-end software suite that can manage booking appointments, generating estimates and processing invoices as well as payroll and dispatching workers. Clients range in size from family-owned contractors to large national franchises totaling more than 100,000 technicians. It charges a subscription fee for its services.
The company, which assists contractors nationwide, says it wants to expand the number of trades and markets it serves. It employed 2,870 workers as of July 31 at its Glendale headquarters and offices elsewhere in the U.S. and internationally.
Competitors include BuildOps, Housecall Pro, Jobber and other companies that charge subscriptions for their web-based business management software.
ServiceTitan had filed confidential paperwork for an $18-billion public offering in 2022, according to Business Insider, but didn’t move forward after the Federal Reserve sharply raised interest rates to combat inflation, freezing up the IPO market.
The company reported revenue of $614 million in the fiscal year that ended Jan. 31, up nearly a third from a year earlier, and an operating loss of $195 million, 28% less than in fiscal 2023. It had about $147 million in cash and equivalents on hand as of Jan. 31 and was carrying $175 million in long-term net debt.
The company’s share structure will ensure that control remains with the founders — Mahdessian is chief executive and Kuzoyan president — who will retain all Class B shares, which are entitled to 10 votes each.
Lead underwriters on the IPO are Goldman Sachs Group, Morgan Stanley, Wells Fargo and Citigroup. They have an option to buy an additional 1.32 million shares, which will be traded under the ticker symbol “TTAN.”
Bloomberg contributed to this report.
Business
Column: GOP and Musk unveil a threat to Social Security
You may have been tempted to believe Donald Trump when he swore, along with some of his Republican colleagues, to protect Social Security. If so, the joke may be on you.
That concern emerged Monday when Sen. Mike Lee (R-Utah) uncorked a tweet thread on X labeling Social Security “a classic bait and switch” and “an outdated, mismanaged system.”
Twenty-three minutes after Lee posted the first of his tweets, it was retweeted by Elon Musk, who has been vested by Trump with a portfolio to root out inefficiencies in the government. Musk led his retweet with the comment “interesting thread”; if that wasn’t an explicit endorsement, it matched his way of amplifying others’ tweets, tending to give them credibility within the Musk-iverse.
It will be my objective to phase out Social Security, to pull it out by the roots.
— Sen. Mike Lee (R-Utah)
Lee’s tweet thread, along with Musk’s apparent concurrence, serves as an outline of the arguments the GOP may use to undermine faith in Social Security, the better to soften it up for “reforms” that will translate into costs imposed on retirees, disabled workers and their dependents.
I recently reported on all the ways that Trump could quietly or secretly undermine his pledge to protect Social Security. Lee’s thread and Musk’s apparent endorsement are different — they amount to a frontal attack on the program.
While delving into Lee’s screed, we should keep in mind that he’s a leader of the cabal with the knives out for Social Security. As I’ve reported, during his first successful Senate campaign in 2010, he unapologetically declared, “It will be my objective to phase out Social Security, to pull it out by the roots.”
Lee said that was why he was running for the Senate, and added, “Medicare and Medicaid are of the same sort. They need to be pulled up.”
So here he is, right out of the box.
Lee’s attack has four basic components. One is to bemoan the fact that Social Security is funded mostly by a tax, which he asserts the government can use for any purpose — not necessarily to cover retirement and disability benefits.
Another is to point out that the program’s reserves aren’t stored in individual accounts with workers’ names on them, but collected in “a huge account called the ‘Social Security Trust Fund.’”
A third is to claim that “the government routinely raids this fund. … They take ‘your money’” and use it for whatever the current Congress deems ‘necessary.’”
And a fourth is to complain that the trust fund is mismanaged: “If you had put the same amount into literally ANYTHING else — a mutual fund, real estate, even a savings account — you’d be better off by the time you reached retirement age, even if the government kept some of it!” He states: “Your ‘investment’ in Social Security can give you a return lower than inflation.”
None of these is a new argument — they’ve been swirling around the conservative and Republican fever swamp like a miasma for decades. They’ve been consistently refuted and debunked. Lee can’t be unaware of that. Some of his arguments have a tiny nugget of truth at their center, but in his hands are twisted and manipulated out of recognition. Consequently, we can label his claims for what they are: Lies.
Let’s examine them one by one. (I asked Lee via a message at his office to justify his tweets, but haven’t heard back.)
Yes, Social Security is funded by taxes. So what? Lee’s salary as a senator is funded by taxes too. Does that make it illegitimate? It’s true that once a tax is collected Congress can decide to spend it however it wishes. But it’s also true that the payroll tax was enacted jointly with the provisions of the Social Security Act that designated the revenue for Social Security benefits.
As Supreme Court Justice Benjamin Cardozo observed in 1937, writing for the majority in a 7-2 opinion upholding the constitutionality of Social Security, it was clear that Congress intended the payroll tax to fund the benefits, for lawmakers “would have been unwilling to pass one without the other.”
It’s proper to note here that no one has ever proposed diverting Social Security revenues for any other purpose without recompense — except Republicans such as Lee. George W. Bush proposed converting Social Security into private accounts, which would have been tantamount to such a diversion — and a gift to Wall Street money managers eager to get their hands on the program’s trillions of dollars.
But Bush’s 2005 privatization plan was stillborn and he quickly abandoned it.
It’s also true that the program’s revenues aren’t stored in individual accounts but in the trust fund. That’s right and proper: Social Security is a shared benefit; no one can know in advance what any worker’s benefits will be. They’re pegged to career earnings, but low-income workers get higher benefits relative to wages than higher-income workers. They’re also related to a worker’s personal and family situation — spouses, dependents, health and so on.
It also makes sense to invest the program’s revenues in a shared account, because large investments tend to perform better over time than those under the control of individuals, not least because that minimizes transaction costs.
That brings us to the notion that the government “routinely raids” the trust funds (there are two, actually — one to cover old-age benefits and the other to cover disability payments — but they’re generally treated as a single combined fund). The trust funds currently hold about $2.8 trillion in assets, all invested in U.S. Treasury securities.
Holding a T-bond, as anyone with the slightest knowledge of government fiscal policy is aware, means the bondholder has lent the money to the government, which can use it for any purpose Congress chooses and which must pay interest on the bond. Over the years, the government has used the money to build roads and other infrastructure and provide services. Using the borrowed money for these purposes allows the government to do so without raising income taxes, which would hit the wealthy harder than middle- or low-income Americans.
Lee should ask his well-heeled patrons if they’d prefer to pay higher taxes because the government couldn’t borrow from the Social Security reserves. Anyone have any doubts about how they’d answer? Me neither.
In any event, the financial transactions related to the buying and redemption of the program’s Treasury holdings are fully disclosed every year by the program trustees in their annual report.
What about Lee’s assertion that investing in “ANYTHING else — a mutual fund, real estate, even a savings account,” would make you “better off by the time you reached retirement age.” This statement is as solid a compendium of financial ignorance as one might wish, even coming from a U.S. senator.
To begin with, if Lee thinks the Social Security trust fund should be invested in something other than Treasurys, he can take that up with his colleagues on Capitol Hill. They’re the ones who have mandated, by law, that the trust fund can be invested only in Treasurys. Over the years, proposals to widen the portfolio have been raised and abandoned, for several reasons. Some were concerned about the potential conflicts of interest inherent in a government program investing in the stock market; others that the returns from market investments are too volatile.
Savings accounts? Is Lee kidding? The rate on savings accounts offered to the average customer of Bank of America, to choose a commercial bank at random, is 0.01% a year. As I write, a 10-year Treasury bond yields about 4.2% annually.
As for Lee’s assertion that “Social Security can give you a return lower than inflation,” the fact is that Social Security benefits are adjusted for inflation every year. They’re also lifetime benefits. Try to find an annuity plan that pays inflation-adjusted benefits for the life of the annuity holder and his or her spouse — for all but the richest people, it would be unaffordable or at least uneconomical.
Lee also reveals a fundamental misunderstanding about Social Security as a program. It’s not just a retirement program, but a combined retirement and insurance program.
Disabled workers — and their dependents — are entitled to benefits well beyond their contributions; the families of workers who die before retirement age receive benefits that include payments for children through age 17 — through age 18 if they’re in school. If those benefits were based on the balances in a worker’s individual account, then the families of those who have suffered untimely deaths could receive a pittance, running out while still needing help.
Lee concludes by urging his followers to “acknowledge the truth: Social Security as it now exists isn’t a retirement plan; it’s a tax plan with retirement benefits as an afterthought.” This is an outright falsehood. As it now exists, Social Security isn’t just a retirement plan, but a disability program. It’s funded by taxes, but to call retirement benefits “an afterthought” is so wrong it’s frightening.
What should we think about all this? Lee is a member of the Senate majority; his proposals could be a real threat to the program. The fact that they garnered an “attaboy” from Elon Musk should be their death knell. Let’s hope so.
-
Technology1 week ago
Elon Musk targets OpenAI’s for-profit transition in a new filing
-
Sports1 week ago
One Black Friday 2024 free-agent deal for every MLB team
-
News1 week ago
Rassemblement National’s Jordan Bardella threatens to bring down French government
-
Technology1 week ago
9 ways scammers can use your phone number to try to trick you
-
World1 week ago
Georgian PM praises country's protest crackdown despite US condemnation
-
World6 days ago
Freedom is permanent for Missourian described as the longest-held wrongly incarcerated woman in US
-
Technology2 days ago
Struggling to hear TV dialogue? Try these simple fixes
-
Sports1 week ago
With F1 entry, General Motors has a shot to become America’s team on the grid