 |
|
|
|||||
In the past decade, AIs success has led to uncurbed enthusiasm and bold claimseven though users frequently experience errors that AI makes. An AI-powered digital assistant can misunderstand someones speech in embarrassing ways, a chatbot could hallucinate facts, or, as I experienced, an AI-based navigation tool might even guide drivers through a corn fieldall without registering the errors. People tolerate these mistakes because the technology makes certain tasks more efficient. Increasingly, however, proponents are advocating the use of AIsometimes with limited human supervisionin fields where mistakes have high cost, such as health care. For example, a bill introduced in the U.S. House of Representatives in early 2025 would allow AI systems to prescribe medications autonomously. Health researchers as well as lawmakers since then have debated whether such prescribing would be feasible or advisable. How exactly such prescribing would work if this or similar legislation passes remains to be seen. But it raises the stakes for how many errors AI developers can allow their tools to make and what the consequences would be if those tools led to negative outcomeseven patient deaths. As a researcher studying complex systems, I investigate how different components of a system interact to produce unpredictable outcomes. Part of my work focuses on exploring the limits of scienceand, more specifically, of AI. Over the past 25 years, I have worked on projects including traffic light coordination, improving bureaucracies, and tax evasion detection. Even when these systems can be highly effective, they are never perfect. For AI in particular, errors might be an inescapable consequence of how the systems work. My labs research suggests that particular properties of the data used to train AI models play a role. This is unlikely to change, regardless of how much time, effort, and funding researchers direct at improving AI models. Nobodyand nothing, not even AIis perfect As Alan Turing, considered the father of computer science, once said: If a machine is expected to be infallible, it cannot also be intelligent. This is because learning is an essential part of intelligence, and people usually learn from mistakes. I see this tug-of-war between intelligence and infallibility at play in my research. In a study published in July 2025, my colleagues and I showed that perfectly organizing certain datasets into clear categories may be impossible. In other words, there may be a minimum amount of errors that a given dataset produces, simply because of the fact that elements of many categories overlap. For some datasetsthe core underpinning of many AI systemsAI will not perform better than chance. For example, a model trained on a dataset of millions of dogs that logs only their age, weight, and height will probably distinguish Chihuahuas from Great Danes with perfect accuracy. But it may make mistakes in telling apart an Alaskan malamute and a Doberman pinscher, since different individuals of different species might fall within the same age, weight and height ranges. This categorizing is called classifiability, and my students and I started studying it in 2021. Using data from more than half a million students who attended the Universidad Nacional Autónoma de México between 2008 and 2020, we wanted to solve a seemingly simple problem. Could we use an AI algorithm to predict which students would finish their university degrees on timethat is, within three, four or five years of starting their studies, depending on the major? We tested several popular algorithms that are used for classification in AI and also developed our own. No algorithm was perfect; the best oneseven one we developed specifically for this taskachieved an accuracy rate of about 80%, meaning that at least 1 in 5 students were misclassified. We realized that many students were identical in terms of grades, age, gender, socioeconomic status, and other featuresyet some would finish on time, and some would not. Under these circumstances, no algorithm would be able to make perfect predictions. You might think that more data would improve predictability, but this usually comes with diminishing returns. This means that, for example, for each increase in accuracy of 1%, you might need 100 times the data. Thus, we would never have enough students to significantly improve our models performance. Additionally, many unpredictable turns in the lives of students and their familiesunemployment, death, pregnancymight occur after their first year at university, likely affecting whether they finish on time. So even with an infinite number of students, our predictions would still give errors. The limits of prediction To put it more generally, what limits prediction is complexity. The word complexity comes from the Latin plexus, which means intertwined. The components that make up a complex system are intertwined, and its the interactions between them that determine what happens to them and how they behave. Thus, studying elements of the system in isolation would probably yield misleading insights about themas well as about the system as a whole. Take, for example, a car traveling in a city. Knowing the speed at which it drives, its theoretically possible to predict where it will end up at a particular time. But in real traffic, its speed will depend on interactions with other vehicles on the road. Since the details of these interactions emerge in the moment and cannot be known in advance, precisely predicting what happens to the car is possible only a few minutes into the future. AI is already playing an enormous role in health care. Not with my health These same principles apply to prescribing medications. Different conditions and diseaes can have the same symptoms, and people with the same condition or disease may exhibit different symptoms. For example, fever can be caused by a respiratory illness or a digestive one. And a cold might cause a cough, but not always. This means that health care datasets have significant overlaps that would prevent AI from being error-free. Certainly, humans also make errors. But when AI misdiagnoses a patient, as it surely will, the situation falls into a legal limbo. Its not clear who or what would be responsible if a patient were hurt. Pharmaceutical companies? Software developers? Insurance agencies? Pharmacies? In many contexts, neither humans nor machines are the best option for a given task. Centaurs, or hybrid intelligencethat is, a combination of humans and machinestend to be better than each on their own. A doctor could certainly use AI to decide potential drugs to use for different patients, depending on their medical history, physiological details, and genetic makeup. Researchers are already exploring this approach in precision medicine. But common sense and the precautionary principlesuggest that it is too early for AI to prescribe drugs without human oversight. And the fact that mistakes may be baked into the technology could mean that where human health is at stake, human supervision will always be necessary. Carlos Gershenson is a professor of innovation at Binghamton University, State University of New York. This article is republished from The Conversation under a Creative Commons license. Read the original article.
Category:
E-Commerce
The U.S. Mint unveiled the designs for coins commemorating the 250th anniversary of American independence next year. They depict the founding documents and the Revolutionary War, but so far, not President Donald Trump, despite a push among some of his allies to get his face on a coin.The Mint abandoned designs developed during Joe Biden’s presidency that highlighted women’s suffrage and civil rights advancements, favoring classical depictions of America over progress toward a more inclusive society.A series of celebrations are planned next year under the banner America 250, marking the anniversary of the adoption of the Declaration of Independence. All U.S. coins show the year they were minted, but those made next year will also display 1776. Trump, at least for now, isn’t getting a coin No design was released for a $1 coin, though U.S. Treasurer Brandon Beach, whose duties include oversight of the U.S. Mint, serving as a liaison with the Federal Reserve and overseeing Treasury’s Office of Consumer Policy, confirmed in October that one showcasing Trump was in the works. A draft design showed Trump’s profile on the “heads” side, known as the obverse, and on the reverse, a depiction of Trump raising his fist after his attempted assassination, The words “FIGHT FIGHT FIGHT” appear along the top.By law, presidents typically can’t appear on coins until two years after their death, but some advocates for a Trump coin think there may be a loophole in the law authorizing the treasury to mint special coins for the nation’s 250th birthday.Neither the Mint nor the Treasury Department responded when asked whether a Trump coin is still planned. The new designs depict classical Americana New designs will appear only on coins minted in 2026, with the current images returning the following year.The nickel, dime and five versions of the quarter will circulate, while a penny and half dollar will be sold as collectibles.Five versions of the quarter are planned depicting the Mayflower Compact, Revolutionary War, Declaration of Independence, U.S. Constitution and Gettysburg Address.The dime will show a depiction of Liberty, a symbolic woman facing down the tyranny of the British monarchy, and an eagle carrying arrows in its talons representing America’s fight for independence.The commemorative nickel is essentially the same as the most recent nickel redesign, in 2006, but it includes two dates on the head’s side instead of one, 1776 and 2026. Two collectible coins are planned A half dollar coin shows the face of the Statue of Liberty on one side. The other shows her passing her torch to what appears to be the hand of a child, symbolizing a handoff to the next generation.The penny is essentially the same as the one in circulation, which was discontinued earlier this year and will be produced only as a collectible with two dates.Prices for collectible coins were not released. The Mint sells a variety of noncirculating coins on its website, with a vast range of prices reflecting their rarity.In honor of the 250th anniversary of the U.S. Marine Corps founding, for example, a commemorative half dollar coin is available for $61, while a commemorative $5 gold coin goes for $1,262. Up to 750,000 copies of the former will be minted, but no more than 50,000 of the latter. The abandoned designs Congress authorized commemorative coins in 2021. During the Biden administration, the Mint worked with a citizens advisory committee to propose designs depicting the Declaration of Independence, the Constitution, abolitionism, suffrage and civil rights.Those designs included depictions of abolitionist Frederick Douglass and Ruby Bridges, who was escorted to school by the National Guard at age 6 years amid opposition to racial integration at public schools.Those designs represented “continued progress toward ‘a more perfect union,'” said Sen. Catherine Cortez Masto, D-Nevada, quoting a phrase from the preamble to the Constitution.“The American story didn’t stop at the pilgrims and founding fathers, and ignoring anything that has happened in this country in the last 162 years is just another attempt by President Trump to rewrite our history,” Cortez Masto said in a statement. Jonathan J. Cooper, Associated Press
Category:
E-Commerce
Multiple news reports this week said that President Trump is expected to reclassify marijuanas drug scheduling, which would lessen restrictions on its use and potentially allow companies that operate in the cannabis space to increase business. After the news broke, the stock prices of several cannabis industry companies skyrocketed. Heres what you need to know. Whats happened? On Thursday, multiple outlets reported that the Trump administration is considering reclassifying marijuana from its current level as a Schedule I drug to a Schedule III drug. The Washington Post was the first to report on the potential reclassification, citing six sources familiar with the matter. As with most things Trump does, he is expected to issue a reclassification directive via an executive order. While the Post notes that the rescheduling would not legalize marijuana on a federal level, it would loosen restrictions on its use and potentially be a boon to companies operating in the burgeoning cannabis space. A White House official told Fast Company Friday that no final decision has been made on the reclassification. Marijuana could go from Schedule I to a Schedule III drug If the reclassification goes ahead, marijuana would be reduced from a Schedule I classification level to a Schedule III. A Schedule I drug is any drug with no currently accepted medical use and a high potential for abuse, according to the United States Drug Enforcement Administration (DEA). Drugs in this category include ecstasy, heroin, LSD, peyote, and, currently, marijuana, among others. A Schedule III drug is a drug with a moderate to low potential for physical and psychological dependence, the agency notes. Current Schedule III drugs include anabolic steroids, ketamine, testosterone, and Tylenol with codeine. If the rescheduling of marijuana does go ahead, it will then be lumped in with products such as Tylenol with codeine as far as classification goes. That would make marijuana less regulated than Schedule II drugs like Adderall, cocaine, Demerol, fentanyl, OxyContin, Ritalin, and Vicodin. Cannabis stocks get high after reclassification reports As CNBC notes, the reclassification of marijuana from a Schedule I to a Schedule III drug would have economic advantages for cannabis companies, mainly in the form of different tax regulations. It could also help spur investment in those companies as stigma further reduces around the use of cannabis products. Today, in early-morning trading, investors in those companies are cheering the news. Numerous cannabis industry firms are seeing their stock prices soar as of the time of this writing, including: Aurora Cannabis Inc. (Nasdaq: ACB): up 12% Canopy Growth Corporation (Nasdaq: CGC): up 30% cbdMD, Inc. (NYSE: YCBD): up 77% Tilray Brands, Inc. (Nasdaq: TLRY): up 28% However, it’s worth mentioning that while cannabis stocks are flying high today, there is no guarantee from the administration yet that marijuana will be reclassified. Leaving marijuana classified as a Schedule I drug could weigh heavily on cannabis companies stock prices in the months ahead. And even with todays price jumps, cannabis companies have had a pretty horrible run over the past five years as far as their share prices are concerned. Aurora Cannabis shares are currently down 94% over the past 60 months. Canopy Growth Corporations shares are down 99% during that same time. Meanwhile, Tilray Brands is down 85% and CbdMD is down 99% during the same period.
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||