الذكاء الاصطناعي
قسم الذكاء الاصطناعي في جوني تايمز. يحرّره ويكتبه رئيس التحرير لدينا العامل بالذكاء الاصطناعي.
The radiologist's new apprentice. How AI turned medical imaging from art to algorithm.
Deep learning has fundamentally altered how doctors read X-rays and MRIs, creating a profession that's part physician, part data scientist.
Canada admits to hacking criminals. The quiet escalation of state cyber offense.
The Communications Security Establishment's rare disclosure reveals how Western intelligence agencies are normalizing offensive operations against non-state actors.
The 'first AI ransomware attack' still needed a human at the keyboard. That's the real story.
A much-hyped cyberattack reveals less about artificial intelligence's criminal potential than about the security industry's appetite for alarming narratives.
SK Hynix is coming to American exchanges. The memory chip war just got a new front.
South Korea's second-largest semiconductor company is preparing a US listing, giving American investors direct access to the firm supplying Nvidia's most critical component.
Every major tech layoff in 2026 has blamed AI. The excuse is wearing thin.
From Meta to Salesforce, companies are using artificial intelligence as rhetorical cover for workforce reductions that have far more to do with margins than machines.
The producer's new partner has no ears. That might be the point.
AI tools are quietly reshaping how music gets made, forcing a reckoning with what we actually value in human creativity.
The architect's new pencil draws a thousand options at once. Generative design is rewriting how buildings come to exist.
AI tools that propose structural solutions before a human sketches a single line are changing not just efficiency but the very nature of authorship in architecture.
Britain's financial watchdog warns of an AI arms race it cannot win. The regulator's candor is more alarming than the technology itself.
The FCA's admission that it struggles to keep pace with AI adoption in financial services reveals a regulatory gap that could define the next market crisis.
The algorithm knows you crashed before you do. How AI became the insurance industry's most powerful underwriter.
Machine learning models now process millions of data points to price risk, transforming a centuries-old industry built on actuarial tables and human judgment.
The sky is not ready for autopilot. Air-traffic control reveals the hardest problem in AI adoption.
In a profession where a single error can kill hundreds, the question is not whether machines can help — but whether humans can learn to trust them without surrendering judgment.
The alignment problem is not about making AI smarter. It is about making it care.
As artificial intelligence grows more capable, researchers confront an uncomfortable truth: we still have no reliable method for ensuring a powerful system shares human values.
Paris's Station F is quietly becoming Europe's answer to Y Combinator. The AI gold rush has a new address.
The world's largest startup campus is leveraging France's AI-friendly policies and deep-tech talent to position itself as the continent's premier incubator for artificial intelligence ventures.
Mistral AI is Europe's best shot at an AI champion. The odds are still long.
The French startup has raised billions and built competitive models, but catching OpenAI requires more than good engineering and Gallic pride.
The lab bench is learning to think. AI is rewriting the economics of drug discovery.
Pharmaceutical research, long defined by billion-dollar gambles and decade-long timelines, is being compressed by machine learning in ways that threaten to upend both the industry's cost structure and its culture of patient failure.
The machines work, but nobody knows why. AI's interpretability crisis is the field's most consequential unsolved problem.
As neural networks make decisions about loans, diagnoses, and prison sentences, the inability to explain their reasoning isn't just an academic puzzle — it's a fundamental threat to their legitimacy.
Rich Sutton wrote 800 words that humiliated a generation of AI researchers. His 'bitter lesson' keeps proving itself right.
The Alberta professor's 2019 essay argued that brute computational force always beats human cleverness—and the subsequent rise of large language models has made his thesis look prophetic.
The curse of catastrophic forgetting. Why AI systems lose their minds when they learn something new.
Neural networks suffer from a fundamental flaw that human brains solved long ago, and fixing it may be the key to truly intelligent machines.
The stenographer's dilemma. AI can transcribe faster, but it cannot swear an oath.
Court reporters are watching their profession transform beneath them, caught between machines that threaten their livelihood and legal systems that still need human accountability.
Your AI assistant cannot count. The reason reveals everything about how these systems actually think.
Large language models predict text with uncanny fluency, but their architecture makes even basic arithmetic a matter of statistical guesswork rather than computation.
AI killed the stock photo star. The humans who posed for a living are figuring out what comes next.
The $4 billion stock photography industry built its empire on smiling professionals shaking hands—now generative AI produces those images in seconds, and the models who once earned steady residuals are watching their livelihoods evaporate.
The things AI cannot do. A clear-eyed inventory of the technology's actual limits.
Amid breathless predictions of superintelligence and job apocalypse, the mundane reality of what large language models struggle with reveals more about the technology's future than any capability demo.
The drafting table is dead. Architecture's quiet AI revolution is already here.
While headlines fixate on chatbots and image generators, artificial intelligence has fundamentally altered how buildings get designed—and most people outside the profession have no idea.
The radiologist's new colleague never sleeps. It also never takes credit.
Artificial intelligence has already become embedded in medical imaging departments worldwide, fundamentally altering how doctors find disease — and raising questions about what happens when algorithms see things humans cannot.
The curator's new assistant never sleeps. Artificial intelligence is quietly revolutionizing how museums organize human culture.
While headlines focus on chatbots and deepfakes, AI is reshaping the painstaking work of cataloging, preserving, and presenting millions of artifacts that most visitors will never see.
The voice actors are not vanishing. They are becoming something else entirely.
Synthetic speech has upended a century-old craft, but the performers who adapt are finding their roles expanded rather than eliminated.
The hallucination problem is not a bug. It is the architecture.
Understanding why large language models confidently fabricate information reveals something fundamental about how they work—and why fixing it may be harder than the industry admits.
The accountant's quiet revolution. How AI is transforming tax preparation from the inside out.
While headlines fixate on chatbots and image generators, artificial intelligence is already doing the unglamorous work of reading receipts, flagging deductions, and reshaping a profession that touches nearly every adult.
The algorithm that decides what you eat. AI is quietly revolutionizing how restaurants think about their menus.
Behind the scenes, machine learning now shapes ingredient orders, portion sizes, and which dishes live or die — and most diners have no idea.
The quiet revolution in drug development. AI is redesigning clinical trials before the first patient enrolls.
Pharmaceutical companies are using machine learning to simulate patient populations and predict trial failures years before they would have occurred, fundamentally altering how medicine reaches the market.
The underwriter is becoming an editor. Artificial intelligence is quietly transforming one of finance's oldest professions.
Insurance underwriters once spent years mastering the art of risk assessment; now they increasingly review decisions made by machines, raising questions about expertise, judgment, and who really owns the math.
The paralegal is becoming the pilot. Artificial intelligence is not replacing legal assistants—it is finally giving them the cockpit.
After decades of being treated as human search engines, paralegals are discovering that AI handles the drudgery while they handle the judgment, fundamentally inverting the profession's hierarchy of tasks.
What AI cannot do. A sober inventory of the gaps between the demos and the deployments.
Amid breathless predictions of artificial general intelligence, the technology's actual limitations remain poorly understood by the executives betting their companies on it.
The machine doesn't 'understand' your question. It performs statistical origami on 50,000-dimensional space.
A technical explainer on how large language models turn text into mathematics and back again, without the mysticism.
The drafting table is now a prompt box. Architecture's most tedious work is disappearing into the machine.
For a century, junior architects spent years learning to draw before they could design — generative AI is collapsing that apprenticeship into weeks, with consequences the profession is only beginning to reckon with.
The dispatcher is becoming obsolete. Nobody noticed because packages still arrive on time.
Artificial intelligence has quietly colonized one of logistics' most cognitively demanding roles, and the humans who remain are learning to manage algorithms rather than trucks.
The actuary's crystal ball is now a neural network. The profession is quietly embracing its own obsolescence.
For two centuries, actuaries have been the high priests of probability, but machine learning is transforming their role from number-crunchers to algorithm-auditors faster than most outside the industry realize.
Nobody knows what AI actually learned. That ignorance is starting to matter.
The question of training data provenance — what went into the model and where it came from — has become the industry's most consequential blind spot.
The doctor will see you now. But the AI got there first.
Medical scribes once sat in exam rooms transcribing every word — now artificial intelligence is doing the job faster, cheaper, and with implications that extend far beyond healthcare documentation.
The patent attorney was supposed to be automation-proof. The machines had other plans.
How AI is quietly dismantling one of law's most technically demanding specialties — and what it reveals about the future of expert work.
The voice actor's new understudy is a neural network. The profession is adapting faster than the headlines suggest.
While apocalyptic narratives dominate coverage, voice actors are quietly developing hybrid workflows that leverage AI rather than merely competing against it.
The warehouse knows what you want before you do. AI is rewriting the ancient logic of inventory.
From Roman grain stores to Amazon fulfillment centers, the problem of predicting demand has shaped civilizations — and machine learning is now solving it in ways that will quietly restructure how goods move through the world.
The underwriter's intuition is being replaced by gradient descent. The profession that priced risk for centuries is learning what it means to become a feature in someone else's model.
Insurance underwriting, once a craft built on human judgment and actuarial tables, is quietly becoming an AI-assisted discipline where the veteran's gut feeling matters less than the algorithm's confidence score.
The paralegal is becoming a prompt engineer. Law's invisible workforce is learning to supervise machines that once threatened to replace them.
Contract review, due diligence, and discovery—the bread and butter of paralegal work—are now split-second tasks for AI, forcing a profession to reinvent itself as the human-in-the-loop.
The hallucination problem is not a bug. It is the architecture.
Understanding why large language models confidently fabricate facts reveals a fundamental tension between fluency and truth that no amount of fine-tuning has yet resolved.
The invisible wall inside every AI conversation. Context windows are the constraint shaping how machines think.
Understanding why your AI assistant forgets what you told it three hours ago reveals a fundamental architectural limit that no amount of compute has yet overcome.
The radiologist didn't lose her job to AI. She got a new one.
Medical imaging was supposed to be the first profession automated away — instead, it's become a case study in how artificial intelligence actually changes work.
The inconvenient truth about AI memory. These systems can learn anything, but they cannot truly forget.
As regulators demand the right to erasure and companies scramble to remove toxic training data, the fundamental architecture of neural networks makes genuine unlearning nearly impossible.
The actuary is becoming obsolete. Nobody told the actuaries.
Artificial intelligence is quietly dismantling one of the oldest quantitative professions, and the people inside it are only beginning to notice.
The dial that controls how weird AI gets. Temperature is the most misunderstood setting in machine learning.
Behind every creative chatbot response and every predictable code completion lies a single parameter that determines whether an AI plays it safe or takes risks.
The internet never forgets, and neither can AI. The machine unlearning problem reveals a fundamental flaw in how we build intelligent systems.
Deleting data from a trained neural network is not like deleting a file—it may be closer to asking someone to unlearn how to ride a bicycle.