knowledge10.net

Category: technology

  • AI Doesn’t Generate Working Products, That’s Still Your Job

    Can humans and artificial intelligence work in one team - Pivot

    A human engineer at a desk, with AI-generated code snippets floating around, but the engineer is actively integrating and testing them, symbolizing human oversight.

    It’s tempting to think that with AI, you can just type a prompt and get a finished app. But the reality is more nuanced: AI tools are excellent at producing prototypes, drafts, and approximations—but they don’t deliver working products. The gap between a demo and a shippable product is where the real engineering work happens, and that’s still your job.

    This isn’t a pessimistic view; it’s a clarifying one. Understanding what AI can and cannot do helps you use it effectively, avoid costly mistakes, and focus your energy where it matters most. In this article, we’ll explore why AI output is a starting point, not an endpoint, and what that means for developers, managers, and learners.

    The Prototype Isn’t the Product

    When you ask an AI to generate code, you get a prototype—a plausible-looking piece of code that might work in a simple scenario. But a product is more than that. It’s reliable, secure, scalable, maintainable, and documented. AI models are trained to produce output that looks competent, but they can’t run, test, or reason about their own output in a production environment. They lack the ability to understand the full context of your system, your users, or your business requirements.

    Consider a simple example: you ask an AI to write a function that calculates the total price of items in a shopping cart. The AI might produce a straightforward loop that sums up the prices. It compiles, and it passes a basic test. But what happens when a user applies a discount code? Or when a product is out of stock? Or when the cart contains a mix of currencies? The AI didn’t anticipate these edge cases because it doesn’t know your business rules. That’s where you come in.

    The Demo Effect: Why AI Output Looks So Good

    AI models are trained on vast amounts of code and text, so they’re great at pattern matching. They’ve seen thousands of similar functions, so they can generate something that looks like a competent solution. This is called the “demo effect”: the output is designed to be plausible, not verified. It’s like a magician’s trick—it looks impressive, but it’s not real magic.

    This effect can fool stakeholders. A manager sees a demo of an AI-generated feature and thinks it’s almost done. But the demo is just a thin slice of functionality, often with hardcoded data and no error handling. The real work—integrating with existing systems, handling edge cases, testing for security, and deploying to production—is invisible and still ahead.

    The Real Work: Integration, Testing, and Maintenance

    The gap between a prototype and a product is where the value of a skilled engineer lies. It’s not just about writing code; it’s about understanding requirements, designing for scale, ensuring security, and maintaining the system over time. AI can help with the initial scaffolding, but it can’t make the countless decisions that go into a production-ready system.

    For example, an AI might generate a REST API endpoint that works in isolation. But to make it a product, you need to add authentication, rate limiting, input validation, logging, and monitoring. You need to write tests that cover not just the happy path but also failure modes. You need to consider how it will perform under load and how it will be deployed. These are not tasks that an AI can do for you—they require judgment and experience.

    The Skeptic’s View: AI as a Liability

    Some developers are more skeptical of AI-generated code. They point out that AI can introduce security vulnerabilities, licensing issues, and technical debt. Because AI models are trained on public code, they may reproduce patterns that are insecure or outdated. And because AI output is often uncommented and unstructured, it can be hard to maintain.

    There’s also a risk of over-reliance. Junior developers who lean too heavily on AI may not develop the debugging and system design skills they need. They might not understand why a piece of code works, only that it does—until it breaks in production. This is a real concern for the industry’s future.

    The Optimist’s View: A Shrinking Gap

    On the other hand, AI tools are improving rapidly. The gap between prototype and product is shrinking. Some argue that the human role is shifting from writing code to specifying intent and reviewing output. That’s still a job, but it’s a different one. In the future, AI might handle more of the pipeline, but for now, it’s a powerful assistant, not a replacement.

    What This Means for You

    Whether you’re a developer, a manager, or a learner, the key takeaway is this: use AI as a tool, not a crutch. Let it help you explore ideas, generate boilerplate, and speed up your workflow. But don’t expect it to do your job for you. The responsibility for the final product—its quality, security, and reliability—rests with you.

    For developers, this means honing your skills in testing, debugging, and system design. For managers, it means setting realistic expectations and budgeting for the productionization work that AI can’t do. For learners, it means using AI to accelerate your learning, but not at the expense of understanding the fundamentals.

    AI is a powerful tool, but it doesn’t generate working products—it generates prototypes. The gap between a demo and a shippable product is where the real engineering work happens, and that’s still your job. Embrace AI as an accelerator, but remember that the responsibility for quality, security, and reliability lies with you. The future belongs to those who can use AI effectively while maintaining the judgment and skills that machines can’t replicate.

    Summary

    • AI tools produce prototypes, not finished products; the human must handle integration, testing, and deployment.
    • The “demo effect” makes AI output look competent, but it lacks verification and context.
    • The real value of an engineer is in judgment, not just writing code.
    • Over-reliance on AI can lead to security issues, technical debt, and skill erosion.
    • Use AI as an accelerator, but keep your skills sharp and your expectations realistic.

    FAQ

    Q: Can AI generate a working product if I give it enough detail?
    A: Even with detailed prompts, AI output is still a prototype. It may work in isolation, but it won’t account for all edge cases, integration issues, or production requirements. You’ll need to test, debug, and harden it.

    Q: Is it true that AI will replace programmers?
    A: No. AI can automate parts of coding, but software engineering involves much more—understanding requirements, designing systems, testing, and maintaining. These tasks require human judgment and are unlikely to be fully automated soon.

    Q: How can I avoid “AI slop” in my codebase?
    A: Review AI-generated code carefully, enforce coding standards, and require tests. Treat AI output as a draft that needs the same scrutiny as code written by a junior developer.

    Q: What skills should I focus on if I use AI tools?
    A: Focus on debugging, system design, and critical evaluation. These skills will help you turn prototypes into products and catch issues that AI misses.

    Q: Should I use AI for learning to code?
    A: Yes, but don’t rely on it exclusively. Use AI to get unstuck, but make sure you understand the underlying concepts. Otherwise, you’ll struggle when the AI isn’t there to help.

  • Demystifying DRAM Read Disturbance: RowHammer and RowPress Explained

    Machine Learning-Based Rowhammer Mitigation: A Breakthrough in Hardware Security - Advances in Engineering

    A stylized illustration of a DRAM memory cell array with a highlighted row being accessed, causing adjacent rows to flip bits, with visual effects representing disturbance.

    Imagine you’re reading a book, and every time you turn a page, the ink on the opposite page smudges slightly. Over time, the words become unreadable, even though you never touched that page. This is eerily similar to what happens inside your computer’s memory with a phenomenon called ‘read disturbance.’

    In the world of computing, data integrity is paramount. Yet, a physical quirk in DRAM (Dynamic Random-Access Memory) can silently corrupt data without any warning. This isn’t a theoretical concern—it’s a real vulnerability that has been exploited by hackers and can cause system crashes. Two specific manifestations, RowHammer and RowPress, have captured the attention of researchers and security experts alike.

    In this article, we’ll peel back the layers of these phenomena, understand how they work, and explore why they matter for anyone who uses a computer—from casual users to data center operators.

    The Basics: How DRAM Stores Data

    To understand read disturbance, we need a quick primer on DRAM. Think of DRAM as a massive grid of tiny buckets (capacitors) that hold a charge representing a 1 or a 0. Each bucket is paired with a switch (transistor) that controls access. The grid is organized into rows and columns. When you want to read a specific piece of data, the memory controller ‘activates’ the entire row containing that data. This raises a wire called a wordline, which opens all the switches in that row, allowing the charges to be sensed and read.

    After reading, the row is ‘precharged’—the wordline is lowered, and the row is deactivated. This process happens millions of times per second. But here’s the catch: the capacitors leak charge over time, so DRAM must be refreshed every 64 milliseconds (the refresh interval) to maintain data integrity. This refresh is like a janitor coming by to top up the buckets before they get too empty.

    The Problem: Read Disturbance

    Now, imagine you activate a row repeatedly—say, you’re reading the same data over and over in a tight loop. Each activation causes the wordline to swing in voltage, creating a tiny electromagnetic disturbance. This disturbance can couple into neighboring rows, causing their capacitors to leak a bit more charge than usual. Over time, this leakage can become so significant that a ‘0’ flips to a ‘1’ or vice versa. This is a read disturbance: data corruption in rows you never intended to access.

    This isn’t just a minor annoyance. Because the corruption happens silently, the system doesn’t know the data has changed. This can lead to crashes, incorrect calculations, or, in the worst case, security breaches where an attacker intentionally triggers these flips to gain unauthorized access.

    RowHammer: The Original Discovery

    In 2014, researchers at Carnegie Mellon University published a landmark paper titled “Flipping Bits in Memory Without Accessing Them.” They demonstrated that by rapidly activating the same row hundreds of thousands of times within a refresh window, they could cause bit flips in adjacent rows. This became known as RowHammer.

    The threshold for RowHammer is roughly 100,000 to 200,000 activations per refresh interval (64ms). That might sound like a lot, but modern CPUs can easily achieve that rate. The attack was quickly weaponized: in 2015, a JavaScript-based exploit called Rowhammer.js showed that a malicious website could trigger bit flips remotely, potentially compromising the entire system.

    RowPress: A New Twist

    Fast forward to 2023. Researchers at ETH Zurich discovered a new variant called RowPress. Instead of rapidly activating a row, RowPress involves holding a row open for a long duration. Think of it as pressing and holding a button rather than tapping it repeatedly. This prolonged activation causes charge leakage to neighboring cells, leading to bit flips with just a single activation—if the row is held open for tens of microseconds.

    This is a game-changer because existing defenses against RowHammer, like Target Row Refresh (TRR), which tracks and refreshes rows that are frequently accessed, are ineffective against RowPress. You can’t track a row that’s only activated once, even if it’s held open for a long time.

    Why Should You Care?

    You might think, “I’m not a hacker or a data center operator, so why does this matter?” Here’s the thing: these vulnerabilities are present in virtually all DRAM chips manufactured since the mid-2000s. That means your laptop, smartphone, and even your smart TV are potentially vulnerable. The implications are twofold:

    • Security: Attackers can exploit bit flips to corrupt security-critical data, like page tables that control memory permissions, or cryptographic keys. This can lead to privilege escalation, where a normal user gains admin access.
    • Reliability: In data centers, silent data corruption can cause system crashes or incorrect results in financial or scientific computations. The cost of such failures can be enormous.

    Current Mitigations and Their Limitations

    To combat these issues, several mitigation strategies have been developed:

    • ECC (Error-Correcting Code): This adds extra bits to detect and correct single-bit errors. However, it’s not foolproof—it can’t handle multiple bit flips, and it adds overhead.
    • Increased Refresh Rate: Refreshing more often reduces the time for charge leakage, but it consumes more power and reduces performance.
    • Target Row Refresh (TRR): This hardware feature tracks frequently accessed rows and refreshes their neighbors. But as RowPress shows, it can be bypassed.

    Despite these efforts, no perfect solution exists. Researchers continue to explore new attack variants, like combining RowHammer and RowPress, and developing more robust defenses.

    The Road Ahead

    As DRAM chips become denser and smaller, the problem is likely to worsen. The physical distance between cells shrinks, making them more susceptible to disturbance. This is a pressing concern for the industry, and it’s driving research into new memory technologies, like MRAM or RRAM, which are less prone to such issues.

    In the meantime, understanding these phenomena is crucial for developers, security professionals, and even end-users who want to protect their data. Awareness is the first step toward mitigation.

    RowHammer and RowPress are not just academic curiosities—they are real, exploitable vulnerabilities that affect the memory in virtually every computer. By understanding how they work, we can better appreciate the importance of robust memory design and the ongoing efforts to secure our digital infrastructure. As technology evolves, so too will the threats, but with continued research and vigilance, we can stay one step ahead.

    Summary

    • Read disturbance is a physical phenomenon where accessing one DRAM row causes bit flips in adjacent rows, leading to silent data corruption.
    • RowHammer (discovered 2014) requires rapid repeated activation of a row, while RowPress (discovered 2023) requires a single prolonged activation.
    • Both are hardware-level vulnerabilities present in most DRAM chips since the mid-2000s, with serious security and reliability implications.
    • Existing mitigations like ECC and TRR are imperfect and can be bypassed, especially by RowPress.
    • As DRAM gets denser, the risk increases, making ongoing research into new defenses and memory technologies essential.

    FAQ

    Q: What is DRAM read disturbance?
    A: It’s a physical phenomenon where repeatedly accessing one memory row causes bit flips in adjacent rows, corrupting data without any explicit write operation.

    Q: How does RowHammer differ from RowPress?
    A: RowHammer requires rapidly activating the same row many times (hundreds of thousands) within a refresh window, while RowPress requires holding a row open for a long duration, even with a single activation.

    Q: Can my computer be affected by these vulnerabilities?
    A: Yes, virtually all DRAM chips made since the mid-2000s are potentially vulnerable. However, exploiting them requires specific conditions and often local access, though remote attacks have been demonstrated.

    Q: What are the main defenses against RowHammer and RowPress?
    A: Common defenses include Error-Correcting Code (ECC), increasing refresh rates, and Target Row Refresh (TRR). However, these are not foolproof and can be bypassed by advanced attacks.

    Q: Why are these vulnerabilities important for security?
    A: Bit flips can be used to corrupt security-critical data like page tables or cryptographic keys, potentially allowing attackers to gain unauthorized access or cause system crashes.

  • How to Do Great Work: A Practical Guide from Paul Graham

    A person standing at the edge of a vast, unexplored landscape with a winding path leading to a distant horizon, symbolizing the journey to the frontier of knowledge.

    In July 2023, Paul Graham—co-founder of Y Combinator and one of the most influential voices in tech—published an essay simply titled ‘How to Do Great Work.’ It quickly became a touchstone for anyone striving to create something meaningful, whether in science, art, or entrepreneurship. But what does ‘great work’ actually mean, and how can you achieve it? Graham’s answer is not a secret formula or a stroke of genius; it’s a deliberate, iterative process that anyone can adopt.

    This article unpacks Graham’s core ideas, translating them into actionable steps. You’ll learn how to choose the right problems, develop a ‘taste’ for quality, and cultivate the resilience needed to see long projects through. Whether you’re a student, a founder, or a creative, these principles offer a roadmap for turning curiosity into impact.

    The Core Idea: Great Work Is a Process, Not a Gift

    Paul Graham’s central thesis is that great work—work that is novel, useful, and impactful—is not the result of raw talent alone. Instead, it emerges from a deliberate, iterative process. He emphasizes that you can’t just wait for inspiration; you have to actively seek out problems, learn the landscape, and then chip away at them with relentless effort.

    Think of it like a sculptor. A block of marble doesn’t become a statue by itself; the sculptor must study the stone, find the shape within, and then work with tools to reveal it. Similarly, great work is revealed through a process of exploration and refinement. Graham’s essay is essentially a guide to becoming that sculptor of your own ideas.

    Step 1: Decide What to Work On

    The first and most crucial step is choosing your problem. Graham advises that you should work on something that interests you deeply and seems important. He calls this ‘ambitious curiosity’—a combination of genuine fascination and a desire to make a difference.

    But how do you find such a problem? Graham suggests you start by asking yourself what you’re curious about, what puzzles you, and what you find yourself thinking about when you’re not forced to. It’s not about picking the ‘hot’ topic or the one that will make you famous; it’s about following your own intellectual passions.

    For example, if you’re a computer science student, don’t just pick a thesis topic because it’s trendy. Instead, think about the problems that make you lose track of time. Maybe it’s how to make algorithms more efficient, or how to build software that’s easier for people to use. That genuine interest will sustain you through the hard work ahead.

    Step 2: Learn Enough to Get to the Frontier

    Once you’ve chosen your area, you need to master what’s already known. Graham calls this ‘getting to the frontier’—the edge of human knowledge in your field. This means reading papers, studying the work of others, and understanding the current state of the art.

    But don’t stop there. The frontier is where you’ll find the ‘interesting anomalies’—the things that don’t quite fit, the contradictions, the unexplored niches. These are the cracks where great work often emerges.

    Imagine you’re a historian studying a well-documented period. You read all the major texts, but you notice a minor event that doesn’t align with the standard narrative. That anomaly is your crack. It’s a chance to ask a new question and potentially uncover something significant.

    Step 3: Look for ‘Cracks’ or ‘Seams’

    Graham uses the metaphor of ‘cracks’ or ‘seams’ to describe the opportunities for great work. These are the overlooked details, the contradictions, and the unexplored areas that others have missed. Finding them requires a combination of deep knowledge and a fresh perspective.

    One way to spot cracks is to question assumptions. Ask yourself: Why is this the accepted view? What if it’s wrong? What if we looked at it from a different angle? This kind of thinking can lead to breakthroughs.

    For instance, in the early days of the internet, many people assumed that online commerce would never work because people wouldn’t trust it. But a few entrepreneurs saw a crack: the possibility of building trust through reviews and secure payments. That insight led to companies like Amazon and eBay.

    Step 4: Work on Projects, Not Just Skills

    It’s tempting to focus on building skills—learning a programming language, mastering a technique, or reading more books. But Graham argues that skills are only useful when applied to projects. Projects force you to produce tangible output, which is where the real learning and iteration happen.

    A project is a specific, finite goal: write a paper, build a prototype, compose a piece of music. It has a beginning and an end, and it produces something you can show to others. This is different from just ‘practicing’ or ‘studying’.

    For example, if you want to become a better writer, you could read books on writing (skill-building), but you’ll improve much faster by writing a blog post or a short story (project). The project gives you a concrete outcome to evaluate and refine.

    Step 5: Iterate and Refine

    Graham is clear that your first attempts will likely be flawed. That’s okay. The key is to keep improving, to work on things that ‘seem like they might be good’ and then refine them based on feedback and your own judgment.

    This is the iterative process: create a rough draft, get feedback, revise, repeat. It’s how great art, great science, and great products are made. The first version of Google was not the polished search engine we know today; it was a rough prototype that was constantly refined.

    Don’t be afraid of imperfection. Instead, embrace it as a starting point. The goal is not to get it right the first time, but to get it better each time.

    Step 6: Cultivate a ‘Taste’ for Quality

    To know what ‘good’ looks like, you need to develop a sense of taste. This is an aesthetic judgment that guides your choices. In science, it’s knowing what a beautiful theory looks like; in art, it’s knowing what a powerful composition feels like; in business, it’s knowing what a compelling product is.

    How do you develop taste? By exposing yourself to the best work in your field and by constantly asking yourself what makes it good. Over time, you’ll internalize these standards and use them to evaluate your own work.

    For example, a young filmmaker might watch dozens of classic movies, analyzing why they work. That analysis builds a sense of taste that will inform their own directing choices.

    Step 7: Embrace ‘Slow Hunches’

    Great ideas often don’t come in a flash of insight. Instead, they develop over years as ‘slow hunches’—half-formed thoughts that you keep in the back of your mind and revisit over time. Graham advises keeping a notebook of these ideas, so you don’t lose them.

    This is why many great thinkers are known for carrying notebooks. They jot down observations, questions, and snippets of ideas, and then later, they connect the dots. The hunch that seemed trivial at first might become the seed of a major project.

    For instance, Charles Darwin’s theory of natural selection didn’t emerge fully formed. It was the result of years of collecting observations and slowly developing the idea, which he finally articulated in ‘On the Origin of Species.’

    Step 8: Collaborate and Seek Feedback

    Great work is rarely done in isolation. Graham emphasizes the importance of sharing your work early and often, and of surrounding yourself with the right peers. Good collaborators are a ‘force multiplier’—they can see things you miss, offer new perspectives, and push you to be better.

    Seek feedback not just from mentors, but also from peers who are working on similar problems. They understand the context and can give you practical advice. And don’t be defensive; use feedback to improve.

    In the startup world, this is why accelerators like Y Combinator are so effective. They provide a community of founders who share their progress and get feedback from each other, which accelerates their learning.

    Step 9: Maintain Resilience

    This is perhaps the most important quality: the willingness to work on hard problems for a long time. Expect setbacks, boredom, and self-doubt. They are part of the process.

    Graham warns that you will often feel like you’re not making progress, or that your work isn’t good enough. This is normal. The key is to keep going, to maintain the discipline to show up and work even when you don’t feel like it.

    Think of it like training for a marathon. You don’t run 26.2 miles on your first day. You build up gradually, and you have days when you want to quit. But if you stick with it, you eventually cross the finish line.

    Step 10: Prioritize Work-Life Balance as a Means, Not an End

    Finally, Graham argues that rest and play are not the opposite of work; they are necessary for sustained creativity. Your brain needs time to relax and wander, which is often when the best ideas surface.

    But he’s careful to say that balance is a means to an end, not the end itself. The goal is to enable more and better work, not to avoid work. So, take breaks, pursue hobbies, and spend time with loved ones—but do it so you can return to your work with fresh energy and insight.

    For example, many writers find that taking a walk or doing a mundane task helps them solve a plot problem. The break allows their subconscious to work on the issue.

    Putting It All Together

    Graham’s advice is not a checklist but a mindset. It’s about approaching your work with curiosity, discipline, and a long-term perspective. You won’t do great work overnight, but by following these principles, you can create the conditions for it to emerge.

    The most important takeaway is to start. Choose a problem that excites you, learn everything you can, and begin working on a project. Iterate, seek feedback, and stay resilient. Over time, you’ll produce work that is uniquely yours and that contributes something new to the world.

    Paul Graham’s ‘How to Do Great Work’ is not a magic formula but a practical philosophy. It reminds us that greatness is not reserved for a lucky few; it’s available to anyone willing to work deliberately and persistently. By choosing meaningful problems, developing your taste, and embracing the iterative process, you can produce work that matters. The journey may be long, but the rewards—both personal and professional—are worth it.

    Summary

    • Great work is a process, not a gift: it requires deliberate effort and iteration.
    • Choose problems that deeply interest you and seem important; follow your ‘ambitious curiosity.’
    • Master the existing knowledge in your field to reach the ‘frontier,’ where you can spot ‘cracks’ and ‘seams.’
    • Work on projects, not just skills, and iterate based on feedback and your own taste.
    • Cultivate resilience and embrace ‘slow hunches’—great ideas often develop over years.

    FAQ

    Q: What is the main idea of Paul Graham’s essay ‘How to Do Great Work’?
    A: The main idea is that great work is not the result of raw talent alone but of a deliberate, iterative process. Graham outlines steps like choosing interesting problems, learning to the frontier, finding ‘cracks,’ working on projects, and maintaining resilience.

    Q: How do I choose what to work on?
    A: Choose something that interests you deeply and seems important. Follow your ‘ambitious curiosity’—the combination of genuine fascination and a desire to make a difference. Don’t pick a topic just because it’s trendy; pick one that you find yourself thinking about even when you don’t have to.

    Q: What does ‘getting to the frontier’ mean?
    A: It means mastering the existing knowledge in your field so you understand the current state of the art. Once you’re at the frontier, you can identify the ‘interesting anomalies’—the gaps, contradictions, or unexplored areas where great work often emerges.

    Q: Why is working on projects important?
    A: Projects force you to produce tangible output, which is where real learning and iteration happen. Skills are necessary, but they’re only useful when applied to a specific goal. Projects give you something to show, evaluate, and refine.

    Q: How can I stay motivated when working on a long-term project?
    A: Graham emphasizes resilience and the willingness to work on hard problems for a long time. Expect setbacks and self-doubt, but keep going. Also, take breaks and maintain a work-life balance as a means to sustain your creativity, not as an end in itself.

  • BMW’s New Dashboard Ads: A ‘Treat’ or a Slippery Slope?

    A close-up of a modern car dashboard screen displaying a digital advertisement, with the BMW logo subtly visible on the steering wheel.

    Imagine sitting in your parked BMW, ready to head out, when a notification pops up on the dashboard screen: ‘Enjoy your drive? Consider upgrading to our premium navigation package!’ It’s not a glitch—it’s an advertisement, delivered directly to your car’s infotainment system. BMW is now showing ads on the dashboard screens of its vehicles, and the company is framing this as a ‘treat’ for drivers, a value-add to the connected car experience. But is this a harmless perk or a troubling shift in the relationship between automakers and car owners?

    The New Reality: Ads in Your Car

    BMW has begun pushing advertisements to the central infotainment displays of its vehicles. These ads appear via over-the-air (OTA) updates, which means they can be delivered remotely without a trip to the dealership. The content so far has been limited to BMW’s own services—like promoting digital features, maintenance packages, or accessories. For example, you might see a prompt to book a service appointment or to subscribe to a premium feature. The ads are designed to appear during idle moments, such as when the car is parked or at startup, minimizing distraction while driving.

    Why BMW Is Doing This

    Automakers are facing shrinking profit margins on vehicle sales. The industry is pivoting to recurring revenue streams: subscriptions, connected services, and data monetization. BMW has been a pioneer in this space, with controversial moves like charging a subscription for Apple CarPlay (later reversed) and offering heated seats as a paid feature. In-car advertising is a natural extension of this strategy. By showing ads, BMW can generate additional income, potentially subsidizing the cost of connected services or boosting their bottom line.

    The Industry Trend: You’re Not Alone

    BMW is not the first to explore this territory. Ford patented technology for displaying ads on in-car screens in 2023. General Motors announced plans for in-car advertising in 2022, but walked back after public backlash. Tesla has experimented with ads and premium connectivity tiers, though it has avoided intrusive ads on the main display. Stellantis (the parent of Jeep and Ram) has discussed ‘data monetization’ and in-car ad opportunities. The infrastructure is already in place: modern cars run on sophisticated infotainment systems with internet connectivity, making it technically trivial to push ads.

    The Consumer Backlash: Why People Are Upset

    For many drivers, the idea of ads in a car they paid $50,000 or more for feels like a violation. It’s a fundamental shift in the ownership experience. When you buy a car, you expect to own the hardware and control what appears on its screens. Ads challenge that expectation. There are also safety concerns. Even if ads appear only when parked, the potential for distraction during navigation or at stoplights is worrying. Critics also see this as a slippery slope: if BMW starts with its own services, third-party ads for coffee shops or gas stations are likely next. And ads require data—about your location, driving habits, and preferences—raising privacy concerns.

    The Pro-BMW Argument: A Value Exchange

    BMW might argue that ads are a fair trade: they subsidize the cost of connected services, allowing features like real-time traffic or remote services to be offered at lower prices. If the ads are relevant and helpful—like reminding you to book a service appointment—they could be seen as useful notifications rather than intrusive marketing. Drivers can dismiss them, and they appear only in non-driving moments. In a world where streaming services and mobile apps have normalized ad-supported tiers, cars may be the next frontier.

    The Regulatory Gray Area

    Currently, there is little to no regulation specifically governing in-car advertising. Privacy laws like GDPR in Europe and CCPA in California may apply to data collection used for ad targeting, but the display of ads itself is largely unregulated. Consumer protection laws could be invoked if ads are misleading or if BMW frames them as ‘features’ without clear disclosure. However, the legal landscape is still catching up to this new reality.

    What This Means for You

    If you own a BMW, you might start seeing these ads soon. You can ignore them, but they’re likely here to stay. The bigger question is whether this trend will spread to other automakers. Given the industry’s financial pressures, it’s plausible that in-car ads become as common as ads on streaming platforms. As a consumer, it’s important to be aware of this shift and to voice your opinions—automakers have reversed course before in the face of public backlash, as GM did in 2022.

    BMW’s dashboard ads are a test balloon for the automotive industry. While the company frames them as a ‘treat,’ many drivers see them as an intrusion. The outcome will depend on consumer reaction and regulatory response. For now, the next time you see an ad on your car’s screen, remember: it’s not just a notification—it’s a sign of where the industry is heading.

    Summary

    • BMW is showing ads on dashboard screens, delivered via over-the-air updates, initially for its own services.
    • Automakers are exploring in-car ads as a new revenue stream due to shrinking margins on vehicle sales.
    • Consumers are concerned about ownership rights, safety, privacy, and the potential for third-party ads.
    • Regulations are currently sparse, but privacy laws may apply to data collection for ad targeting.
    • The trend is industry-wide, with Ford, GM, Tesla, and Stellantis all exploring similar ideas.

    FAQ

    Q: Will I see ads while driving?
    A: BMW says ads appear only during idle moments, like when parked or at startup, to minimize distraction. However, the potential for ads during navigation or at stoplights remains a concern.

    Q: Can I opt out of seeing these ads?
    A: Currently, there’s no clear opt-out mechanism. BMW frames the ads as a value-add, so they may be part of the connected services experience. You can dismiss them, but they may reappear.

    Q: Are these ads for third-party products?
    A: So far, ads are for BMW’s own services and products. But the infrastructure could support third-party ads in the future, which is a major concern for critics.

    Q: How does BMW deliver these ads?
    A: Through over-the-air (OTA) updates, which allow BMW to push content to the car’s infotainment system remotely, without a dealer visit.

    Q: Is this legal?
    A: There’s little regulation specifically on in-car ads. Privacy laws like GDPR and CCPA may apply to data collection, but the display of ads is largely unregulated for now.

  • Why Oil Giants Are Posting Record Profits During Wartime Crude Prices

    An illustration showing a large oil pump jack silhouette against a dramatic sunset with a rising stock market graph overlay, symbolizing record profits during wartime crude prices.

    In the second quarter of 2026, the world’s largest oil companies reported staggering profits, collectively tens of billions of dollars. ExxonMobil, Chevron, Shell, BP, and TotalEnergies all posted earnings that dwarfed their historical averages. The cause? Crude oil prices spiked to levels not seen in years, driven by ongoing geopolitical conflicts that have disrupted supply chains and added a ‘war premium’ to every barrel.

    For everyday consumers, these headlines can be baffling and frustrating. How can companies rake in record profits while families struggle with high energy bills? Is this price gouging, or is something more complex at play? Understanding the mechanics behind these windfalls requires a closer look at how global oil markets work, the role of conflict in shaping prices, and the various ways these companies actually make money.

    This article breaks down the key factors behind the profit surge, separates myth from reality, and explores the broader implications for the economy, consumers, and the energy transition. By the end, you’ll have a clearer picture of why oil company profits soar during wartime and what it means for the rest of us.

    The Perfect Storm: How Conflict Drives Oil Prices Up

    Oil prices are not set by any single company or government; they emerge from a global market where supply and demand meet. When geopolitical tensions flare, the market reacts to the fear of supply disruptions, even before any actual barrels are lost. This is known as the ‘war premium.’

    In Q2 2026, several factors converged to push crude prices well above $100 per barrel, with Brent averaging around $110–120. Key shipping lanes like the Strait of Hormuz and the Red Sea faced threats, raising insurance costs and rerouting tankers. Sanctions on major producers removed millions of barrels from the market, and OPEC+ spare capacity was limited, meaning there was little buffer to absorb any shortfall.

    At the same time, global demand remained surprisingly robust, especially from Asia. When supply tightens and demand holds steady, prices rise—sometimes sharply. This is basic economics, but it’s the foundation for the profit surge.

    Inside the Numbers: Where Do the Profits Come From?

    Oil companies are not monolithic; they operate across different segments, each with its own profit dynamics. The three main areas are:

    • Upstream (Exploration & Production): This is where crude is extracted. When oil prices jump, the revenue from selling each barrel increases directly. For a company like ExxonMobil, which produces millions of barrels per day, even a $10 increase in price can add billions to quarterly revenue.

    • Downstream (Refining & Marketing): Refineries turn crude into gasoline, diesel, and other products. When crude prices spike, product prices often lag behind, temporarily widening refining margins. This means refineries can buy crude at a lower cost relative to what they sell their products for, boosting profits.

    • Trading: Many oil giants have proprietary trading desks that profit from volatility. When prices swing wildly, these desks can make speculative bets that pay off handsomely.

    In Q2 2026, all three segments contributed to the windfall. Upstream profits soared due to high crude prices, downstream margins expanded, and trading desks capitalized on the chaos.

    The Recurring Cycle: A Historical Pattern

    This isn’t the first time oil companies have posted massive profits during conflict. In 2008, amid the Iraq War and other tensions, prices hit record highs. In 2022, after Russia’s invasion of Ukraine, ExxonMobil reported its highest annual profit ever. The pattern is consistent: conflict → supply fear → price spike → record earnings.

    Why does this keep happening? Because oil is a globally traded commodity, and geopolitical instability directly threatens its supply chain. When that threat is perceived, prices rise, and companies that hold oil inventories or have production capacity benefit enormously.

    The Critics’ View: Profiteering or Market Reality?

    Consumer advocacy groups and politicians often cry foul when oil companies report huge profits while households struggle with energy bills. They argue that these windfalls are a form of profiteering, especially when companies use the money for stock buybacks and dividends rather than increasing supply or investing in renewable energy.

    However, industry executives counter that they are price-takers, not price-setters. They argue that profits are a function of global markets, not unilateral price hikes. They also point out that reinvestment in supply is needed to stabilize prices, and that windfalls fund research into cleaner technologies.

    Both perspectives have merit. While companies don’t control oil prices, they do make strategic decisions about how to allocate profits. Critics argue that during a cost-of-living crisis, prioritizing shareholder returns over consumer relief is morally questionable.

    The Investor’s Perspective: Cheers and Concerns

    For shareholders, record profits are a welcome boon. Buybacks and dividends increase, rewarding investors who have weathered years of volatility. However, some ESG-focused investors are questioning whether these windfalls are being used to accelerate the energy transition. If fossil fuels remain this lucrative, the incentive to pivot to renewables diminishes.

    This tension is at the heart of the climate debate. High oil prices make clean energy more competitive in relative terms, but they also make oil companies richer and more powerful, potentially slowing their transition efforts.

    The Macroeconomic Impact: A Tax on the Global Economy

    High oil prices are inflationary. They raise the cost of transportation, heating, and manufacturing, which feeds into consumer prices. Central banks, already battling inflation, may be forced to keep interest rates higher for longer, slowing economic growth. In this sense, wartime oil prices act like a tax on the global economy, transferring wealth from consumers to oil-producing companies and nations.

    This dynamic creates a political dilemma. Governments face pressure to act—either by imposing windfall profit taxes, releasing strategic reserves, or pressuring OPEC+ to increase production. But such measures are often slow and politically fraught.

    Separating Myth from Reality

    Several misconceptions cloud the public debate:

    • Myth: Profits equal price gouging. In reality, profits are largely driven by global commodity prices, not unilateral price hikes. However, vertical integration can amplify margins beyond simple price pass-through.
    • Myth: Companies control oil prices. Oil prices are set by global supply and demand, OPEC+ decisions, and geopolitical events. Even the largest companies are price-takers in the spot market.
    • Myth: Windfall equals cash on hand. Accounting profits include non-cash items like inventory gains and impairment reversals. Cash flow is a better measure of actual liquidity.
    • Myth: All oil companies benefit equally. Profits vary widely by geography, asset mix, hedging positions, and exposure to specific conflicts. Some may even lose money on refining if price spikes outpace product prices.
    • Myth: Profits are immediately available for spending. Much is earmarked for debt repayment, dividends, buybacks, and capital projects; only a fraction is discretionary.
    • Myth: Wartime prices are purely speculative. While speculation amplifies moves, physical supply disruptions are real—lost barrels from sanctioned nations, damaged infrastructure, and rerouted tankers.

    The Road Ahead: What Happens Next?

    The duration of high oil prices depends on the conflict’s trajectory. If de-escalation occurs, prices may fall, and profits will normalize. But if tensions persist, we could see continued windfalls, along with ongoing inflationary pressure and political backlash.

    For consumers, the key takeaway is that oil company profits are a symptom, not the cause, of high energy prices. The root cause is geopolitical instability and the resulting supply constraints. Until the world diversifies its energy sources and reduces dependence on volatile regions, this cycle is likely to repeat.

    Oil companies’ record profits during wartime are a direct consequence of global supply disruptions and price spikes. While these windfalls are legitimate market outcomes, they raise important questions about fairness, corporate responsibility, and the pace of the energy transition. As consumers, understanding the mechanics behind these profits can help us engage in more informed debates about energy policy and the future of our economy.

    Summary

    • Oil company profits in Q2 2026 soared due to wartime crude prices, driven by supply disruptions and geopolitical tensions.
    • Profits come from upstream production, refining margins, and trading desks, all of which benefit from price volatility.
    • This is a recurring pattern seen in 2008 and 2022, where conflict leads to price spikes and record earnings.
    • Critics call for windfall taxes, while industry argues profits are market-driven and fund reinvestment.
    • High oil prices have inflationary effects, acting as a tax on the global economy and complicating central bank policy.

    FAQ

    Q: Are oil companies price gouging when they report record profits?
    A: Not necessarily. Oil prices are set by global supply and demand, not by individual companies. Profits rise because the market price of crude increases, which is a direct result of supply disruptions and geopolitical tensions. However, some companies may benefit from refining margins or trading activities that amplify profits beyond simple price pass-through.

    Q: Why don’t oil companies use their windfall profits to lower prices at the pump?
    A: Oil companies are price-takers in the global market; they cannot unilaterally lower prices without selling below market rates, which would be unsustainable. Additionally, much of the profit is allocated to dividends, buybacks, debt repayment, and capital projects. Lowering prices would require subsidizing consumers, which is not typical corporate behavior.

    Q: Do all oil companies benefit equally from high crude prices?
    A: No. Profits vary based on a company’s asset mix (upstream vs. downstream), geographic exposure, hedging strategies, and how much they rely on trading. Some may even lose money if refining margins compress or if they are heavily hedged against price increases.

    Q: What is a windfall profit tax, and could it be applied?
    A: A windfall profit tax is a levy on profits deemed excessive or unexpected. Some governments have proposed or enacted such taxes during periods of high oil prices. However, they are controversial because they may discourage investment in supply and are difficult to design without unintended consequences.

    Q: How do high oil prices affect the average consumer?
    A: High oil prices increase the cost of gasoline, heating, and goods that rely on transportation. This contributes to inflation, which can erode purchasing power and force central banks to raise interest rates, potentially slowing economic growth.

  • Cursor Removes Cost Data from Usage Dashboard: What Users Need to Know

    A screenshot or illustration of a usage dashboard with cost data removed, showing a blank or missing section where costs used to be.

    If you’re a Cursor user, you may have noticed something missing from your usage page recently: the detailed token counts and dollar costs that once helped you track your spending. This change, which also affects CSV exports, has sparked frustration and confusion across the community. In this article, we’ll break down exactly what happened, why it matters, and what you can do to stay informed about your usage.

    What Changed and What Didn’t

    Cursor, the AI-powered code editor, has removed cost-related information from its usage tracking page and CSV export. Previously, users could see detailed token counts (input and output tokens) and the corresponding dollar cost for their subscription plan. Now, that granular data is gone, replaced by more generic metrics like the number of requests or a usage percentage.

    It’s important to clarify what hasn’t changed: the usage page still exists, and you can still see some metrics. The CSV export still works, but it no longer includes the token and cost columns. So, this isn’t a removal of usage tracking altogether—it’s a removal of the cost transparency.

    Why Users Are Frustrated

    The reaction has been largely negative, especially among users on usage-based plans or those who rely on Cursor for professional work. Freelancers and small teams often use the cost data to budget and avoid surprise overage charges. Without it, they feel like they’re flying blind.

    One user on the Cursor forum put it bluntly: “I need to know how much I’m spending to decide if I should upgrade or cut back. Now I have no idea until the bill arrives.” This sentiment echoes across Hacker News, where the topic gained significant traction.

    Why Did Cursor Do This?

    Cursor hasn’t issued an official statement, so we can only speculate. Here are a few plausible reasons:

    • Simplification: Token math can be confusing for non-technical users. A percentage bar is easier to understand at a glance.
    • Anti-gaming: Some users might optimize prompts to minimize token usage, which could degrade code quality. Removing cost data discourages this behavior.
    • Pricing Overhaul: Cursor might be preparing to shift to a value-based pricing model where raw token counts matter less. Removing the data now could be a precursor to that change.

    It’s also possible that Cursor wants to steer the conversation away from cost and toward the value the tool provides. But without transparency, that’s a hard sell for budget-conscious users.

    What Can You Do?

    While you can’t force Cursor to bring back the data, you can take steps to manage your usage:

    • Monitor your usage manually: Keep a log of your sessions and estimate token usage based on your prompts. It’s not perfect, but it gives you a rough idea.
    • Use third-party tools: Some users have built browser extensions or scripts to track usage. These are unofficial, so use them at your own risk.
    • Contact support: Let Cursor know that cost transparency matters to you. If enough users complain, they might reconsider.
    • Check your plan limits: Review your subscription terms to understand what’s included and what happens if you exceed it.

    The Bigger Picture

    This change highlights a growing tension in the AI tools space: how much transparency should companies provide about the underlying costs of their services? Competitors like GitHub Copilot and OpenAI’s API dashboards still show detailed token and cost breakdowns, which makes Cursor’s move stand out.

    Whether this is a step toward a better pricing model or a misstep in user trust, only time will tell. For now, the best you can do is stay informed and vocal about your needs.

    Cursor’s removal of cost data from its usage dashboard is a significant shift that has left many users feeling in the dark. While the company hasn’t explained its reasoning, the change is deliberate and affects both the UI and CSV exports. If you rely on Cursor for work, it’s worth taking proactive steps to monitor your usage and voice your concerns. Transparency matters, and users have the power to influence product decisions.

    Summary

    • Cursor removed token counts and dollar costs from its usage page and CSV export.
    • The usage page still shows some metrics, but not the granular cost breakdown.
    • Users are frustrated, especially those on usage-based plans who need cost data for budgeting.
    • Cursor hasn’t explained the change, but speculation includes simplification, anti-gaming, or a pricing overhaul.
    • You can manually track usage, use third-party tools, or contact support to voice your concerns.

    FAQ

    Q: Did Cursor remove all usage tracking?
    A: No. The usage page still exists and shows some metrics like request counts or a usage percentage. Only the detailed token and cost information was removed.

    Q: Is this a bug?
    A: Unlikely. The change is consistent across the UI and CSV export, suggesting it was a deliberate decision.

    Q: Does this affect all subscription tiers?
    A: Reports suggest it affects multiple tiers, but the exact scope isn’t confirmed. Check your own usage page to see what’s visible.

    Q: Can I still export my usage data?
    A: Yes, the CSV export still works, but it no longer includes token and cost columns.

    Q: Is Cursor trying to hide costs to overcharge me?
    A: There’s no evidence of overcharging. The issue is about visibility, not billing accuracy. Your bill should still reflect your actual usage.

  • Canada Quietly Signs UN Cybercrime Treaty: A Surveillance Pact in Disguise?

    A stylized Canadian flag with a digital surveillance eye overlay, symbolizing the tension between national identity and digital privacy.

    In late 2025, Canada quietly signed the United Nations Convention on Cybercrime, a treaty that aims to harmonize cybercrime laws globally. But critics warn that beneath its crime-fighting surface, the treaty contains provisions that could enable mass surveillance and undermine civil liberties. The signing, which occurred with little public debate or parliamentary scrutiny, has raised alarms among privacy advocates who see it as a backdoor to expanded state powers.

    This article unpacks what the treaty actually says, why Canada signed it, and what it could mean for your digital rights. We’ll explore the fine print on data collection, the vague ‘prevention’ clause, and the geopolitical chess game that led to this moment. By the end, you’ll understand why this seemingly technical treaty is anything but mundane.

    What Is the UN Cybercrime Convention?

    Formally known as the ‘United Nations Convention on Countering the Use of Information and Communications Technologies for Criminal Purposes,’ this treaty was adopted by the UN General Assembly in December 2024. It’s a broad agreement that requires signatories to criminalize a range of cyber offenses—from illegal access to data interference, fraud, and child sexual abuse material. It also sets up frameworks for international cooperation, including mutual legal assistance and extradition.

    But the treaty goes beyond simple crime-fighting. It includes provisions for real-time collection of traffic data and preservation of electronic evidence. These are tools that law enforcement agencies love, but they come with significant privacy implications. The treaty also has a controversial ‘prevention’ clause that critics argue could be used to justify broad surveillance or content moderation mandates.

    The Quiet Signing: Why No One Noticed

    Canada signed this treaty in 2025–2026 with almost no public fanfare. There were no major press conferences, no parliamentary debates, and no consultations with civil society. This is a stark contrast to how Canada typically handles major international agreements. The government’s silence has led to accusations that it’s trying to sneak a surveillance-friendly treaty past the public.

    Why the secrecy? One possibility is that the government knows the treaty is controversial. Another is that it’s part of a broader strategy to engage with the UN process to counter Russian and Chinese influence. But whatever the reason, the lack of transparency is troubling for a treaty that could affect the digital rights of every Canadian.

    The Surveillance Provisions: What’s in the Fine Print?

    Let’s break down the most concerning parts of the treaty. First, there’s the real-time collection of traffic data. This means that internet service providers (ISPs) could be required to hand over information about who you’re communicating with, when, and from where—in real time. This is different from wiretapping, which captures the content of communications. Traffic data is metadata, and it can reveal a lot about your life, even if the content of your messages remains private.

    Second, the treaty requires signatories to preserve electronic evidence. This sounds benign, but it can mean that companies must store data for long periods, even if there’s no ongoing investigation. This could lead to data retention mandates that force companies to keep logs of your online activities for months or years.

    Third, the ‘prevention’ clause is vague. It says that countries should take measures to prevent cybercrime, but it doesn’t define what those measures are. This could be interpreted to require ISPs and platforms to monitor content for illegal activity, which would be a form of mass surveillance. It could also be used to pressure companies to weaken encryption, which would make everyone less secure.

    The Budapest Convention: A Better Alternative?

    Canada is already a party to the Budapest Convention on Cybercrime, which has been the gold standard for international cybercrime cooperation since 2001. The Budapest Convention has strong human rights protections and requires that any data collection be subject to due process. The UN treaty, in contrast, has weaker safeguards, which is why many experts see it as a step backward.

    Why would Canada sign a weaker treaty when it already has a better one? The answer may lie in geopolitics. The UN treaty was a Russian-led initiative, and by signing it, Canada can have a seat at the table when the rules are being written. But critics argue that this legitimizes a treaty that could be used by authoritarian states to justify surveillance of dissidents and journalists.

    What Does This Mean for Canadians?

    If Canada ratifies the treaty, it will need to update its laws to comply. This could mean changes to the Criminal Code and the Privacy Act. The government might argue that existing laws already meet the treaty’s requirements, but the treaty’s vague language could be used to push for more expansive surveillance powers.

    For ordinary Canadians, the most immediate impact could be on your online privacy. If ISPs are required to collect and store traffic data, that information could be accessed by law enforcement without a warrant in some cases. The treaty also creates a framework for sharing evidence across borders, which could make it easier for foreign governments to request data about Canadians.

    The Geopolitical Angle: Why Canada Signed

    Canada’s decision to sign is not just about cybercrime; it’s about international relations. The UN treaty was adopted with support from many Global South countries, who see it as a way to get technical assistance and capacity building. By signing, Canada can help shape how the treaty is implemented, potentially pushing for stronger human rights protections.

    But there’s a risk: by signing, Canada lends legitimacy to a treaty that could be used to justify authoritarian surveillance. Some argue that boycotting the treaty would be worse, as it would leave the field open to Russia and China to define the norms. It’s a delicate balance, and the Canadian government seems to be betting that it can influence the treaty from within.

    The Path to Ratification: Still a Chance for Debate

    Signing is just the first step. The treaty will only enter into force after 40 countries ratify it, and as of early 2026, fewer than 20 have done so. In Canada, ratification requires parliamentary approval, which means there’s still time for public debate. Civil society groups are already calling for hearings and consultations, and it’s possible that the government will face pressure to add reservations or interpretative declarations to protect Canadians’ rights.

    If you’re concerned about this treaty, now is the time to speak up. Contact your MP, join privacy advocacy groups, and demand that the government be transparent about its intentions. The treaty may have been signed quietly, but its impact could be loud and lasting.

    Canada’s quiet signing of the UN Cybercrime Convention is a wake-up call for anyone who cares about digital rights. The treaty’s surveillance-friendly provisions, combined with the lack of public debate, make it a dangerous precedent. While signing doesn’t mean immediate ratification, it sets the stage for a potential erosion of privacy protections. Canadians must demand transparency and accountability before this treaty moves any further.

    Summary

    • Canada signed the UN Cybercrime Convention in 2025–2026 with little public or parliamentary scrutiny.
    • The treaty includes provisions for real-time traffic data collection and electronic evidence preservation, which could enable mass surveillance.
    • The vague ‘prevention’ clause could be used to justify content monitoring or weakened encryption.
    • Canada is already a party to the stronger Budapest Convention, raising questions about why it signed a weaker treaty.
    • The treaty is not yet ratified; there is still time for public debate and parliamentary oversight.

    FAQ

    Q: What is the UN Cybercrime Convention?
    A: It’s a UN treaty adopted in December 2024 that requires countries to criminalize cybercrimes and cooperate internationally. It includes provisions for data collection and evidence sharing that worry privacy advocates.

    Q: Why is Canada’s signing controversial?
    A: Because it happened quietly, without public debate, and the treaty’s provisions could be used to justify surveillance and data retention that infringe on privacy rights.

    Q: How does this treaty differ from the Budapest Convention?
    A: The Budapest Convention has stronger human rights protections and due process requirements. The UN treaty is seen as weaker, with vaguer language that could be exploited by authoritarian governments.

    Q: What can I do to stop it?
    A: Contact your Member of Parliament, support privacy advocacy groups, and demand that the government hold public consultations before ratification.

    Q: Will this affect my online privacy?
    A: If ratified, it could lead to laws requiring ISPs to collect and store traffic data, which law enforcement could access. This could make it easier for authorities to track your online activities.