AI can feel nearly free to the person using it while its wider costs fall on workers, communities, and infrastructure. In her Cybernews editorial, Chief Editor Jurgita Lapienytė argues that those costs deserve scrutiny—but that scrutiny should not turn hard-to-test predictions of catastrophe into established fact. Her title, “Let’s make AI way harder than it needs to be,” is a challenge to how we discuss AI, not a technical guide or a claim that every worst-case forecast is true.
What Lapienytė means by making AI “harder”
The editorial begins with an ironic line: “I love the thrill of thinking the world is about to end.” Lapienytė is not endorsing panic. She is examining the pull of dramatic AI predictions alongside less dramatic costs that may already be accumulating.
Her central contrast is between the low immediate price a user sees and the broader consequences AI may impose. A person can make an image or use a chatbot without seeing the full electricity demand, local infrastructure pressure, labor disruption, environmental burden, or security exposure associated with AI systems. The editorial raises these as concerns; it is commentary, not an original study that measures each one.
Why a tiny user cost does not settle the question
As a personal example, Lapienytė writes that an “80s-style picture” of herself cost her four cents in tokens. That is one reported anecdote from the editorial, not a typical price for AI images and not a measure of the total social or environmental cost of producing one.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →#1 Best Overall
The distinction matters: a token charge describes what a user paid under particular circumstances. It does not, by itself, account for electricity, data-center infrastructure, or impacts borne by others. Conversely, pointing to those wider concerns does not establish the scale of any particular impact. The editorial’s argument is strongest as a reminder to ask who pays and what is being counted—not as a quantified accounting of AI’s costs.
Separate observable costs from dramatic forecasts
Lapienytė names electricity use, job disruption, environmental strain, security risks, and the scanning of books for AI training as concerns. She also discusses public predictions of catastrophe, including claims involving enormous death tolls. The article does not provide a full evidence review for those forecasts, so its critique should be read as a point about how such claims are debated, not as proof that catastrophic risks are impossible.
Her objection is that some predictions are difficult to prove or disprove and may therefore be assessed largely through the authority of the person making them. She compares that dynamic to conspiracy theories. That is the author’s assessment of the debate’s framing; it does not establish that every long-range risk claim is unfalsifiable or unsupported.
- Present-day effects: Ask what is being measured, by whom, where, and over what period. A specific claim about electricity or employment needs evidence tied to the relevant system and location.
- Local versus national effects: National electricity-price trends do not necessarily reveal pressure on a particular grid or community. Cybernews has reported on this distinction in its coverage of AI data centers and electricity prices; that reporting is context, not a substitute for checking the cited local utility or grid sources.
- Long-range predictions: Ask what would count as evidence for or against the forecast, what assumptions it depends on, and whether the speaker distinguishes possibility from likelihood.
Take security and book-scanning claims in context
The editorial refers to reporting about an alleged AI-agent access incident and to book scanning. Its linked coverage offers context, but the editorial does not independently establish those underlying claims. Treat them as reported examples rather than verified findings unless the original record or publisher confirms the details.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesRank #3
This distinction avoids two errors: dismissing a concern simply because the most alarming version is uncertain, and repeating an allegation as fact because it appears in commentary. For security incidents, the useful questions are what system was involved, what access occurred, and what evidence has been made public. For book scanning, establish who scanned the books, under what circumstances, and what is known about their use before drawing broader conclusions.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.A more useful way to argue about AI
The editorial’s tension is worth preserving. AI can have real costs and risks even when some predictions about its future are hard to test. Skepticism should apply both to sweeping claims of imminent disaster and to easy reassurances that a low user price means there is no wider cost.
Quick Recap
- Distinguish a measured effect from a forecast or allegation.
- Keep the scope attached to the claim: one user, one facility, one community, or a national trend are not interchangeable.
- Separate the author’s opinion from the evidence offered to support it.
- Do not treat uncertainty as proof of either safety or catastrophe.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
