Each one is a single self-contained page built to kill one specific wrong idea about AI and machine learning that slides reliably fail to kill. No install, no account, no network, no real model. Save the page and it still works. Companion to Security Toys.
Kills: “overall accuracy means it works”. The prettiest headline on the dial ships 29 of 50 faults, and a detector that detects nothing scores 95%; the threshold is a business decision about which error hurts more.
ReadyKills: “a document is just data”. Drag one hidden sentence into a document and watch the summary flip. Nothing on screen looked different.
ReadyKills: “no names, so it’s anonymous”. The nameless table opens at k=1, and one row is uniquely the CFO. Coarsen the columns and watch them vanish into a crowd.
ReadyKills: “the model knows things”. Same prompt, two runs, two answers; Sydney was always sitting there at 24%. It weighs, it doesn’t know.
ReadyKills: “higher training accuracy is better”. Drag complexity up: train climbs to 100% while test falls to 73%. The model memorised the noise.
ReadyKills: “95% per step is basically reliable”. Across a 12-step agent workflow it’s a coin flip, and the failures arrive polished. Checkpoints are placement, not diligence.
ReadyKills: “more efficient means less used”. Make it 10× cheaper per task and the bill quadruples; the whole paradox hangs on one toggle.
ReadyKills: “the clusters are in the data”. Step the algorithm by hand; two starts, two confident truths, ten documents that change sides. A cluster is a choice until a human names it.
ReadyKills: “AI decisions are mysterious”. Three factors, three weights, one threshold: change a weight with the data untouched and the verdict flips. The decision was never hiding in the data.
ReadyKills: “the data speaks for itself”. Three small charts, and in each one more than one confident story fits the numbers. The data never argued; your interpretation did the work.
ReadyKills: “there is a default AI technique”. Five clients, five briefs, five different right answers: the business question, the constraints, and the shape of the data pick the model every time.
ReadyKills: “technique selection is a technical detail”. Four client briefs, one 60-minute team sprint: frame the problem, pick the technique, defend the call in a three-minute pitch.
ReadyKills: “the summary statistics tell you what the data looks like”. Thirteen datasets — a dinosaur, a star, a bullseye — with identical means, spreads, and correlations. The stats panel never moves. Plot the data.
ReadyKills: “the average describes the typical person”. Drag the owner’s pay past two million: the mean sails beyond every real person while the median doesn’t blink. Which average, and whose money moved?
ReadyKills: “a line through the middle is the best fit”. Draw your line; every point grows a square of its error. The machine’s line looks wrong and carries the smallest total square area any line can. Try to beat it.
ReadyKills: “training figures out the right answer”. The ball checks one number — the slope under its feet — and steps the other way. Where it lands is an accident of start and step size, not of knowing.
ReadyKills: “survival went up, so the treatment works”. Start screening earlier and survival after diagnosis climbs while every death date stays put. A statistic measures where you started counting.
Ready