Back in the early pandemic days, a comment from a statistician online struck gold for me.

It was an online discussion on NYTimes about the mortality rate of COVID-19.<br><br>I wish I had saved the exact quote, but in essence, he said that we shouldn't trust the mortality rates being quoted because of how they're calculated. We're not dividing the real number of dead people by the real number of people with the disease. We're dividing the number of officially recognized deaths by the number of confirmed cases.<br><br>And he stopped right there, without saying a word about testing capacity.<br><br>But to me, this comment was a goldmine. I knew that there's a big gap between the real number of cases and the number of confirmed cases. And I also knew that there's a big gap between the real number of deaths and the number of officially recognized deaths.<br><br>It's easy to see why there's a gap between real cases and confirmed cases. Sick people couldn't get tested. There was a shortage of tests. <br><br>It's harder to see why the NYTimes wouldn't recognize a death from COVID. It's not like you need a test to confirm that a dead person was sick. But the same obstacles still exist. Since sick people can't get tested, how would we know that they're sick? Without a test, how would we know whether they died of COVID?<br><br>Here's the goldmine: if sick people couldn't get tested, **and** dead people can't get tested, **then the confirmed cases and confirmed deaths were both missing the same population of people.**<br><br>What does that mean?<br><br>It means that if we know that 50% of cases are going untested, **then we also know that 50% of deaths are also going unreported.**<br><br>And if 50% of both cases and deaths are going unreported, then **the mortality rates being quoted were accurate.**<br><br>The proportion of unreported cases and deaths were the same, so dividing one by the other cancelled them out.<br><br>This brings me to AI. Many of us believe that AI has already taken over many jobs, but how can we know how many jobs have been displaced by AI if most of them aren't being reported? How would we even know? It isn't like there's a statistician at the NYTimes to tell us how many jobs are being displaced.<br><br>The truth is that there are tons of NYTimes articles, from 2023 onwards, about how AI has already taken over tons professional jobs. Most, or possibly the overwhelming majority, of these articles aren't about how AI **will** take over these jobs in the future. They're about how AI **is** taking over these jobs **now.**<br><br>Maybe we can't know how many **total** jobs have been displaced by AI. But that's not the point. The NYTimes isn't just reporting how many jobs **will** be lost in the future. **We already know, from mainstream media reports, that most of these jobs have already been displaced. **<br><br>AI has already been displacing jobs **for years. **<br><br>This isn't some alarmist narrative connecting dots with no evidence. Every single day, we're getting more and more reports from **mainstream media** and **mainstream academia** about how **nearly every** single professional job has already been displaced in **some** capacity.<br><br>We already know that AI is a substantive threat to every single job. But how can we know the extent of that threat?<br><br>The answer is **we don't need to.**<br><br>It doesn't matter how many jobs have already been displaced. We don't need to know the statistic. **It's irrelevant.**<br><br>It's irrelevant **because there's a statistic that we do know. **<br><br>**One.**<br><br>We know that **one** single job has already been displaced by AI. **And that's enough.**<br><br>It was an online discussion on NYTimes about the mortality rate of COVID-19.<br><br>I was online when I first heard that a deepfakes voice AI program replaced an actress in a Pixar movie.