[1] https://thehill.com/homenews/campaign/4920827-60-minutes-tru...
[1] https://thehill.com/homenews/campaign/4920827-60-minutes-tru...
But it is. Numbers can be twisted, but it they can easily be verified. Bias, bad sources and cherry picking can allow you to tell stories, but the data will allow you to verify those stories are indeed facts. Brain can’t really fact check things that don’t have any data.
Even if the numbers are accurate, nearly any situation has a nearly infinite number of potential data points, and deciding which ones are relevant isn't as straightforward as people act like it is.
This is easy to see play out; you can look at the same stories being reported on both Fox News and MSNBC. Usually both sources' raw facts will be basically "correct" in the sense that they're not saying anything explicitly false, but there can be bias in determining which facts are actually useful or how they're categorized.
You can see how the reporting of the January 6th stuff varied between news outlets.
Scientists: X number of people died of Covid in the US according to CDC data.
US Government: you can't prove that number. That data doesn't exist on government servers, the data in the copies is fake and can't be trusted.
Numbers are extremely useful, but numbers alone mean absolutely nothing.
It's a simple example, that's why it's relevant. All the facts are available for anyone to see, to process, to analyze. There is no disputed or hidden data. And yet nobody, including any AI, can produce a "true" answer to the question, because it's reliant on one's personal biases.
Even with Covid, did a 92-year old die because of Covid, or because of a multitude of existing conditions that Covid triggered? Probably impossible to know medically, and AI isn't going to tell you definitively one way or the other.
- COVID-19 Death Overcount: In 2022, a coding error led the CDC to overcount 72,277 COVID-19 deaths across 26 states. Source: https://www.theguardian.com/world/2022/mar/24/cdc-coding-err...
- Maternal Mortality Data: Changes in death certificate reporting, particularly the addition of a pregnancy checkbox, resulted in overcounts of maternal deaths due to false positives. Source: https://www.theatlantic.com/podcasts/archive/2024/08/materna...
- Lead Exposure Report: A 2004 CDC report underestimated the impact of lead-contaminated water in Washington, D.C., leading to criticism over its data accuracy. Source: https://en.wikipedia.org/wiki/Morbidity_and_Mortality_Weekly...
- Property System Data: An audit revealed that the CDC's property system data was neither accurate nor complete, with an estimated $29.2 million of property at risk of being lost or misplaced. Source: https://oig.hhs.gov/reports/all/2016/centers-for-disease-con...
These instances highlight that data, even from reputable sources, can be subject to errors, misinterpretation, or manipulation, underscoring the need for critical analysis beyond face-value acceptance.
If the question was who scored the most points in the year, that can be answered factually by data.
If the NFL was deleting all their data at the end of the season with the goal of creating arguments and sowing disinformation, that would be a more relevant example.
Barring said data being fabricated, deleting data seems to be a sign of bad faith.
A more realistic example: we can theoretically predict the weather weeks in advance. In reality, it's pointless because there so much data needed to collect for that, and so many events to away the weather, that's its impractical past a few days in the future.
That's the point of data. To get us closer to the truth. Gravity will keep making you cling to the earth no matter your opinion. Even though as we speak we are still trying to develop models to properly understand the particles or forces behind gravity.