You need more skepticism in your life.
It wasn’t always like this. You used to be able to trust what you read and watched without needing to categorize the authenticity. Humans are naturally gifted at separating visual fact from fiction. In more primitive times, survival depended on our ability to make fast, accurate decisions about potential threats.
It’s not only physical, but mental, as Daniel Kahneman states in Thinking, Fast and Slow -
A general ’law of least effort’ applies to cognitive as well as physical exertion. The law asserts that if there are several ways of achieving the same goal, people will eventually gravitate to the least demanding course of action.
Generative AI easily bypasses this lazy evaluation. For years, spotting generated content was laughably easy. But today, you’re much more likely to accept false opinions and situations as reality simply because they appear normal.
You didn’t actually believe that Will Smith was eating spaghetti , or that Harry Spotter was a legitimate Warner Brothers project. Your mind distinguished fact from fiction right away, and it didn’t take any energy because most people using the technology haven’t fully crossed the uncanny valley.
But that’s quickly changing.
To nobody’s surprise, generative AI is capable of making content that bypasses the fast judgments your brain makes to distinguish reality. You can’t rely on the visual presentation to determine legitimacy anymore. In fact, you might not even think to question whether something is real or not.
The absurd and goofy isn’t a problem - we can (usually) filter that out with context clues. It’s the subtle, seemingly harmless content to watch out for. The TED talk. A podcast snippet. A strong opinion from someone filming in their car. We’re likely to buy into realistic settings without effort because it’s literally exhausting to question the legitimacy of common situations we’ve trusted our entire lives.
For years, we’ve trusted our minds to filter reality, but pseudo-reality has finally peeked over the horizon of judgment and it’s here to conquer our discernment.
But the invasion of our cognitive filters doesn’t stop at our eyes and ears. It extends into the very numbers we use to measure reality.
I recently followed an urge to deepen my understanding of probability and statistics to improve my mastery over data. As a proper dork, I bought a textbook to work through and several high profile books on the subject: The Signal and the Noise by Nate Silver and Naked Statistics by Charles Wheelan.
My innocent adventure into the field revealed horrifying insights around how information is communicated and interpreted.
Statistics is a powerful tool for communicating truth, both in how it’s grounded in the trustworthy institution of mathematics, and the subtle way it can easily trip up even the most careful connoisseurs of the craft.
Data and statistics are not the objective, trustworthy representation of truth we assume they are. They’re easily shaped by human interpretation, agendas, methods, and philosophies - regardless of how pure the intentions behind them.
This isn’t surprising - people and institutions twist reports to fit their needs all the time. Our minds are already on guard when we hear data we don’t like. Or from someone (or something) we don’t trust.
It’s the statistics we agree with or don’t even think about that we ought to spend more time challenging, when our minds don’t question the presentation. Statistics have several layers worth knowing about, simplified by my amateur understanding of the process:
First, the hypothesis. What is the report trying to actually accomplish before any data is collected or reviewed? Good science suggests a hypothesis should be repeatable and falsifiable. Some schools of thought say to build a null hypothesis, which means the study should attempt to overturn the belief that the subjects in the study are not related. The hypothesis forms the basis of how a study will be conducted. If you enter a study already believing you’ll find evidence to support a result, you may unconsciously (or directly) favor positive and hypothesis-affirming methods, populations, and outcomes even when there may be more suggestive evidence to the contrary.
Second, data collection. Where do you look to find evidence related to the hypothesis? What constants do you control for? How much data is required in later steps? Expert statisticians exercise extreme care when selecting the right groups and variables to evaluate. Studying the wrong population builds inaccurate assumptions directly into the results, regardless of how perfectly researchers execute math during later analysis.
Finally, the analysis. How should the raw data be analyzed? How well does it relate to the hypothesis? Is it something worth publishing in a journal? Human judgment collides again with raw statistics and math.
There’s a difference between reporting an average and knowing if the average is even fair to evaluate. In Naked Statistics, Wheelan demonstrates an example where the average income in a bar seating 9 middle class patrons might be about $55,000 per year. But if Bill Gates walks in and sat down, the average would skyrocket -
If I were to describe the patrons of this bar as having an average annual income of $91 million, the statement would be both statistically correct and grossly misleading.
Extending beyond the nature of the numbers, the conclusions are also subject to misinterpretation. When research disproves a null hypothesis - essentially finding out that there are no associations between two variables - that doesn’t automatically prove the alternate hypothesis true. It just means it’s unlikely there is no relationship between the variables but doesn’t define what that relationship might be.
Rejecting the null hypothesis doesn’t automatically prove your theory true. If a study proves that school funding and literacy rates are not unrelated, it doesn’t automatically mean throwing money at a school fixes reading levels. It just means a relationship exists. The exact nature of that relationship is where human bias easily creeps back in.
(the “chance”, by the way, is determined in advance by the researchers… another point of subjectivity injected into a relatively objective field of study)
The trustworthy application of statistics to raw data does not inherently create trustworthy research.
Our minds don’t want to do the legwork to question everything we receive. Especially favorable research or reports from trustworthy, published sources. Honestly, we don’t have time. But the institutions we typically trust to conduct trustworthy work on our behalf - the published scientific community - aren’t immune to the pitfalls of poor statistics. Dr. John Ioannidis published research in The Journal of the American Medical Association suggesting that roughly half of all scientific papers published will eventually turn out to be wrong.
I’m raising awareness to live with more skepticism. More scrutiny. To exercise discernment when consuming content or reading reports.
It’s not that creators and researchers are out to manipulate us. At least not on purpose. But we no longer live in a world where trusting your natural senses to suss out good from bad, real from fake is a viable strategy. If you aren’t careful, the algorithm will dictate your POV instead of challenging it.
Here are a few strategies for navigating modern information:
Consume with care. It’s increasingly rare, and not even trivial, to curate your own content. An overwhelming majority of people allow platforms to decide what to consume instead of making a deliberate choice. This allowance comes with a much higher risk of encountering shadow generated content that easily bypasses your normal judgment. Be smart about the amount of time you spend here and the level of influence you surrender to platforms.
Think like a fox. Nate Silver argues that a more productive relationship with our beliefs and data comes from acting like a fox rather than a hedgehog. Foxes allow incoming data to challenge and shape their beliefs, constantly working towards approaching and refining the truth. Hedgehogs start with an outcome in mind, and seek data and inputs that affirm what they already believe instead of allowing it to challenge their assumptions about the world. Unfortunately, most people are hedgehogs and don’t even realize it.
Filter information in, not out. Foxes refine their beliefs based on new information, but not all incoming information is useful or trustworthy. Change your stance to make it hard for information to even be considered in the first place. Viral information, big claims, and proven studies all deserve extra scrutiny. It used to be relatively harmless to trust most things. Blind trust comes with more risk than ever before because the source of information is deceptively legitimate.
Actively participate in the ideas that influence your thinking and be slower than ever before to trust anything you can’t validate or touch & see in real life. Yes, it’s exhausting to be a skeptic. It takes a lot of work to validate what you see, read, and experience.
But the alternative is surrendering your worldview to an algorithm or flawed statistic.