The Arms Race for Truth

Or: How to be your own fact checker in the Disinformation Age.

Unless you live under a digital rock, by now you’ve read a dozen or more posts about Meta’s recent decision to replace independent fact checkers with something they’re calling “Community Notes.” Zuckerberg’s statement says that this is all about promoting more free expression. Now I loves me some free expression - in art and satire and humor - and I believe everyone is entitled to their opinion. But I’m struggling to connect the dots between facts and free expression. The last time I checked, no amount of freedom can make truth out of claims like “2 + 2 = 5” or “Abraham Lincoln is alive and working the last checkout line at Walmart.”

The timing of this latest frontier in social media enshittification is…concerning, coinciding as it does with the swearing-in of a POTUS whose primary bully pulpit is The App Formerly Known As Twitter and who has made enough provably false public statements to merit a dedicated Wikipedia page. It feels like we're being blindfolded while walking into traffic.

He who controls the facts controls the universe.

This whole thing got me thinking about the role "knowing the facts" plays in society—and how much it matters for how we function as a unit (or dysfunction, as the case may be). We humans figured out pretty early on that we’re better at surviving in groups than on our own. Whatever ground we’ve covered as a species over the last 200,000 years is the direct result of the ongoing argument over how society should work. And for all that arguing to matter, people have to agree on basic facts.

Humanity has improved our collective lot through the relentless pursuit of understanding. Understanding is the thing that evolves as new facts come to light. The dissemination of facts has always been an arms race. When fewer people have access to the best factual information available, power concentrates. It’s just harder for the few to dictate the lives of the many when information is free.

When information could only be transmitted through parchment and quill, the knowledge was hoarded by the church and the state. Then the printing press comes along and boom - information everywhere. Fast forward to the consolidation of newspapers and later television into relatively few hands, when the manufacturing of consent was so effective that we still lament the passing of “the good old days” that we know to be largely mythical. We know because the rise of the internet took us into an Information Age where facts were just a click away, at least for those of us willing to run them down. The problem is, we’ve gotten so into our free expression that we’ve forgotten what a fact is. This makes us easy pickins for disinformation.

You keep using that word.

So, let’s get something straight here: there’s a huge difference between objective facts, hypotheses about causality, and plain old opinions.

• Objective facts: These are the things that happened. “The Titanic sank after hitting an iceberg on April 15, 1912” is an objective fact—no debate, no spin.

• Hypotheses about causality: This is when we try to explain why things happen. For example, “The Titanic sank because the iceberg breached the ship’s hull and caused it to flood.” This is a reasonable hypothesis, backed by tons of evidence, but it’s not a fact—it’s an explanation. We can prove or disprove it, but we have to dig for evidence (more facts) to do so.

• Opinions: These are the fun ones. Opinions are personal. “The Titanic disaster was an example of humanity’s hubris and disregard for nature’s power.” Opinions are what we all love to share at dinner parties and on Facebook.

But here’s the thing: free expression, in this context, only really means that everyone is entitled to their opinion. That’s it. If you want to express an opinion about pineapple on pizza (I personally think it’s a travesty), go for it. Let that be the guiding light of your life. Nobody cares. But don’t mistake that for a universal truth.

Yes, Virginia, there is an intersubjective reality.

Here’s where it gets interesting. While the stakes are low when it comes to opinions of personal preference where facts are irrelevant, when it comes to the big issues—justice, fairness, and how society should function—opinions matter because they influence how a given society functions. In a democracy - assuming that’s still the goal here in the US - the policies that affect our lives are shaped by public opinion. The “free expression” disinformation as fact is a problem because most people form their worldview based on the news they consume. And for better or worse, social media is where most people get their news. According to a Pew Research Center report, just over 50% of US adults get at least some of their news from social media.

People feel strongly about their opinions. The more disinformation circulates, the more misinformed opinions are formed. And that leads to policy options that might be completely at odds with what’s best for society. In other words, misinformation doesn’t just make people look foolish; it distorts democracy. If you think the country’s running off the rails, it might be because the tracks were never properly laid in the first place - thanks to a bunch of half-truths and alternative facts on your Facebook feed.

Meta’s decision to ditch independent fact-checkers comes at a time when fake accounts and disinformation campaigns are rampant on social media. It’s not hard to imagine how crowd-sourced truth could be subverted by bad actors. Read this longish transcript of my conversation with ChatGPT about it. If fake accounts can spread conspiracy theories and manipulate public opinion in elections (we saw it in 2016), why wouldn’t the same thing happen with Community Notes?

The absence of clear standards and guidelines in Meta’s new system means that bias could go unchecked, and both the facts of an event and the hypothesis behind it could be cherry-picked to support whatever narrative someone wants to spin. How will they handle bias? How will they differentiate between the facts (what happened) and the causal hypotheses (why it happened)? Will people just vote on everything? How do you prevent the facts from being hijacked by organized groups with an agenda? This is not a recipe for objective truth—it’s a formula for a hot mess of disinformation.

You’re gonna need a bigger boat.

So what’s a person living in the Disinformation Age to do? Our public education system has spent decades teaching kids how to punch a clock, not how to think critically. They’re teaching us how to become cogs in the machine instead of well-informed citizens capable of distinguishing fact from fiction. PS: Googling fringe articles to back up your opinions isn’t research.

I don’t know about you, but my day job doesn’t leave enough time to do my own investigative journalism. In an arms race for truth, the only ready answer to disinformation technology is… more technology. Lately, I’ve been experimenting with Gen AI tools like ChatGPT and Claude AI for both fact-checking and opinion testing. Let me tell you, these tools are like having dinner with a superpowered librarian who’s tired of all the bullshit. You’ll find complete transcripts of some lightweight fact checking I did with both of these bots below.

Like any weapon, AI is not without risk. The heavy hitters are for-profit companies who - spoiler alert - may not have the best interests of humanity at heart. That’s a whole other debate (Bruce Schneier's take is worth reading) - but for now, these tools are doing a pretty good job of helping me filter through the noise. I’m still learning, and I encourage you to try this for yourself and see what happens. Share your experience in the comments here, so we can all get smarter together.

We need to question everything, and we need to be aware that right now, we’re losing the arms race for truth. In a world where everyone’s screaming their opinions, and truth is increasingly hard to find, it’s on each of us to separate the facts from the noise. It’s up to us to be the truth-seekers, the fact-checkers, and the critical thinkers. Because, at the end of the day, if we’re going to make the right decisions as a society, we need to know what’s actually true.

---

---

Transcript of fact checking with ChatGPT and Claude AI

I provided the same prompts to both bots. I’ve merged their responses to make comparison easier. ChatGPT provided links to sources in many places (not included below), which is really nice despite being limited by the age of the training dataset resulting in the occasional dead link.

Brad: Are you capable of fact checking?

ChatGPT: Yes, I can fact-check by searching for and verifying information using trusted sources. If you have something specific you'd like me to verify, let me know!

Claude AI: I can help analyze claims using my knowledge (trained through April 2024), but I cannot access current information or external sources. For accurate fact-checking, especially of recent events, I recommend consulting reliable fact-checking websites and primary sources directly.

Brad: Define "fact checking."

ChatGPT: Fact-checking is verifying information for accuracy and truthfulness by consulting reliable sources and cross-referencing evidence.

Claude AI: Fact checking is the process of verifying the accuracy of claims, statements, or information by comparing them against reliable sources, data, and evidence to determine if they are true, false, or misleading.

Brad: How do you assess whether claims of causation are factual?

ChatGPT: To assess whether claims of causation are factual, follow these steps:

1. Establish correlation: Confirm a consistent relationship between variables.

2. Check for temporal precedence: Ensure the cause occurs before the effect.

3. Rule out confounding variables: Identify and eliminate alternative explanations.

4. Look for mechanisms: Verify a plausible, evidence-based explanation for how the cause leads to the effect.

5. Review controlled studies: Evaluate experimental designs (e.g., randomized controlled trials) for strong evidence.

6. Examine replication: Ensure findings are consistent across multiple studies.

7. Assess source credibility: Verify the expertise and reliability of the source making the claim.

Would you like help applying this to a specific claim?

Claude AI: To assess causation claims, I examine:

1. Correlation strength and consistency

2. Temporal sequence (cause precedes effect)

3. Control for confounding variables

4. Evidence of mechanism (how A leads to B)

5. Reproducibility of findings

6. Alternative explanations ruled out

7. Dose-response relationship if applicable

Statistical methods like regression analysis and controlled experiments provide stronger evidence than observational studies alone.

Brad: Fact check the following statements: "No healthy individual should die or be injured from a vaccine. We are now approaching a time when the injuries and deaths from COVID vaccines will likely surpass those from the virus. Between Dec. 14, 2020 and Aug. 13, 2021, a total of 595,622 total adverse events were reported to the government-run Vaccine Adverse Events Reporting System (VAERS)."

ChatGPT: Let's examine each statement for accuracy:

1. "No healthy individual should die or be injured from a vaccine."

   While vaccines undergo rigorous testing to ensure safety, no medical intervention is entirely without risk. Rare adverse events can occur, even in healthy individuals. However, the benefits of vaccination in preventing disease generally far outweigh these risks. For instance, reports of death after COVID-19 vaccination are rare, with more than 331 million doses administered in the U.S. from December 14, 2020, through July 6, 2021, and 5,946 reports of death (0.0018%) among vaccine recipients.

2. "We are now approaching a time when the injuries and deaths from COVID vaccines will likely surpass those from the virus."

   This claim is not supported by current evidence. As of December 15, 2024, over 7 million COVID-19 deaths have been recorded globally, with the World Health Organization estimating the real death toll to be at least 20 million. In contrast, while adverse events, including deaths, have been reported following COVID-19 vaccination, they are exceedingly rare compared to the mortality caused by the virus itself. For example, during December 14, 2020–January 13, 2021, VAERS received 6,994 reports of adverse events after COVID-19 vaccination, with 640 (9.2%) classified as serious, including 113 deaths.

3. "Between Dec. 14, 2020 and Aug. 13, 2021, a total of 595,622 total adverse events were reported to the government-run Vaccine Adverse Events Reporting System (VAERS)."

   This figure aligns with data from that period. It's important to note that VAERS is a passive reporting system, and reports can be submitted by anyone. The reports are not verified and do not establish causality. They serve as an early warning system to detect potential safety issues that require further investigation.

In summary, while adverse events following vaccination can occur, they are rare, and the benefits of COVID-19 vaccination in preventing severe illness and death far outweigh the risks. Claims that vaccine-related injuries and deaths will surpass those caused by the virus are not supported by current evidence.

Claude AI: Let me fact check these claims:

1. "No healthy individual should die or be injured from a vaccine"

This oversimplifies vaccine safety. While severe adverse events are extremely rare, any medical intervention carries some risk. Even well-tested, safe vaccines can occasionally cause reactions in healthy people.

2. "Injuries and deaths from COVID vaccines will likely surpass virus"

False. COVID-19 has caused millions of deaths globally. Verified vaccine-related deaths are extremely rare. The claim lacks evidence and contradicts extensive safety data.

3. VAERS data citation

The numbers are accurate but misleading without context. VAERS:

- Accepts all reports without verification

- Cannot establish causation

- Explicitly states reports alone cannot determine if vaccines caused adverse events

- Is intended as an early warning system, not a verification tool

For accurate vaccine safety information, refer to peer-reviewed studies and established medical institutions rather than raw VAERS data.

Brad: Is childrenshealthdefense.org a reliable source? Why or why not?

ChatGPT: Children’s Health Defense (CHD) is not generally considered a reliable source for scientific or medical information. Here’s why:

1. Bias and Advocacy

• CHD is an advocacy organization led by Robert F. Kennedy Jr., known for promoting anti-vaccine narratives and skepticism of public health measures. Its content often reflects a strong bias against vaccines and government health agencies.

2. Misinformation

• The organization has been criticized for spreading misinformation about vaccines, including unsubstantiated claims about vaccine safety and effectiveness. Public health experts and fact-checking organizations have frequently flagged CHD for disseminating misleading or false information.

3. Lack of Scientific Rigor

• CHD often relies on anecdotal evidence, cherry-picked studies, or out-of-context data to support its claims. It does not consistently adhere to the standards of peer-reviewed scientific research.

4. Reputation Among Experts

• Reputable health organizations, such as the Centers for Disease Control and Prevention (CDC), the World Health Organization (WHO), and numerous fact-checking sites, have explicitly countered CHD's claims, emphasizing that its content undermines public health efforts.

5. Legal Challenges

• CHD has faced lawsuits and criticism for spreading harmful narratives that contribute to vaccine hesitancy, which can lead to public health risks.

Conclusion:

While CHD may present itself as an advocate for health and safety, its information is often misleading and not supported by the broader scientific consensus. For reliable health information, it is better to consult peer-reviewed studies or reputable sources like the CDC, WHO, or major medical institutions.

Claude AI: Children's Health Defense should not be considered a reliable source because:

1. It consistently promotes debunked anti-vaccine claims and conspiracy theories

2. Its content contradicts well-established scientific consensus

3. It misrepresents or incorrectly interprets scientific studies

4. The organization has been cited by multiple fact-checking organizations for spreading misinformation

5. The site's founder, Robert F. Kennedy Jr., has been banned from some social media platforms for repeatedly sharing vaccine misinformation

While the site presents itself as an advocacy organization for children's health, its claims and interpretations frequently fail basic fact-checking criteria and contradict evidence from peer-reviewed medical research and public health institutions.

For reliable health information, consider sources like:

- Peer-reviewed medical journals

- Major medical institutions

- Public health organizations

- Academic research centers

---

Creative Process Transparency: This article was collaboratively written using generative AI. The initial argument and framing were developed by me, with ChatGPT assisting in drafting certain sections. The final structure and narrative choices remained with me. Based on our established attribution model, this article reflects an approximately 75/25 division of contributions.