Article Details

Bots, bias, and bunk: How can you tell what's real on the net?. You can improve the odds by combining skepticism, verification habits, and a few technical checks. Opinion Liars, cranks, and con artists have always been with us. It's just that nowadays their reach has gone from the local pub to the globe. Take, for example, when X, the social network formerly known as Twitter, introduced a new feature, "About this account," which includes information about a person's location. The idea was to make it easier to spot bots. It was successful. Way too successful for the taste of some, as it quickly became apparent that many Donald Trump-touting troll accounts, such as MAGA NATION, TRUMP_ARMY_, and CharlieK_news, for all their American-flag waving, are from non-US locations. They and their ilk are bot-driven propaganda sites. Thinking of X's brother service, the AI system Grok, its training includes such lies as former first lady Michelle Obama is a man and COVID-19 vaccines caused millions of unexplained deaths. Grok owner Elon Musk has also ordered its answers to reflect his opinions. For instance, the Brookings Institution reported that when an X user asked Grok to identify "the biggest threat to Western civilization," it responded with "misinformation and disinformation." Unhappy with this answer, Musk responded: "Sorry for this idiotic response. Will fix it in the morning." The next day, he released a new version of Grok that responded with "sub-replacement fertility rates" to the same question. This is a popular right-wing talking point. As for Musk's Wikipedia clone, Grokipedia, even Grok, when asked whether Grokipedia was trustworthy, replied: "Grokipedia is not a fair and unbiased source of information." So, what can you do about this? It's not like X and its associated services are unique. Far from it! Bots, AI spam, and good old-fashioned liars are everywhere. The first thing you should know is that, despite what some people will tell you, everyone is biased. There's no such thing as objective news. Sure, there are objective facts – the world is round, the speed of light in a vacuum is 299,792,458 meters/second, and the sky is blue. The trick is to recognize your own bias and the bias in the information you're consuming. So, let's start with a quick gut check. If something sounds too good or crazy to be true, it's likely not. No one's going to give you a million dollars if you just give them your bank account number. Nor did Hillary Clinton run a pizza-restaurant child-sex ring. So, always ask yourself, "who benefits if I believe or share this?" In the cases above, it's someone trying to get their hands on your cash and someone wanting you to vote for Trump. Next, "is anyone reputable reporting the same thing?" Who's reputable? The best guide currently out there is Ad Fontes Media's Media Bias Chart. Also, click through to the actual site or profile. Is it a known outlet, an official organization, or a verified account, and does the same item appear on their site? If it's just a friend passing along something that a friend sent, that's a red flag. If you're getting the "facts" from someone you know, you should also consider the source. For instance, I know a fair amount about some things –Linux, film noir, baseball, chess – so if I say something about those topics, I usually have a clue about what I'm talking about. If I'm yakking about cooking, American football, or poker, I don't have a clue. You should also look for precise details: named people, dates, locations, and links to primary documents instead of vague "experts say" or "scientists proved." For example, you may have noticed my articles tend to be stuffed with links. That's because while I have my biases – I'm liberal, an AI-cynic, and a big open source fan – I want you to know where I'm getting information from. Additionally, you should check the date and context. Far too many "new" viral posts are recycled, old content attached to a current event. For instance, if you look closely at a lot of photos and videos of terrible things happening during ICE protests in Portland, Oregon, you'll find that many of them were taken from the 2020 George Floyd protests. How can you tell? Look for visible evidence of the times. If some things, such as signs and graffiti, are blurred out, as Fox News did, that's very suspicious. Still not sure? Run the key claim through such dedicated fact-check sites as PolitiFact, FactCheck.org, Snopes, or AP Fact Check. You may also want to dig deeper by cross-checking names, numbers, location, and quoted laws or studies by searching them directly and looking for primary sources such as government sites, court documents, company releases, and peer-reviewed academic papers. I'm sorry to say, however, that lately you can no longer trust American government sites. Under the current regime, they're being twisted to reflect Trump and friends' prejudices. For example, under Robert Kennedy Jr., the US Centers for Disease Control (CDC) website has become an anti-vaccine site. Another thing that has fallen out of fashion, especially in the US, is respect for expertise. When it comes to vaccines, I respect real doctors' opinions more than I do political hacks with an ax to grind. Sounds like a lot of work to figure out what's true and what's not, doesn't it? Well, you're right. It is. That's one reason I make a living as a journalist. This work is part of what I do. Unfortunately, tools like AI text detectors, such as those from Grammarly, Copyleaks, and GPTZero, which are supposed to make it easier, aren't that good. They analyze language patterns and assign probabilistic scores, but they can mistake human writing for AI and vice versa. Sometimes an em-dash just means the writer likes to use em-dashes and not that it's the product of an AI chatbot. I'm one of those people. In the meantime, deepfakes and AI images are getting ever more difficult to detect. Last year, I could still spot most fake AI images, but I can't anymore. They're too good. We can still try. Do a reverse image search with tools like Google Images, Google Lens, and TinEye to see where and when a photo first appeared and whether it's being reused. You should still look for such visual red flags as inconsistent lighting or shadows, unnatural skin textures, warped backgrounds, mismatched reflections, or strange details like jewelry, hands, or text in the scene. For deepfake videos, watch the eyes and mouth for micro-movements: odd blinking; off lip‑sync; frozen or mask-like expressions; or glitches around hair and edges. These are all tells. I'm sorry to say that nothing on the internet can be trusted by default anymore. I can only suggest that you treat every hard to believe, too good to be true, or just downright fantastic claim as a hypothesis to test, rather than a fact to forward. In a world of bots and AI where we're drowning in AI slop, we can only keep paddling as best as we can in search of the truth.