Hold the Hallucinations

A look at the new reality check.

Jun 10, 2025

"In an age of automated answers, our most valuable skill is the courage to question them. Trust, but always verify."— Nadina D. Lisbon

Hello Sip Savants! 👋🏾

When did we start taking an AI's word as fact? It’s a question we need to ask ourselves, because it’s already happening with serious consequences. This past Friday, England's High Court had to formally warn lawyers for blindly trusting their AI, which had simply invented 'hallucinated' legal cases for a court filing ¹. This isn't a rare glitch; it's a built-in risk of the technology we're rushing to adopt. It exposes a strange contradiction: recent polls show that while over half of us use AI, a staggering 66% are highly concerned about getting inaccurate information ⁵. If we don't trust it, why are we taking its word? The courtroom warning is a clear sign that our blind faith needs an immediate reality check.

3 Tech Bites

⚖️ Contempt of Court, by AI
- The legal world is grappling with AI-generated fiction. In a stark warning delivered last Friday, England's High Court declared that lawyers who submit 'hallucinated' legal cases invented by AI could face serious sanctions ¹. This follows similar incidents in the US and sets a crucial precedent: professionals are responsible for the facts.
🔍 Google's AI Stumbles
- The public rollout of Google's "AI Overviews" has become a lesson in digital absurdity. Users have documented it confidently advising them to put glue on pizza or claiming a US president graduated from a university he never attended ². This high-profile example shows that even the biggest names in tech are struggling to prevent their AI from confidently presenting dangerous nonsense.
🤖 Can We Fix The "Dreaming"?
- Is there a cure for AI hallucinations? Some experts believe so, and they call it Neurosymbolic AI. Unlike current models that just predict patterns, this hybrid approach combines pattern-recognition with a rules-based reasoning ability, similar to human logic ³. The goal is to teach AI why something is true, not just to mimic what it has read online.

5-Minute Strategy

🧠 The "Fact vs. Fiction" AI Check

Before you trust or share any AI-generated content, run it through this simple 3-step reality check. It takes less than five minutes and can save you from serious errors ⁴.

Demand a Source: Ask the AI directly: "What is the source for that statement?" If it can't provide a specific, real URL or publication, treat the information as unverified. Remember, AIs can also "hallucinate" sources, which leads to step two.
Cross-Reference the Claim: Paste the claim into a search engine like Google or Bing, but immediately scroll past the AI-generated summary at the top. Your goal is to look at the actual list of results. Click on multiple, independent, reputable sources (like established news organizations, academic sites, or official reports) to see if they all confirm the information. If you can't find it anywhere else, it's almost certainly a hallucination.
Perform a Sanity Check: Read the answer out loud. Does it sound logical and reasonable? Or is it just a string of plausible-sounding jargon? AI often lacks real-world common sense (like advising you to eat rocks or put glue on pizza). If it feels off, it probably is.

1 Big Idea

💡 The New Digital Literacy: From "Knowing" to "Verifying"

For the last two decades, digital literacy meant knowing how to find information. We mastered the art of the search query, learning how to ask the right questions to navigate a sea of data. The AI era fundamentally inverts this dynamic. Now, the answer is given to us first, often packaged with a confident, authoritative tone. This demands a radical shift in our skills: from being an expert searcher to becoming an expert skeptic. The new literacy is no longer about the discovery of information, but about the rigorous validation of it.

This shift places a new and profound ethical burden directly on us, the users. We have become the final checkpoint between a plausible-sounding fabrication and a shared reality. It requires us to actively fight against the convenience of a ready-made answer. The challenge is as much psychological as it is technical; we must overcome the AI's confident tone and our own natural inclination to accept the easiest path to information. Every time we accept an AI's answer without question, we are not just risking being wrong, we are ceding a piece of our own critical authority ⁴.

This brings us to a cognitive crossroads, and the stakes couldn't be higher. In a future saturated with AI-generated content, will our critical thinking muscles atrophy from disuse, making us more vulnerable to misinformation and manipulation? Or will this challenge force us to evolve? We have an opportunity to develop a more sophisticated and resilient form of digital wisdom—one that is constantly questioning, cross-referencing, and using technology as a starting point for inquiry, not a final destination for truth. How we choose to engage with these tools today will determine whether they dull our minds or ultimately make them sharper.

TechSips with Nadina: The Newsletter

Discussion about this post