Knowing how to assess online sources is no longer optional in this digital era—it’s essential.
Algorithms can trap users in echo chambers, reinforcing existing beliefs and filtering out opposing views.
Guys, we have been using Google search for quite a while. The results were mixed: some information was accurate, but some was the result of fake articles posted online, and some links pointed to infected websites.
Since AI entered the web space, we can use a variety of super tools like ChatGPT, Grok, and many others. Thanks for the advances in technology.
More and more people are using those tools and find them amazing. AI is software that uses data warehousing (or data collection from a large database) through specially designed algorithms. That software is being “educated” by exposure to a large amount of information from various sources. There are many other features available in algorithms, such as image processing, project management, writing articles and letters, and more.
So, there is nothing magical about it, but the speed at which the answers are received always amazes me. It is the result of highly efficient, powerful data centers with server farms equipped with high-end microprocessors.
Unfortunately, those data centers are power-hungry, and our energy sources are stretched to the limit.
But this article is not about the energy problem; it is about getting the right answers by asking well-constructed questions.
I recently used ChatGPT to ask health-related questions that were very important for making an informed decision about how to treat the illness. When I constructed my question, I used a short phrase. I received an answer that I forwarded to my doctor, but later I discovered that the information I received was NOT correct.
Why did it happen?
Here is what I found.
If the algorithm has minimal information in your question, it is looking for answers based on HOW your question was constructed. Sometimes, AI “assumes” the answer. It has been found many times that AI is not an ideal tool. If it cannot find the correct reference, it may provide a completely wrong answer or simply supply you with incorrect information that may look like the right one.
It is especially true if you research stocks for investments. I have corrected ChatGPT on multiple occasions when I wanted to find the right stocks and ETFs for my portfolio.
So, how can you ensure that when using AI, you get accurate information?
First of all, you need to know that the paid version can perform an in-depth search and deliver more accurate information, although it may take a little longer. For ChatGPT, it is $21/month.
You need to be more precise in your questions. Here is an example:
Conduct in-depth research on both approaches to treating xxx based on my past input.
Use only proven data and sources. Do NOT guess or assume; use only reliable resources.
Show the source of your information (the link) so I can verify it.
Show the statistics for each side effect, if available from a reliable source: what percentage of patients experienced each side effect with the corresponding treatment.
Find the percentage of successful treatments without recurrence within 5 years for both approaches to treat xxx.
Be absolutely precise regarding the treatments for my preexisting condition. Elaborate, using reliable sources, on why xxx treatment is recommended or not recommended.
If you cannot find a resource in your database, show “NOT ENOUGH DATA”.
At the end, build the table that reflects your findings.
This approach produced reliable data, along with a list of the websites and other sources used to generate the PDF, which yielded a nicely compacted table with the needed answers.
The key phrase was “Do NOT guess or assume; use only reliable resources”
I hope you got the idea.
Now, let’s talk about the problems with the sources of information.
I want you to read about how fake news is created by giving you a real-world example from one fellow:
“I once worked in the communications department of a large organization and witnessed firsthand how quickly and easily falsehoods can spread online. In one instance, the organization wanted to capitalize on the arrival of a significant public figure on the world stage. Little was known about this figure, and everyone scrambled to dig up information about their past and views on hot-button issues.
In the fight to stand out among the deluge of information and the cacophony of voices online, every outlet wants to be the first to break significant stories—in this case, the first to unearth relevant quotes by the person in question. Credibility, popularity, and—ultimately—money depend on it.
For this reason, one social media staffer at my organization took a shortcut and used AI to find quotes from the public figure on key issues that we could repost on our social channels. The staffer submitted a query along these lines to some AI chatbot. The AI generated a few alleged quotations from the public figure, and the staffer, unfortunately, posted the content on our social pages as genuine without verifying its authenticity. As became clear later, the AI had made a mistake, and the quotations were either entirely made up or had been said by someone else and misattributed.
Nevertheless, within a few hours, the fake quotations our organization had unwittingly generated and published were picked up, copied, redistributed, and passed off as legitimate by at least half a dozen other organizations and outlets. Like multiplying bacteria, the alleged quotations appeared on numerous websites by the end of the day, convincing a casual researcher of their authenticity.
Yes, folks, simply like that…
Like that guy, I learned on my own not to blindly trust any information, whether it was generated by AI, posted on a website, Facebook, or other social networks, or shown in a YouTube video. Since then, I view any information with a grain of salt until I verify the sources and cross-references.
You should know that AI can generate videos using the faces of known political figures or imaginary people. The quality of the videos is amazing, and at first glance, you might think the information is accurate. In this case, ask the same ChatGPT if that information is proven (and ask about the sources if the answer is YES).
When evaluating information online, first do the following:
- Consider where it comes from
- Who wrote it?
- Who published it, and why?
- Was the source information in question created by someone with credentials or not?
- Does the author have a known bias? Understanding the author’s motive can further illuminate the information and add to or detract from its reliability. Is the author trying to make money? To control others? To fuel fear?
- If the information references no sources, just bluntly throwing out claims with no evidence to back them up, then it’s immediately suspicious.
- If the information provides verifiable data from reliable sources and evidence.
Remember, it is easy to create a professionally looking website that convinces you with crafted information. It can be generated by AI or designed manually, but it will always be questionable. In fact, hackers are masters at duplicating websites and luring you into either paying for a fake product or infecting your computer or smartphone.
A fake website’s domain name or URL often looks very similar to the original, with just one or two letters different. Be the judge of the information. If various independent sources confirm the same information, it is more likely to be true. That’s why it’s essential to trace the claim's origin—if everyone cites the same source, the “consent” carries less weight, while several sources may provide more weight.
As you and millions of users interact more with content that supports your beliefs, the website or search engine will provide more of it, further fossilizing your presuppositions. BE AWARE! Yes, one of the quite challenging aspects of evaluating online content is being aware of your own biases and how digital algorithms may amplify them
In addition to algorithmic suggestions, even our own search habits can reinforce our existing views. People tend to choose search terms designed to produce results that match what they already think.
Example? If you are politically liberal and interact more with certain content—content that supports your beliefs—the website or search engine will provide more of that content. Or if you ask the search engine or AI: “the benefits of consuming curcumin” vs. “the dangers of consuming curcumin”, you may get a biased answer.
So, my friends, by being aware of the tendency to seek only information that reinforces us, we can uncover more objective information and build a more comprehensive picture of reality.