Gary’s Q&A

Listener question & Gary’s response · Transcript passage

Kevin O’Keefe from the Facebook groups asks, Gary, How do you use AI in your research now, and how do you know if your sources are accurate? Just very curious how you evaluate all the information you’re getting from the internet in terms of what sources are more accurate than others?

There were many AI-related questions, and I’ll try to address them all here. As most of you know, I use AI to create the cover art for many episodes. This is because certain topics don’t have any images available, or the topics are so broad that it is difficult to find an image to represent them.

I’ve had people complain about this and have said that I should hire humans to create it. That will never happened. I’ve worked with many graphic designers in my life, and the process of getting custom artwork would take days at a minimum, if not weeks, and cost at least hundreds of dollars per image.

Both the budget and time constraints render that impossible for this show. If I couldn’t use AI images, I just wouldn’t use cover art for episodes.

I use AI as tool to understand things better. I use it as a tutor, a topic I’ll be addressing in my book, which I am working on.

For example, I’ve been working on an episode about quantum computers. Before I do the quantum computing episode, I feel that, on a scale of 1 to 10, I have to understand it at a 6 or 7 level to do an episode that explains it at a 4 or 5 level.

The beauty of using AI is that you can ask it questions about anything you don’t understand. No question is too dumb.

I’ve also used it to solve things I couldn’t find. I could swear I once read a quote by President James Monroe about the ‘tyranny of space’, that being the vast overland distances in the United States were a huge problem. I asked an AI agent where the quote came from, and it couldn’t confirm it. It was probably just my imagination and the Mandela Effect.

Likewise, I had an episode idea about the five laws of stupidity, which I had read were attributed to Dietrich Bonhoeffer. Again, no matter how much I searched, I couldn’t find a source. When I asked an AI agent, it told me that it wasn’t Dietrich Bonhoeffer, it was actually the Italian economist Carlo Cipolla. Mystery solved.

I’ve had some weird comments over the last month, in which people have accused me of using an AI voice to create the podcast. Those of you in the completionist club who remember me recording episodes while I was almost hoarse from a bad cold might chuckle. But it has gotten to the point where anything posted online can be accused of being generated by AI, and I’m not sure how you can prove a negative.

I’ve used Grammarly for over ten years and have used it on every episode since episode one. It was just software that was a glorified spelling and grammar checker, but now they claim it is powered by AI. I don’t know how much of what it does is actually AI or if it is just marketing.

I’m also working on an episode about artificial intelligence, but it is such a rapidly changing field its hard to pin down. I’m looking at installing an AI agent called Open Klau on one of my spare computers just to play around with it. I’ve found the best way to learn things is usually just to jump in with both feet.

As to the biggest question you asked, how do you verify a source, this is actually an issue that has existed for thousands of years. Myself, Joel, and Olivia have all encountered the same problem that different sources will often give different dates or names for the same event. Usually, the difference is only a few years apart, say 1842 or 1844. Rather than spending hours trying to pin down the correct date, the simpler solution for the purpose of script writing is usually to just say “early 1840s.”There are many historical sources, often primary sources, that differ in their accounts of basic facts. Basically, you just try to look for consensus. It is an issue that is actually further up the food chain from AI and existed well before it.