Greetings all, I hope you’re well as you are reading this post.
We are, again, at the last day of the month and I realized, earlier, I didn’t publish at least one blog to reach my goal. At least this serves as that achievement. The milestone I speak of may be more of a pre-milestone, but nonetheless I’ll count it as a regular one.
This is the 450th published blog since I started self-publishing via an official blog on my own domain in Dec 2008. The math tells me I’ve averaged two blog posts a month since I first started, not bad when I consider myself an average blogger. Pretty cool. I’ll do this type of post again when I reach 500 blog posts.
As for the first part of the blog title. I came across an article earlier this evening that was shared on Bluesky by someone I follow who is a postdoc neuroscience researcher. The title of the article is Learning via ChatGPT leads to shallower knowledge than using Google search, study finds. The title and topic is not new to me and, I suspect, the same for you. I’ve purposely felt that it bares some truth for sure, however, it’s just the latest, but same type of, discussion that has occurred during some technological advances in the past. What it could be, in the sense of ML and AI, is that it now seems accelerated but, in principal., not much different than before. Since the age of computing (in the broadest sense), it’s been happening – technological features that do things, automate things, for us that require less thinking, resulting in more cognitive ability offloading. This requires, most of the time, a level of trust and comfort that tells is it’s one less thing to have to think about. While there are many examples, I’ll use GPS for navigation. I’m old enough to remember having to rely on paper maps to get from point A to point B for routes I have no familiarity with. At some point, physical GPS units were commercially available to put in your vehicle, even if maps had to be periodically downloaded to them in order to provide the latest accuracy. Today, it’s on our consumer mobile devices and doesn’t require anything but download your app of choice and for us to trust what the GPS tells us for navigation – this is cognitive ability offloading at best, compared to having to use that Rand McNally (or similar) paper map.
As far as learning via ChatGPT vs using a web search, how different are they, really, from a basal standpoint. Off the top of my head, I would say that AI saves more time than doing a general web query, however, the more specific (with both) you are in regards to your level of “prompt engineering”, the better your results will be. We’re still at the point of ML/AI that AI still needs to be fact checked as a result of providing wrong/inaccurate answers at times because, well, look at what it’s trained on via webscraping alone.
I was having a discussion with a good friend/music composer/producer and large advocate of AI tools, just before starting this post. My belief at the end of the day is, multiple studies or not, how does this affect you as an individual? We all learn, comprehend, ascertain, differently, but at the same time, whether it’s via a web search or AI, how much fact checking, how much time and energy do you regularly invest to ensure that the information is correct? AI cannot fact check itself today. On the contrary, you could best believe that large strides were taken for information accuracy before that set of World Book or Encyclopedia Brittanica volumes for the year (fill in the blank) were purchased by your parents for the home library.
Things to ponder as technology evolves.
Happy Holidays to all of you.
Oceans of rhythm,
Fresh!


