I am not an expert in SEO or AI, so I’d love to read comments from such specialists here at DK — if I’m off-base anywhere or missed anything important in this diary.
I am however, a skilled web information researcher. I use web research in my profession and also when I’m writing my amateur political opinion pieces. My primary mission in performing such research is to find the most trustworthy, well-documented, and clearest information sources to rely on.
Often, I try to find original sources, such as (a) government publications (especially Code of Federal Regulations), (b) documents authored by SDOs (Standards Development Organizations) such as NFPA (National Fire Protection Association) and ASTM (American Society for Testing and Materials), (c) dependable news organizations such as Washington Post and Los Angeles Times, and (d) well-researched opinion outlets, such as The Atlantic, Slate, and The Hartmann Report.
When I execute a search, I’m well aware that the first two to six “hits” are likely to be “Sponsored” and I pretty much always ignore those links, or perhaps try to find an un-sponsored link to the same content below the first several hits.
Furthermore, when all the hits I receive are from “lower-quality” sources (e.g., sites that promote a service, or chat sites where the contributors may not be well-educated in the subject they are opining about), I re-frame the query to open up more (hopefully reliable) sources.
I am linking an article that appeared a few days ago in Slate, because I didn’t see anyone writing about it here at DK. It’s about the subtle yet monumental changes Google is making to their search engine. Here are some points the article makes:
Google Search … is becoming something else entirely: a self-ingesting singular webpage of its own, powered by the breadth of web information to which it once gave you access.
Gemini chatbot will spit out a general “AI Overview” answer at the top of your search results. These answers will be informed by (or even plagiarized from) the very links now crowded out by a chatbox.
On May 3 … [a SEO researcher published] how tweaks made from September onward had reduced the basic Google search visibility of several big-name information websites by up to 75 percent. [emphasis mine]
Reporters at the Washington Post have found that Google’s most recent search generative experiences are even worse now than they were last year, failing at even basic tasks like finding restaurants in your vicinity.
It appears that the best Google is doing is to make it harder to find stuff you’d trust.
Other bots like OpenAI’s ChatGPT and Meta’s Llama 2 keep “hallucinating”—that is, making stuff up based on haphazard, predictive interpretations of the data they are trained on.
I don’t know about you, but when I’m in a new city, and I type in “nearest gas station” or “nearest McDonalds”, I often get the first hit to be a store that is 25 miles away. (This always made me wonder if that particular store had bribed Google to have their search hits higher up in the results listing.)
And the last thing I need is for some hallucinating AI bot to dish out information that sounds authoritative, but is actually based on “deep learning” it performed on Quora or Reddit — without telling me that’s where it came from.
I wrote a DK diary about Chat GPT being a “pathological liar” about a year ago. I still believe that “hallucination” is a very kind (and inaccurate) term for the fabrications AI is capable of dishing out as fact.
The alternatives? Not too sure whom to trust here either. Here is a website that seems to give a neutral analysis of alternatives, but when you look closely, each of the suggestions has some/most of their roots in Google or Bing (Microsoft). Maybe I’ll try Brave, because it seems to be “based on” Google’s “chromium”, but not reliant upon it. It does, however, have an AI assistant onboard, so who knows whether you can turn that off or not.
I’m not as concerned about search privacy as I am about search accuracy. However, since Google has so effectively monetized all our searches, I think loss of accuracy is probably not far behind loss of privacy.
Why do so many businesses think lying to their customers is a good way to make money?
P.S. Does anyone know the underlying meaning of the Gemini artwork I posted at the top of this diary? It’s kind of cool, but also a bit ominous. It feels to me like Gemini is playing a “shell-game” with information. It takes in information, stirs it up a bit, and spits out something that looks the same, but probably isn’t. The artwork also reminds me of those old Chinese Finger Puzzle toys you could get in the 1970s.