Google's "Generative AI" lies about junk DNA

Every now and then I check Google to see if there's any news about junk DNA. I use "junk DNA" as my search query.

The first thing I see at the top of the results page is a summary of the topic created by Google's Generative AI, which it claims is experimental. The AI summary is different every time you start a new search but all of the responses are similar in that they criticize the idea of junk DNA. Here's an example from today,

"Junk DNA" refers to non-coding DNA sequences in the genome that were once thought to have no function, but are now recognized to play important roles in regulating gene expression and other cellular processes.

Here's a more detailed explanation:

Historical Context:

The term "junk DNA" emerged from the idea that most of the genome, beyond the protein-coding regions (genes), was simply "filler" or "waste".

Non-coding DNA:

This "junk DNA" is the vast majority of the genome, comprising about 98% of the human genome.

Functions Beyond Coding:

Research has revealed that this non-coding DNA is vital for various cellular functions, including:

• Gene Regulation: "Junk DNA" can act as switches, controlling which genes are turned on or off, and at what level.
• Structural Roles: Some non-coding DNA contributes to the structure and organization of chromosomes.
• Evolutionary Importance: "Junk DNA" can include remnants of past viral infections, transposons (jumping genes), and other elements that can influence evolution.
• Disease Involvement: Variations or mutations in "junk" DNA have been linked to various diseases, including cancer, autism, and schizophrenia.

This description is wrong on so many levels, but here's the important point: it conflicts with the Wikipedia article on junk DNA and it conflicts with some of the scientific and popular literature. The true value of AI would be to sort through all the trash on the internet and return accurate, factually-correct, information in spite of the fact that the majority of stuff it finds is garbage. Modern AI algorithms can't do that. They are not the least bit "intelligent"; instead, they simply collect and summarize all the data they find without regard to whether it's true or not and whether it agrees with the scientific consensus. Then they present it using a language that sounds human-like.¹

The first sign of improvement will be when AI algorithms begin to recognize that there are multiple points of view and it needs to summarize both sides. The next step will take place when AI is able to sort the wheat from the chaff. Future AI algorithms will have to determine which sources are reliable and which ones aren't. We're not even close to being there, yet.

Meanwhile, AI is doing more harm than good by propagating misinformation as fact. This is a problem.

---

Image credit: This image of a garbage truck carrying junk DNA with DNA molecules was generated by AI.

1. The general public has been tricked into thinking that an algorithm must be intelligent just because it can mimic human speech and writing. Anyone who watches FOX news should know that's not true.