gemini

The danger of AI isn’t what it knows, it’s how it decides what’s true…

Saw 60 Minutes with Google’s CEO Sundar Pichai on the current state of AI.

Scott Pelly, the interviewer, asked Sissie Hsiao, Google’s VP, what Google’s new chatbot, Gemini, does.

“It’s really here to help you brainstorm ideas to generate content like a speech, or a blog post, or an email,” she said with confidence, which made my skin crawl.

Let’s all stop exercising our neural connectivity to do tasks like writing an email, or a blog like this one, that requires disciplines in linear thinking, quantitative and qualitative reasoning, real news research, and engagement of my MIND to construct. Ms. Hsiao is suggesting that doing these tasks that demand and promote brain power are worthless wastes of our time.

So, Ms. Hsiao, where does that leave most of us who aren’t all paid the big bucks by Google to fuck up humanity even more than you already do? A hint, honey: STUPID. Do that research Sissie, and even Google’s AI will find recent data that humanity is getting dumber:

…etc.*

Next, Scott Pelley was “speechless” that Gemini made up a story from Hemingway’s six-word flash piece.** Mr. Pelley was so overwhelmed, he said, “Gemini appears to possess the sum of human knowledge.”

BULLSHIT.

There is NO INTELLIGENCE in Artificial Intelligence. The software NEVER, EVER THINKS. Gemini, ChatGPT, MS Copilot SCRAPE THE INTERNET to construct their responses to our queries. Every publicly available article, news feed, video, ad, post online are SCRAPED and stored in data centers like AWS (Amazon server farms).

Everything YOU do online, every Google Doc you create or read, to your social media posts, to what you see, watch, interact with, buy, to every email and text exchange you have are SCRAPED, stored and analyzed to “return better responses to queries.” How? Google, Microsoft, Verizon, Xfinity, T-Mobile…etc., all have access, and all sell, exchange, or share their data to improve their AI.

After SCRAPING the internet for data, these LLMs (large language models) WEIGH THE IMPORTANCE of the data they’ve collected according to your query. They analyze PATTERNS of behavior as it relates to the query, but also as it relates to all it knows about you, and it knows a LOT, as stated above. It simply follows its programming to return what you requested in a way you will likely hear it—i.e. siding with your existing opinion, or writing an email in your unique ‘voice.’ But AI DOES NOT THINK. It cannot distinguish between fact, propaganda, marketing, or popular opinion. It collects, stores, and analyses the data it finds, then regurgitates the LOUDEST SIGNALS that aligns with your query. The more a viewpoint dominates the data, the more likely the AI will present it as fact.

Which brings me to this pic I clipped from the 60 Minutes AI exposé:

Next, Scott Pelley asked Google’s chatbot to “Summarize the New Testament.” In one second, Gemini came back with: “The New Testament is the story of God’s love for humanity, which was revealed through Jesus Christ.”

BULLSHIT.

The actual Bible is filled with a jealous, angry, vengeful God (Corinthians 10:22; 2 Corinthians 11:2), who murders millions of people at his whim. Jesus is hateful to Jews (John 8:44) and others. He tells parables in which beatings, and even killings of household slaves are affirmed as ‘disciplinary measures’ (Luke 12:45–47). Revelations, the last chapter in the New Testament, tells of God, and Jesus inflicting the “punishment of eternal destruction,” (2 Thessalonians 1:5–10) on anyone who doesn’t agree with, or believe in them. Strip away blind faith and this is called TOTALITARIANISM.

My human interpretation of the New Testament is: “The New Testament is a collection of violent stories that center around two jealous, angry narcissists who inflicts hardships, loss, plagues, and other forms of gruesome violence on humans.”

So why did Google’s chatbot call the New Testament a “story of God’s love for humanity?” Google’s AI engine is trained on most ALL THE DATA ON THE INTERNET. This includes digital marketing — all those annoying ads — but also what people are saying, via texts, and posts, and blogs…etc., about any given subject. At least 80+%† of the U.S. identify as religious or spiritual. Christianity alone touts 64% of the U.S. as believers, and, by far, put out the most advertising. Christian marketing is close to a trillion-dollar-a-year industry.

Gemini, and all LLMs today are a combination of Machine Learning, Natural Language Processing, Deep Learning…etc, algorithms. That’s it. Garbage In, scraped from the internet and into the AI software, like Christian advertising, and people chatting up their spiritual beliefs to each other, led Gemini to spew Garbage Out — its positive, loving spin on the New Testament — even though the book itself, well… isn’t.

Since 99.999%…etc. of all that data Google’s collecting confirms both Bibles are good, righteous, and loving, Gemini LEARNS (not THINKS) that this is a FACT. AI of today WEIGHTS the importance of data by CONSENSUS, NOT TRUTH, or even FACTS. Truth by SOCIAL CONSENSUS, like Germans who became Nazis, and today’s hate on immigrants, Jews, and Muslims. If only 30% of the data Google collected on Christianity was positive, Gemini would likely not have come up with the nonsense it did. If Scott Pelley had bothered to question Gemini’s translation of the New Testament, or the CEO of Google had pointed out that their AI is a WEIGHTING SYSTEM that places more ‘value’ on the masses than the FACTS, perhaps those of you who’ve read this far will now understand how dangerous today’s AI really is.

Another question from Scott Pelley: “Is Gemini safe for society?”

Sundar Pichai: “I THINK so…”

—

*Human Idiocracy:

How many of you remember (or ever learned) phone numbers, now that you have them on speed dial. (Why does it matter? Try calling your kid in an emergency without your contacts list).
Who remembers, without the help of Google Maps, how to get to a place you’ve only been to once? Or even 20 times? For that matter, which of you even knows how to read a real map?
How many of you even know that the “news,” and information you’re getting through Google (or any) Search is only a fraction of what is on the internet, and worse, it is a reflection of how YOU think, delivered to you via recommendation engines that reinforce your own perspective? Essentially rec engines make you THINK you’re smart, but only make you dumber by serving up no other perspective than your own.

**Gemini’s AI construct from the words of Hemingway’s 6-word tale: “For sale. Baby shoes. Never worn,” missed the subtext of the words entirely. No one buys shoes for a child yet to be born, which Google’s AI story suggests. Hemingway’s story, (and the original he stole it from), is about the loss of a baby already born. Infants get booties. Babys get shoes. So, it’s likely the baby was born but never made it to the walking stage, neonatal deaths a rather common occurrence until the 1960s. Gemini missed all of this completely and made up a story virtually unrelated to Heminway’s 6-word tale.

† People who identify as nonreligious, but claim to be spiritual, are known as “NONES”.

Tag: gemini

How AI Works IRL