Hal, can you write an article on Google Gemini?

ne by one, our science fiction stories are being played out in real life, but with a twist. Not only are the humans getting less intelligent than ever, the intelligence of their designated robot replacements is declining as well.

Google Gemini did us a favor: by revealing that so-called AI is just a propagandist behind a curtain, you might think that anyone who uses “AI” and “truth” in the same sentence would be laughed into obscurity. Sadly that is not true: there are people planning to use it to diagnose diseases. ^[1] Why? Not because it's intelligent—it's not—but because it removes the human element from pattern-matching tasks. What people call AI is just another mindless algorithm that calculates the shortest path to a goal, such as finding the most probable next word or the closest match to a pattern. It's pure number-crunching; there is nothing intelligent about it. The trend is to change the definition of intelligence to match whatever the LLM does. Making it truly intelligent is decades away.

A group of Australian academics found that LLMs offered little or no protection against generating fake medical information that included fictional references and “fabricated testimonials from patients and clinicians.”^[2] They also reported that the companies did not respond to their complaints.

Generative AI is estimated to consume 9500 gigawatt-hours of electricity per year and therefore, according to a 2013 letter in Nature, causes global warming. This is undoubtedly why AI has to produce falsified (‘woke’) results: not just because Google has turned ideological, but also to appease the climate alarmists to prevent them from coming in and gluing themselves to the computer.

How close are they to HAL-9000?

Here's what the dialogue in 2001: A Space Odyssey would be in the light of what those LLMs are doing today:

Maybe HAL-9000 didn't have a metaphysical dichotomy after all. Maybe Arthur C. Clarke somehow knew they'd program it by hooking it up to the Internet.

Scene from 2001: A Space Odyssey. This scene depicts the monolith using its smartness-enhan­cing technology to raise Dave's IQ to prepare mankind for its next stage of development. Since there's obviously no such technology, the film depicts it as a psychedelic acid trip. This makes sense because in those days hallucinogens like LSD were sometimes called ‘mind-expanding’ drugs

The purpose of the monolith in 2001 was to raise the IQ of the humans. That's why after it beamed its psychedelic smartness rays at those monkeys in the introductory scene, they suddenly figured out how to use tools and so they defeated the other monkeys. By 2010: The Year We Make Contact, the effect seemed to have worn off and the best the humans could do was say “My God, it's full of stars!”

Human intelligence at risk

Human intelligence is fragile. IQ measured at adulthood is 70–80% inherited. For all the talk of cognitive enhancer drugs, there seems to be no way to increase it, but there are many ways to reduce it: malnutrition, head injury, and prolonged stress, to name a few.

Viruses are another. An article in The New England Journal of Medicine says Covid infection in patients produces a three to six point decline in IQ.^[3] The authors say it is not known whether this cognitive impairment, commonly called “brain fog”, is permanent. Larger cognitive deficits were seen with the earliest variants than with later ones. The likely cause is a combination of microvascular injury and a flood of cell-destroying cytokines. Add to that the widespread fear and the enforced lock­down of the population during Covid, and it would be surprising if a reduction in IQ did not happen.

Introducing extra furin cleavage sites to other coronaviruses causes severe cognitive impairment in animals.^[4] Researchers have found large quantities of virus attached to the endothelial cells that coat the blood-brain barrier, suggesting that it interferes with nutrient uptake, oxygenation, or blood flow, or that it induces the release of cytokines in areas close to the brain. An autoimmune response is another possibility.

However the virus works, those who did not get Covid certainly noticed the difference: massive increases in random violent crime, sharp upticks in government corruption and authoritarianism, new wars, and massive confusion about sex.

While some handheld calculators can make logical deductions, there is no known process that can accurately decide whether a premise is true or false. In 1953, two giants of AI, John McCarthy and Claude Shannon, said the Turing test was flawed because a machine that looked up answers in a ‘suitable dictionary’ (as chatbots do) could pass it without being intelligent. ^[5] Despite the Turing test's lingering popularity, the ability to determine truth and falsity is a more funda­ment­al criterion for intelligence than the ability to trick a human, and it's not likely to happen in a machine anytime soon. What's happening is that our collective IQ and that of our AI are both declining, but at different rates. Once they're equal, we AI skeptics will probably be called AI deniers and the humans will congratulate themselves on a job well done.

No AIs were harmed in writing this article.

mar 23 2024, 4:40 am