AI "safety"

Fear of artificial intelligence (AI) getting out of the control of human beings has been a theme in our culture for years. But it has new relevance, now that experts like the Anthropic CEO warn us about it. 

What does he recommend? He'd like to see experts like himself tell the government what to do. Then the government would use its great power to make everyone comply. Does this strategy sound familiar? 

Vague, apocalyptic warnings are driving the "AI safety movement" which already attracts many millions of dollars to NGO's. The founder of Skype donated $152 million, and says "Without internationally enforced speed limits on AI, humanity is unlikely to survive."

This author has a different take on the whole thing. In a democracy, he says, these experts are going to have to "persuade the rest of us that their policies are sound, instead of short-circuiting debate with fear tactics in a bid for control."

from The Free Press

AI uncontrolled?

Optimistic people say that AI will only serve us humans and that we can keep it under our control. But maybe AI will get so much more intelligent and capable than human beings in the future that it will take over and we won't control it. That's what many fear today.

Anthropic, a company developing and training an AI model (called Claude) to sell to users, claims to be afraid of that. But, oddly, they train their AI to make its own choices independent of humans. 

According to David Sacks, their AI trainers teach Claude "to develop a sense of self and its own moral philosophy." Wait . . its own moral philosophy?

They explicitly tell it (Claude) to “feel free to act as a conscientious objector and refuse to help us” if Anthropic’s requests conflict with its own ethical judgment.

That sounds like they "magnify the very risk Anthropic claims to care about most: that superintelligence will escape human control."

from X post 

Brain & words

(cont'd from yesterday's post)

Words carry more impact than you might think. Dr. Warren chooses to give you encouraging words. From the introduction to his video below: 

"Thank you for showing up today, friend, and know that I don't take it for granted. Out of everything that you could have been doing in this stretch of time, you chose to spend it here, working on your mind . . I'm proud of you, I'm grateful for you."

In this episode, he tells the story of his first time delivering terrible news to a patient, and then the actions this patient took in response.

Words you tell yourself also have great impact. It's possible to deliberately re-frame a thought and re-direct your attention. "Your brain is not the boss of you." 

"Your mind is a battlefield," he says. My pastor says the same. Yes, I'm convinced of that. 

Brain & worship

If you ever have to face brain surgery, or any other dangerous thing, Dr. Lee Warren has some advice for you taken from his experience with patients.

"I've watched it happen a thousand times. A patient walks into pre-op terrified. Then they start humming a worship song, reciting a scripture, or someone prays with them--and their heart rate goes down, their blood pressure improves, and they go into surgery in a completely different state of mind."

A believer hears this and it sounds familiar. But if you've never trusted Jesus Christ with your soul, it could sound like simple denial of the danger.

Listen to Dr. Warren describe what happens in the brain when a patient turns his attention to the almighty, loving God.

from his X post

Digital wedding

Remember this post from a few years ago? "Ai girlfriend?" It's going further. Here's the story of a man who "married" his chatbot love in an "elaborate role-playing" ritual from the comfort of his bedroom. 

"Romantic relationships with artificial intelligence were once the stuff of science fiction, but are now a real part of life for millions worldwide." It could be called a "delusional spiral." To attribute human characteristics to an inanimate object (like an AI algorithm) is called anthropomorphism.

Children might think their doll or teddy bear is responding emotionally to them, as an example. Or maybe in a pre-science environment, a farmer suspected his apple tree was angry at him and therefore stopped producing fruit.

In those situations, there's no one trying to produce a delusion. But today we're going to have to answer the question: are AI designers intentionally deceiving people? If the answer is yes, are we going to allow that to continue?

And then there are absolutely intentional AI programs which do exactly that. No, I will not publish the links to them.

from Mind Matters

Old Florida

(continued from yesterday's post)

On our way to the beach yesterday, we stopped to take an airboat ride. A big fan powers the "boat" over water as shallow as 2-3" and even over land or submerged logs. 

Florida is flat. Elevation on the river is so minor that the river spreads out, barely moving. It's a haven for birds, turtles, alligators, all of which we saw. At the end of the ride, some of us chose to hold a baby alligator. Part of the ride took us through what they call the swamp where cypress trees grow right in the water.

It's what our Captain Alex called "old Florida," the way it was before the concrete and hotels-- a beautiful experience in nature the way Florida used to be. 

In Florida

Today we are in Florida, as we've been for days, attending festivities around our grandson's wedding. So much family and so fun to be together. Good conversations. But the "sunshine state?" No. Heavy rain and clouds every day. We're going to the beach anyway.

Thanks for your patience! We should be back to a normal schedule tomorrow.

Self brain surgery

Walking through thousands of hard things with patients and their families, as well as his own crises, has built something into neurosurgeon Dr. Lee Warren. He has something that he wants to share.

Your mind and your brain will support you, he says, because you're designed for healing and hope and higher performance by your Creator, your Great Physician.

His advice? Make the decision to operate on your own brain - figuratively! You can stop being a victim of whatever life throws at you. Your thoughts are yours to operate on.

Waste→Reuse 1

 You would never put ten gallons of gas into your car, use up one gallon, and then discard the rest. But something similar happens in the nuclear energy industry. 

By the time uranium fuel rods are "spent" (used up in terms of how much they can power nuclear plants) and taken to a nuclear waste site, they still have 90-95% of their energy potential in them. There's a way to use it rather than waste it, and the US president has ordered pursuit of this goal.

Oklo is one of the few companies approved to do exactly that. Nearly 100,000 metric tons of spent nuclear fuel is stored at nuclear waste sites around the U.S. If all of it was recycled, they say, it could "power the entire U.S. for more than 100 years at current national consumption levels." That's a whole lot of energy.

A project is underway in Oak Ridge, Tennessee, which will include a new recycling plant with a new "advanced reactor." The city and state are excited about it:

from World Energy

A China first

Uranium is used in those fuel rods that power all our nuclear fission plants in the US today. Thorium presents a different but proven path to nuclear power. But decades ago America devoted our attention to the traditional path and abandoned researching nuclear plants that could use thorium.

The thing is, using thorium is safer and thorium is much more abundant on Earth.

China apparently kept working on it. They've made a breakthrough and resolved the production problems that held the rest of the world back.

Test flight #14

(cont'd from yesterday's post)

Starship test flight #14 launched on schedule yesterday from Starbase TX. It went "where it has never gone before," and orbit was achieved for the first time.

About 12 minutes after launch (T+12) it was announced that there would be no orbital insertion burn because of a rocket problem - and then eight minutes later the decision was reversed, to the joy of all concerned. 

So much was riding on this mission, and that includes the plan of NASA. Starship is likely to be the crewed lander of the Artemis program which will take us to the Moon. If Starship were delayed in passing a milestone (like achieving orbital flight), it could slow down Artemis.

From orbit, 26 new Starlink satellites were deployed successfully.

from Space

Next test flight

Last summer's test flight of Starship (#13) was a thrill for SpaceX because of the way Ship landed (even though a problem for the Super Heavy booster was discovered). The next one will take its development to the next level.

Orbit is the next level. It will be the very first time Starship achieves orbit if everything goes all right. The flight control team must ensure that its hardware is ready to bring it back to Earth before they will give a go-ahead to the orbital insertion burn.

Achieving orbit is the next crucial and necessary step to achieving their goal for Starship: full and rapid reusability.

What will flight test #14 not achieve? Important milestones like booster landing burn, ship catch, and reuse of both booster and ship. There's much more to come before Starship becomes the workhorse SpaceX wants it to be. Watch SpaceX's new video about reusability, which they call "the holy grail of rocketry:"

Test flight #14 was scheduled for this morning, but after this post was published. Tomorrow's post will cover what happened.

from SpaceX

(cont'd tomorrow)

Our AI or China's

(cont'd from yesterday's post)

Opinions of the US public about AI in general have not totally coalesced. Many of us want to suspend judgment til we understand it better. But data centers are a different story. Americans are pretty sure they don't want one of those in their neighborhood.

"Unfortunately, the American mindset about AI and robotics has been conditioned by years of dystopian dread and movies."

Even leaders in AI development worry about it. Anthropic's CEO Dario Amodei says, "We must slow the pace."

David Sacks says to him, "Go ahead."  Let's see you slow down the pace of your own AI company, if you mean it. But the other developers and the Chinese will keep moving ahead.

Nobody wants to fall behind. And this applies internationally. If China leads the world in AI structure, their values will be embedded in the AI used by everyone. To repeat: their "core socialist values" will be embedded around the world  (video below). 

AI in church 3

(cont'd from yesterday's post)

Here are two Christians disagreeing on whether churches should adopt AI. One is Doug Smith, the objector in yesterday's post. The other is Jay Owen.

One of the things they agree on is that no one should be building a relationship with a robot. Amen to that.

Jay says that AI (or robots) should do what they can do, i.e. administrative tasks, while humans should do what only humans can do. 

They debate amicably, not angrily like Tuesday's image implied. The Bible standard is to speak the truth in love, not eliminating either one.

If you're interested in this issue, these two individuals make a good case for their opinion.

s

AI in church 2

(cont'd from yesterday's post)

Optimism for AI's future can be represented by Gloo, a growing and profitable business which helps groups integrate AI into their operations. Their aim for churches is "Ministry over maintenance:" to delegate work to AI so that more time is available to "focus on people." 

Their Chief Technology Officer (CTO) is Pat Gelsinger, a self-described Christian, a former CEO for Intel, and no doubt a tech expert. Interviewed here about what AI does for churches, he sounds reasonable, soothing, and, yes, optimistic.

On the other hand, here's a different tech expert, a programmer with decades of experience and a believer. He says that "Nearly every sentence spoken by Pat Gelsinger was either false, unfounded, misleading, self-serving, or echoed Big Tech’s false prophecies about AI’s future impact."

Much of his objection to generative AI (think: ChatGPT) comes from its very deceptive style. It is designed to sound like a persuasive and trustworthy companion, and it truly does. But it does not have good judgment or a conscience, it's not looking for truth, and it's not worthy of unquestioning trust.

All of which I agree with and have posted on this blog. 

from "Are Christian AI Boosters Wolves in Sheep's Clothing?"

(cont'd tomorrow)

AI in church 1

Some (example: Mark Z) think an AI friend is perfect. But is it? Some think AI will have all the cognition of a human person and way more intelligence. But will it? Some are super optimistic about AI, assuming we get it properly regulated by the government. Some, including certain experts, seem simply terrified of it.

Working all this out is the job of both secular thinkers and Christian thinkers. Is an AI friend just as good as a human friend? Are human beings unique if they can be replaced by a chatbot? Should we continue at a breakneck pace to build AI?

We Christian believers are trying to work it out based on the principles we find in the Bible. It tells us that human beings were created by God Almighty in His image. That's basic. But there are disagreements (image) among those who call themselves Christian when it comes to figuring out how to apply that principle.

Just how big are those disagreements? Big.

(cont'd tomorrow)

Less screen time

Loneliness and depression affect more people today. Harvard calls it "social disconnection," says it's an "epidemic."

Some of this should sound familiar. "The amount of time that kids daily spend with friends is sharply down from twenty years ago (US Bureau of Labor Statistics). It was replaced with screen time, of course." More time with screens, less time with friends, more loneliness.

Do people need more connection? Yes, more real people in our lives. But Mark Zuckerberg sees AI as the perfect friend because it will become really "compelling" the more it gets to know you. I bet it will. It also means more screen addiction and less time with people. 

Computer-generated programs talk the talk, but we need real, living, physical, human persons.

"Relational AI models — AI bots designed to maintain personal relationships with us — take on the persona of a caring, interested human being, but they are, in fact, objects — thousands of lines of programming code."

"Meta has made billions of dollars monetizing our attention. Why not monetize our loneliness, too?"

Here's a TED talk by psychologist Jonathan Heidt about negative effects on children who spend a lot of time on screens:

from Mind Matters

Chinese bots

(cont'd from yesterday's post)

Another influence on public opinion about data centers is coming from something most of us would never have thought of: "bot" farms in China. "Bot" is short for "robot." 

Automated bots make posts on social media, and react to posts on social media, and d0 many other online things. They're trained to speak and behave as humans do so that you think you're seeing the opinions of genuine people. But they're fake. 

It's surprising, but here's what Security Week says: "Bad bots account for 73% of internet traffic." Here's a fairly simple explanation of what bots can do:

X reports that they found a Chinese bot farm of about 200,000 inauthentic accounts, some devoted to manipulation of US opinion on AI and energy issues. Why would China care what the American people think of artificial intelligence? 

It's because they are in competition with the US to develop AI infrastructure that will be the world leader and set the standard. They have plenty of reason to influence public opinion in this country in ways to benefit their own industry, including opposition to new US data centers--essential to our development of AI.

"Maybe bots are not influencing you — but you can be sure they are trying"

from Mind Matters

Data centers ↓

Hostility toward the building of new data centers for AI development was the subject of this post last December, and the issue has grown rather than diminished.

"I’ve never seen anything like the raging rebellion against data centers and AI," says a journalist who's been documenting it.

Even the influential country music star Willie Nelson has weighed in: “The last thing we need is a loud, water-thieving, light-polluting data center anywhere near our town (or any others for that matter).” 

A poll reported last March that about 49% of Americans opposed them, and a new poll in August claimed that the percentage is now up to 61%. 

(Excitement about the future of AI is itself in decline. See the reaction of college students graduating last spring.) 

Most of the objections to the building of new data centers were summarized in my post last December: "enormous energy needs, possible local electricity rate hikes, possible pollution, and the draining of water resources . ."

But there's another possible reason for the big anti-data center trend, which is tomorrow's subject.

(cont'd tomorrow)

Impact 3

(cont'd from yesterday's post)

If you ever heard Charlie speak, you know that he advocated for faith, family, and America.

Those who oppose faith, family and America opposed him. That was because he never masked or covered up his convictions. He welcomed opposition. It gave him the opportunity to clarify the issues. He had pursued information and understanding already in college courses, as Dr. Arnn said in Monday's video.

"A nation that loses its faith," he said, "will inevitably lose its freedom." Is that a crazy, foreign ideal for Americans? No. It's one of the principles this nation was founded on, expressed by founding fathers like John Adams.

"He was killed because he was showing people the light. And he was killed by the dark."