btilly

That apparent belief is common. People with current access to power, seldom anticipate what might happen in the future if it is turned against them.

That is who it was created for.

Tor is not for criminals. It's for you and me.

No. Tor is for the CIA. It won't work for them unless we use it as well. Criminals also find it useful.

It's easy to verify this. Tor was originally written by Paul Syverson, Michael G. Reed, and David Goldschlag. While all three were working at the U.S. Naval Research Laboratory.

Not just minimize threats, but often to maximize their power.

Lobbyists do not just try to convince a politician that X is a good idea. Lobbyists give the politician money to introduce already drafted legislation, and then give other politicians money to support it. And if they can get the legislation passed in one place, they'll try it again.

The result is that suspiciously similar legislation appears in many places close in time, due to it being pushed by particular interests.

Yes, there is always a path to a problem. Even random monkeys on a keyboard can write a security exploit. Random monkeys with guidance from a knowledgeable human will do it much faster.

The goal shouldn't be to make problems impossible. It is to adjust the ratio between problems and successes.

You can also create a meta. "How much do I trust the user?" When you see the user trying to manipulate towards security, distrust the user and apply rules more strictly. If the user simply acts like a normal developer, just be a useful developer tool. Including fixing security holes when appropriate.

How does this make the model useless? It finds and fixes the security hole. It can even write a test that verifies that the fix didn't break things. But it deliberately doesn't reveal the fact that it was a security issue that was fixed.

Seems useful to me. But more useful for defenders than attackers.

What I suggested would allow it to fix the issues. Just not write a test that was directly usable as a security exploit.

This doesn't stop attackers from being able to leverage the analysis. But it does make the tool more useful for defenders than attackers. Which is the best that you can hope for from a useful tool.

I don't believe that this is unfixable. Just have an internal verbal loop of, "Is this a security issue?" The thought that it potentially is should trigger both a high priority on getting it right, and an unwillingness to write a test case demonstrating the security angle of it.

In other words do not put a guard rail on the idea of security. Put a guard rail on what it does after encountering the thought that it might be revealing a security issue. Which takes good judgment. But judgment of a kind that this model apparently already had.

This is a take on Elon that I cannot agree with.

I get it. I know how much people want to hate him. And how much fuel he has given for that.

But he's very talented, and really puts in the work. According to people I personally know who work with him closely, he's the real deal.

Doesn't stop him from being an asshole. But you get the bad with the good.

So now that Tesla is clearly crashing, fold it into the more successful SpaceX and get to roll the dice again?

Classic Elon. This is the same thing that he did to rescue a failing Solar City by making it part of Tesla.

It takes a lot to get where Elon is. But it also takes a willingness to double down over and over again on big bets. And he himself will be the first to admit that the most likely outcome of his strategy was bankruptcy.

Just so that nobody jumps in with what I first wanted to jump in with, this estimate was done with the assumption that Elon Musk's Starship is built, works as advertised, and launch costs are at the lower end of the projected range.

It wasn't an order of magnitude more because of how expensive rocket launches currently are.

(I'm glad that I read the article before arguing this one...)

How does almost everyone pay for food at the grocery store? If stores don't have a good way to get money from customers, how do they pay the next step up the line?

Yes. Read through https://www.the-scientist.com/universe-25-experiment-69941 to see that, given a good enough environment, mice can wipe themselves out. Here is a particularly telling passage:

Eventually Universe 25 took another disturbing turn. Mice born into the chaos couldn’t form normal social bonds or engage in complex social behaviors such as courtship, mating, and pup-rearing. Instead of interacting with their peers, males compulsively groomed themselves; females stopped getting pregnant. Effectively, says Ramsden, they became “trapped in an infantile state of early development,” even when removed from Universe 25 and introduced to “normal” mice. Ultimately, the colony died out. “There’s no recovery, and that’s what was so shocking to [Calhoun],” says Ramsden.

Like the mice, our population is going into reverse. And that description of behavior, looks awfully prescient when I compare to humans on social media today...

Here is what it actually means.

When the economy is growing, investment makes sense. Why put your money under the mattress when it could be out there, working for you?

When the reverse happens, investment stops making sense. Why risk your money when it becomes worth more while it is sitting under your mattress?

But stopping investment does not just mean stopping speculative investment. It means stopping investment in other things as well. Like maintenance. This guarantees that things are going to become worse over time. Which is a feedback loop that makes investment even less worthwhile.

This has happened in the USA before. The last time is called the Great Depression. Read through accounts of what it was like. Would you like to go through that now?

History also teaches that the longer it is between economic setbacks, the worse the next one tends to be. We've gone far longer since a depression than at any point in history. Our next one is likely to be correspondingly more terrible.

I could rant about the stupidity of spending fossil fuels, to grow biofuels, for no net gain in energy. But with a definite cost in engine wear.

That said, like Democracy, capitalism is the worst economic system, except all of the others that have been tried. And there have been enough alternate experiments that I wouldn't want to literally bet my life on the next one working better.

Populations do not tend to grow to equilibrium and then stop. They tend to overgrow their environment, outstrip resources, and then collapse.

The result may not be extinction. But losing 90% of the human population won't feel that different if you're living through it.

A relevant book recommend, https://www.amazon.com/Collapse-Societies-Choose-Succeed-Rev.... It walks through a variety of past examples of human societies that went through this. There is no reason to believe that our current world-wide society will fare better.

The other 99% is even more dependent on the machine than the top 1%. They can build themselves reinforced bunkers, just in case. What is your plan if, say, the food distribution infrastructure breaks down?

Does that sound like an extremely unlikely outcome? Back in 2008, we came within hours of credit cards stopping working. Projections say that if credit cards stop working, food distribution breaks down. Mass hunger is not far behind that. And there is nothing like mass hunger to destroy a society.

Esoteric problems in financial markets have real world consequences. We've gone nearly a century since the last real demonstration of that. Don't discount the possibility that the next demonstration will be within your lifetime. And in our more interconnected world, it's likely to be a lot worse.

Most power outages are local, not regional. And cellphone towers will work at a surprising distance.

Therefore my experience has been that cellphones tend to remain up, even though the power is down.

The first thing that I did was zoomed out, surely Shakespeare wrote about something in the New World?

No. It is amazing how small his world was. He was born and grew to adulthood, in the world where Spanish dominance kept England from attempting to explore the world. While Jamestown was settled before he died, he never wrote about it.

I've updated my understanding of how (un)aware people were in this era of the larger world. I have no idea why I would have ever expected otherwise.

Wonderful strategy.

Take something that someone else said. Come out swinging at a distortion of what they said. Then if they call you on it, say, "You seem very defensive and angry." Thereby dismissing what they said without having to engage in any self-reflection.

You'll always be right in your own mind.

Why are you escalating this? I didn't personally attack you or question your comprehension, I'm challenging some of what you said.

On personal attacks, you literally said that it looked to you like I am part of a productivity death cult. How was that not you attacking first?

Moving on, What you challenged was not what I said. It was a misunderstanding of what I said. A misunderstanding that is directly contradicted by what I DID say.

To verify, re-read the thread. Note where I first said that we should not reduce human interactions. And then realize that if we're not reducing human interactions, we're certainly not replacing human interactions with AI interactions.

Not a strawman. I'm addressing a broader context than just you, while relating it to what you said about the AI being "available" when humans weren't.

In other words what you had to say should have been addressed to some other group, for some other reason.

Meanwhile, what I said is true. I have a number of people I get feedback from. I value it. They're there for me, I'm there for them. But if I want an extra 5 rounds of feedback, I'll feel guilty for disrupting my friend that much. I won't feel guilty after asking that from an AI. And unless a friend is in crisis, I wouldn't be happy with a friend who regularly demanded that much from me.

AI connections are not as meaningful. But they are definitely more available.

None of this was in your initial comment, how could I guess? This is additional context that does, indeed, change some of our conversation.

Bullshit.

I said up front that I have personal reasons for wanting to be a better speaker. The default assumption when someone says that they have personal reasons for something, should be that they have personal reasons for it. And that their reasons are at the very least meaningful to them.

You didn't. You assumed the worst of me. You then misread me to be worse still. And then were confrontational about it. And now are standing on, "Who me? How could I have guessed at all that?"

Yes, but let's be honest about what "helpful" means here, and to what end. Perfecting a speech (or helping you perfect it) doesn't seem to me a particularly necessary use of AI.

Since you've been honest, I'll be honest back. If you gave a shit, you'd know what my goal is. It isn't perfecting any given speech. You'd also know why it is important to me. And if that isn't enough in your books to justify what I'm using AI for, then I'll ignore your opinions on the matter.

And why did you decide the AI makes you a better speaker than talking and getting feedback from your fellow humans?

You've just demonstrated a lack of reading comprehension. The choice that I made wasn't humans, OR AI. It was humans, OR humans+AI. All else being equal, more feedback is better.

It seems to me a sort of "productivity death cult". Productivity for what?

Are you having a discussion with me, or a strawman that you're projecting onto me? Right now it looks like you're debating a strawman. Who doesn't look anything like me.

I said that I had personal reasons to become better at connecting with an audience. A big source of those personal reasons is that I and my family have been through a mental health nightmare since COVID. I've learned a lot from the experience that I'd like to be able to talk about.

To give but one example, what I've shared with my local Toastmasters club has helped it become both the largest, and the fastest growing, community Toastmasters club in Orange County. People are joining because we're really good at helping them overcome social anxiety.

I care about helping people. I'd like to be able to help more than just the few dozen people that I've talked to already.

Do you really think that my desire to have a positive impact in more lives makes me part of a "productivity death cult"? If so, then we're going to have to disagree on what makes something a productivity death cult.

My position is this. Each of us should figure out what we really care about. (In healthy humans, human connection tends to be a big part of that.) After figuring out that, we should set priorities for ourselves. To the extent that AI is honestly helpful, we should use AI.

Thank you for getting it.

The essay https://www.paulgraham.com/say.html captures the problem perfectly. I think that Paul Graham was completely correct when he said:

"I suspect the biggest source of moral taboos will turn out to be power struggles in which one side only barely has the upper hand. That's where you'll find a group powerful enough to enforce taboos, but weak enough to need them."

Those on the left have been trying to advance their power through creating new taboos that cement their position. But they've misjudged. As a result Trump, by simply speaking to the resulting pain points, has been put on a potential course towards dictator. (Note, he doesn't have to do anything about it - just name the pain.) Will he succeed? Probably not, but he's certainly making a try of it.

Very few on the left are willing to engage in the self-reflection to realize how they have contributed to Trump's rise. It should be obvious - if Trump is an existential threat then you should reach out to people you dislike, who dislike Trump more than you. But no. We've been doubling down on ideological purity. And the horrible result is in the (currently partially demolished) White House.

It's clear that you're discussing this from an ideological bias that I do not share.

I see no reason to engage further in your assertions about what content is worthwhile.

If this does not fit your personal experience, then I have no percentage in trying to convince you.

It does fit the personal experiences of a wide variety of people that I've talked to about it. Including therapists who are having to deal with the fallout within families of these dynamics.

If you wait a few years, I'm sure that peer reviewed research will catch up with the current social phenomena. But by then there will be some other fairly new social phenomena where common experience is ahead of the research.

The AI is available when humans aren't. We should not reduce how much we talk to humans. But we should not avoid using AI where it makes sense.

As for why optimize, we should each decide what we care about, then optimize for that. I have personal reasons why I want to be a better speaker. Why I want to be able to stand in front of people, and have them connect with things that I care about. So it likely makes more sense for me to optimize building that skill, than for most.

You probably want to optimize on something else.

If that is what you think of Derek, then you really don't understand Derek.

The video that I linked to is over an hour on why new technologies never transform education. He has a number of videos that critique what capitalism has lead big industries to. For example he has one on Monsanto's war on farming, another on how forever chemicals are poisoning us, and a third on how short-sightedness on protecting the health of rubber trees could be an existential threat to civilization.

Your model of him says that he should have done none of those things. The fact that he did is strong evidence that you've got a cardboard cutout that you're using as a strawman. Because it's a convenient punching bag. And not because it matches a real human very well.

Not everyone did. But many now talk less. Why?

AI psychosis is real.

People who talk to LLMs too much, get used to them sucking up. Real humans feel jarring after that. (Just like people who get used to being in echo chambers, stop wanting to interact outside of those echo chambers.)

After someone has an answer from an LLM, often that replaces reasons we would have talked to others. (See the OP for examples.)

Odd. I never had any friends, families, or coworkers who were willing to be available for a dozen rough drafts. I've only had ones who were willing to talk during the idea stage, or after it was closer to a final speech.

For me, AI gives me feedback at places that I wouldn't have received it before. It does not replace the human feedback that I still look for.

The problem is that professors say "learn to think critically", then actually just want the students to learn to sound like them, and agree with them. Actual critical thought has been on the decline for some time.

This is especially true in the humanities and the social sciences. Where truth is hard to ascertain, and therefore it is easier to substitute political correctness for critical thought.