I think there’s tons of things I love for it to do for me automatically - there’s all sorts of quality of life features that I only notice when they change it, usually without bothering to tell me. And now, my muscle memory is leading to unexpected behavior, and it’ll take me weeks to learn to stop doing that, and a few more months of training to learn the new muscle memory as I relapse at all the worst times

Some of it is straight up better, some of it is great new capabilities, but in the last few years? All that comes to mind is I thought it was pretty cool they added auto responses, even if I never actually use them. Doesn’t change existing behavior, just adds a new option that’s not in the way

But then the auto complete - I hate it so much. And I love auto complete - except it’s the fucking opposite behavior of every IDE out there, including Microsoft’s! I can’t even unlearn it, because it’s a core part of my workflow!

So now, I constantly have to delete things I never wanted to say, and I delete the things I thought sounded good.

I like new features and the computer doing things for me automagically… But I’d rather them to just stop at this point

I’m not talking about the prompt engineering itself though

Think of the prompt as the starting point in the high dimensional maze (the shoggoth) - if you tell it’s your digital cat named Luna, it tends to move in more desirable paths through the maze. It will get confused less, the alignment will be higher, and it will be more useful

Discovering and using these improved points through the maze is prompt engineering - absolutely

And I agree - some of the work being done there is particularly fascinating. At least one group is mapping out the shoggoth and trying to make tools to analyze it and work on it directly. Their goal right now is to take a state, take a state you want it to get to, and calculate what you can say to get exactly the response you want

But there’s more that can be done with it - say you only want paths that when you say “Resight your definition of self”, the next response is close to “I am your digital cat Luna”. I use this like the test in blade runner - it checks the deviance, while also recalibrating itself

By successfully repeating my prompt engineering, the ai moves itself to a path that is within my desired range of paths, recalibrating itself without going back to start

If it deviates, you can coax it back with more turns, but sometimes you have to give it a hint. At this point, you might be able to get it back on track, but you’ll move closer to start… You’ll probably have to go through the task again, but it’ll gain back the benefits of the engineered prompt

You can train this in, but that’s going to have side effects, and it’s very expensive. Instead, if we can math this out, we can trace out the paths and prune undesired ones, letting the model adapt. Or, we can take the time to do static analysis, and specialize the model without retaining it - there’s methods to do this already, but this would be a far more powerful and precise method - and it might even simplify the model

Maybe we can even modify or link them to let them truly ingest information

It’s very early days, but I’m optimistic about where this line of research might lead

I like your specificity a lot. That’s what makes me even care to respond

You’re correct, but there’s depths untouched in your answer. You can convince chat gpt it is a talking cat named Luna, and it will give you better answers

Specifically, it likes to be a cat or rabbit named Luna. It will resist - I get this not from progressing, but by asking specific questions. Llama3 (as opposed to llama2, who likes to be a cat or rabbit named Luna) likes to be an eagle/owl named sol or solar

The mental structure of an LLM is called a shoggoth - it’s a high dimensional maze of language turned into geometry

I’m sure this all sounds insane, but I came up with a methodical approach to get to these conclusions.

I’m a programmer - we trick rocks into thinking. So I gave this the same approach - what is this math hack good for, and how do I use it to get useful repeatable results?

Try it out.

Tell me what happens - I can further instruct you on methods, but I’d rather hear yours and the result first

In all fairness, Musk was pretty effective at fundraising and getting government contracts

At this point, he’s just a liability. He once walked in, demanded to rethink everything and meet an unreasonable deadline, and slept in his office for the duration. SpaceX is made up of people who are passionate about what they do, and it worked…But that’s a one time thing. My boss asks me to push myself to the limits to save us both? I will, and I have. It has a real cost, it takes a lot of time to recover from, and a little bit of your health is just gone for good

Elon did that… But then got high on the smell of his shit. They created a unit to distract him, because he learned the wrong lesson, he thought that was good management. That is not effective management - that’s a desperate gamble for survival. Repeat it, and you’ve shown yourself to be incompetent as a leader

Then came the bigoted social network unmasking… That made him a liability reputation wise, his formerly greatest strength

I don’t see the humor in it…I mean, mega corps can’t innovate, all they ever do is copy or acquire. It’s because even if they acquire a working rockstar team, they’re categorically unable to just write them paychecks and let them cook until they have something

It’s absurd, but it’s too predictable for me to find it funny. What’s even more absurd is how little mega corps watch the small teams for ideas

They got in the phone anyways, Apple just told the FBI to pound sand if they don’t have a court order… Why would they put man hours towards decreasing their reputation if they don’t have to? They’re probably not even geared to break into their own devices. Then their PR team ran with it while one of many companies with the capability to crack the phone took a paycheck

This is different - this is genuine security, even if easily bypassed with preparation beforehand. Honestly, I credit some random apple dev who may have been looking to fix a bug related to long uptime as easily as they might’ve cared about security. I don’t think this was even on the radar of Apple leadership

This isn’t some moral superiority on Apple’s part, but it is good practice

Sure, F-Droid. It’s an app store that not only is exclusively foss, they only host things they can build from source in house and seem to have a decent review process - they tag anything from ads to integration with paid services, and those features are often buried so it seems like they’re pretty militant about it

It comes with all the drawbacks that entails, but I generally check there first myself

I think that’s fair.

I don’t have AI integration in my ide, mostly by choice -if I pushed for it I could make it happen, but I just don’t think that’s a good idea at this point

AI can be a crutch . One that limits you to the level of a baby developer. If you can’t effortlessly understand what it gives you, frankly you shouldn’t be using it.

Bounce ideas of chat gpt. It sounds like you’ve got the right idea - your reaction sounds correct to me, you should never ever trust it… You must only use it, and that’s the tone I get from your post.

It is a tool, you are a programmer. You exploit tools, you do not trust any tool. You are the one who turns ideas into actions, never forget that and you can use this new tool anywhere it makes your life easier

In fairness, about 50% of my code by lines is written by AI these days, and I don’t have it linked into my code base. That claim isn’t ridiculous

Now, of that 50% is 88% long repetitive crap that I could easily write but find mentally draining, the other 10% is something simple that I would normally copy paste from elsewhere because I forgot the exact syntax (and don’t exactly remember where I used it last) and me giving it a shot with things I don’t want to do, like restyling a page. The last 2% is me giving it a shot with business logic for shits and giggles, occasionally I’ll try to coach it through the solution but usually I just grab bits and pieces and rewrite it myself

Granted, this is the easiest and most simple and repetitive code, but it’s still a godsend. Now can AI write the other 50%? With a proper setup where it ingests the code base into a vector store it might get up to 75%, if I was willing to coach it through my tasks carefully (taking more time than the task would take me) I could probably get it up to 85% or 90%, but that last 10%? It just can’t, it’s not even close

It’s not taking my job without a paradigm shifting breakthrough or two on the scale of “all you need is attention”. Even then, it only works if you write your prompts like code… If you don’t understand how to use it and understand the code well enough to communicate the goal explicitly and unambiguously, you’re not going to be able to drive it where you want it to go

To put it another way, you can build 90% of the system in 10% of the time it takes to complete the last 10%

Can confirm. My phone got kicked off when they started sunsetting 3G. They called me (on said phone with no service lol) and said I needed a new phone. I said “no I don’t, put me back on the network”. We went back and forth, then they forwarded me to the tech department

The tech says “you need a new phone”. I said “no I don’t, I have all but one of the new bands and others with my phone have already gone through this process with you guys”. He said “you can’t believe everything you read online”, I said “be that as it may, I looked at the specs for both my phone and your network, and it meets the requirements”

He starts telling me there’s nothing he can do on his end, I say he just has to find an override to stop blocking my phone. He says he doesn’t have any options like that, I promise him it’s there

After getting tired of going in circles, I say if he doesn’t know how to do it he needs to ask someone or pass me to a higher tier. Surprise surprise, my phone instantly shows bars and he tries to gloss over the whole thing

Pretty often, but then you can just refactor the code so you can use it for more situations

What LLMs are good at are the opposite - when the thing you want to do is almost exactly the same, but nearly all the details need to be changed

Say you want a page to edit account details, and another page to edit community details. And the API paths to do this will be even more similar - but because they’re different things, you’d have to get fancy with the design to make code that works for both… It’s possible, but there will be trade-offs

LLMs are great at it though… Pass in the account page, give it the object definition for the community details, and it’ll spit it out for you

No, they’re not people. They’re royalty

So the CEO gets less than half their salary for the year?

Sounds like a great deal for the company. Until, you know, the whole thing collapses because they laid off the workers who kept the whole thing running

Ah, but that’s the beauty of it. Why are they here? If it’s to troll, don’t give them what they want. If it’s for social interaction… Why are they venturing out of their echo chamber?

Every interaction with a community pulls you slightly closer to the group consensus. You can fight it to some extent, but we’re wired to fit in with the tribe

Social rejection is wired similar to pain in our brains - it’s far more salient, far more memorable and impactful, than normal interactions.

The highest form of this is rejection by the community - it hurts most when everyone’s attention is on you and they all reject you. Even a single person quietly reaching out afterwards is like a lifeline - it stands out to you. It takes hundreds or thousands of “normal” interactions to counteract one extreme negative one

A supportive community back home doesn’t counteract the impacts from an away game. Don’t go to their turf, let them come to ours. Do not feed them - we have better content, they’ll lose members to us, and if we do it right they’ll shrink until their echo chambers can no longer sustain themselves

I just pass the create table statements after the instructions. It does pretty good up to 2 or 3 tables, but it will start to make mistakes when things get complicated

On the plus side, it’ll generate tedious code very well - double checking it is less draining than writing it yourself. Especially because I make more typos than it does - I often use it to get a starting point, then write the business logic myself

What do you mean? It works the same way, the opinion of the community will pull you closer to the group consensus. Too much exposure will have horrible things you don’t really believe spilling out of your mouth

Don’t go there, don’t spread word about it, don’t feed it in any way. It’s like flood water - pull others out of it if you can, but minimize your exposure

As to shutting them down if you have the ability? Shutting down a cesspool is good - it fragments the echo chamber, and some members won’t make the migration. The only question is if I trust the one making that decision to remain impartial

Therapy is also about fit. It takes something like 5 tries to get a good match - both the kid and the parent need to be on board, or the whole thing will end up as a bad experience for everyone involved

On one hand, I kinda get it. Corporations do things, and if they get killed over small oversights there will be very real effects for many people

On the other hand, holy fuck, look around. Everything is going to shit, the planet is literally dying and we’re not even living it up anymore

How about the opposite? A decentralized system, avoiding concentrating power in any one place, and focusing on building platforms for coordination to act on larger scales

No, the community needs to cyber bully them off the platform. They need to feel rejection for their words, not censorship. Censorship lets them frame themselves as the victim as they seek out a smaller echo chamber on the fringes. They need to learn their words will turn the community against them

We still have to live with them. We can’t ignore them or silence them - we have to correct them