Lol strong “I don’t want buyer’s regret” energy from this guy. Or maybe “I am way out of my league when evaluating how good something is” with perhaps a dash of “boots are delicious”.

Like he literally mentions that he can hear water sloshing around in the frame somewhere but then immediately concludes that it’ll probably go away on its own sometime in the future. I had a period in my life when I was like that. I consider it my “I had no fucking idea how naive I was or how things worked or how to take care of them” phase, and I was the last person anyone should have taken advice about anything from.

No, the exact % depends on how stable everything else is.

Like a trivial example, if you have 3 programs, one that sets a pointer to a random address and tries to dereference it, one that does this but only if the last two digits of a timer it checks are “69”, and one that never sets a pointer to an invalid address, based on the programs themselves, the first one will crash almost all the time, the second one will crash about 1% of the time, and the third one won’t crash at all.

If you had a mechanism to perfectly detect bit flips (honestly, that part has me the most curious about the OP), and you ran each program until you had detected 5 bit flip crashes (let’s say they happen 1 out of each 10k runs), then the first program will have something like a 0.01% chance of any given crash being due to bit flip, about 1% for the 2nd one, and 100% for the 3rd one (assuming no other issues like OS stability causing other crashes).

Going with those numbers I made up, every 10k “runs”, you’d see 1 crash from bit flips and 9 crashes from other reasons. Or for every crash report they receive, 1 of 10 are bit flips, and 9 of 10 are “other”. Well, more accurately, 1 of 20 for bit flip and 19 of 20 for other, due to the assumption that the detector only detects half of them, because they actually only measured 5%.

Who is talking about average consumers? We’re not trying to market something here.

Improved overall system stability and data accuracy? With error correction, you can also push performance farther, since you can tolerate a certain amount of errors, instead of needing to aim for 0% error rate.

The outside would be burnt and the inside raw. There might be a layer of well-cooked chicken between them, though just cutting through it to see that will contaminate the cooked bit from the raw bit. That’s why the penicillin sauce is so important.

A decent number of them have already sent cash gifts. Which is already spent paying off the loan for the wedding planner and the artists that made some concept art of the theme which is being presented as actual footage of the venue, despite a few of the visuals being things that cutting edge technology cannot physically produce.

Oh and after the last meeting where some of this was laid out, the wedding planner just started laughing hysterically, left the room, and isn’t answering or returning calls. The kids who were playing out front know a bit more information but are worried they’d get in trouble if they repeated what she said about the groom.

Just flashbang every room before you enter it, just to be safe.

Yeah, I appreciate the duck duck go ai chatbots, since they are free and don’t have the limit that using chatgpt directly for free had. Plus you can select other models (haven’t even been using chatgpt).

Had a presentation to make yesterday so figured fuck it, maybe the copilot integrated into power point can make this task I don’t want to do easier.

So I made a title slide and explained what was going on, then tried to get it able to edit the current file. Wasn’t able to but it was able to create a new file.

Took one look at the result and decided fuck it, I’ll just do it myself. How much it tried to play up everything without really saying anything was almost nauseating.

So I guess thanks copilot for sucking so much it inspired me to just spit out a better presentation. At least better to my standards; I bet an MBA would have preferred the other one that acted like the whole category the task was in was my own work and needed to be explained to the rest of the team that has been doing related tasks longer than I have.

Better build the whole OS around this tech which basically just exploits how easily we’re fooled by confident words.

Any orbit resulting from a collision will pass through that collision point unless there’s another collision to change it’s velocity again. The higher a collision sends an object, the more likely the “orbit” intersects with more atmosphere to cause drag, or it might even collide with the ground without drag.

Lol I thought your link was “here’s a rocket designed by an LLM” rather than one designed by the non-LLM AI.

LLMs are a local minimum that tech bros are stuck trying to optimize to a generally useful point because its language abilities are able to fool so many (just like how a real person talking with confidence can fool so many).

This obsession with LLMs is making me question general human intelligence more lol. It’s looking more and more like we are just dumb apes but get lucky and every now and then a smart ape is born and teaches the other dumb apes how to bring their stupidity to whole new levels.

Yeah, it’s good enough that it even had me fooled, despite all my “it just correlates words” comments. It was getting to the desired result, so I was starting to think that the framework around the agentic coding AIs was able to give it enough useful context to make the correlations useful, even if it wasn’t really thinking.

But it’s really just a bunch of duct tape slapped over cracks in a leaky tank they want to put more water in. While it’s impressive how far it has come, the fundamental issues will always be there because it’s still accurate to call LLMs massive text predictors.

The people who believe LLMs have achieved AGI are either just lying to try to prolong the bubble in the hopes of actually getting it to the singularity before it pops or are revealing their own lack of expertise because they either haven’t noticed the fundamental issues or think they are minor things that can be solved because any instance can be patched.

But a) they can only be patched by people who know the correction (so the patches won’t happen in the bleeding edge until humans solve the problem they wanted AI to solve), and b) it will require an infinite number of these patches even to just cover all permutations of everything we do know.

Here’s an example I ran into, since work wants us to use AI to produce work stuff, whatever, they get to deal with the result.

But I had asked it to add some debug code to verify that a process was working by saving the in memory result of that process to a file, so I could ensure the next step was even possible to do based on the output of the first step (because the second step was failing). Get the file output and it looks fine, other than missing some whitespace, but that’s ok.

And then while debugging, it says the issue is the data for step 1 isn’t being passed to the function the calls if all. Wait, how can this be, the file looks fine? Oh when it added the debug code, it added a new code path that just calls the step 1 code (properly). Which does work for verifying step 1 on its own but not for verifying the actual code path.

The code for this task is full of examples like that, almost as if it is intelligent but it’s using the genie model of being helpful where it tries to technically follow directions while subverting expectations anywhere it isn’t specified.

Thinking about my overall task, I’m not sure using AI has saved time. It produces code that looks more like final code, but adds a lot of subtle unexpected issues on the way.

Going for less known names can also help, as they are trying to build/maintain a reputation in addition to sales.

IKEA is an interesting brand because it spans from incredibly cheap to nice quality, and personally, I find the cheapness is more in the material selection than the design. Like the furniture I got from them at my last place all survived the move to my current place, even the one I got frustrated with and stopped caring if it made it when taking it apart, it still stands solid today. They are one of the few that has decent value, though their prices can get pretty high at the high end.

Yeah, it’s more of a late stage capitalism “luxury” where the difference isn’t so much in the quality as in the price because people conflate “price” with “quality” and “desireability”.

And I do understand it, at least to a degree. I try to do research on more expensive items or ones I’m looking for quality in, but it’s kinda exhausting, and often a cycle of “I want thing, see it in store and remember I want it, look at options, no idea which (if any) are decent and which suck, start looking online, decide I don’t want to do this right now, move on, forget to do research, repeat next time I’m at that store”.

The easy mode of doing that would be look at options, assume cheapest ones suck, most expensive is too much, get one of the ones a little cheaper. At which point, the seller just needs to set a higher price to get a sale on the crappy ones.

Also all those glass parking lot comments.

Yeah, I thought it might be a different kind of AI, at least, until it fucking said “LLM”.

They don’t assess risk, they correlate words. Even if they can be massaged to use a tool to assess risk in a more accurate way, they don’t evaluate risk assessments and determine how that should affect strategy or tactics, they correlate words. They don’t even do math that puts a value on human life to determine if an action is worth the cost, they just correlate fucking words. All based on given training data, so anything they can offer for real is already out there, and everything else is suspect because it’s purely based on correlations of words.

It’s like reading the Art of War and thinking that means you’re ready to be a general.

But something AI might do is introduce uncertainty that might get used to try to excuse a nuclear strike a human wanted to do.

If you want a demo on how bad these AI coding agents are, build a medium-sized script with one, something with a parse -> process -> output flow that isn’t trivial. Let it do the debug, too (like tell it the error message or the unwanted behaviour).

You’ll probably get the desired output if you’re using one of the good models.

Now ask it to review the code or optimize it.

If it was a good coding AI, this step shouldn’t involve much, as it would have been applying the same reasoning during the code writing process.

But in my experience, this isn’t what happens. For a review, it has a lot of notes. It can also find and implement optimizations. The weighs are the same, the only difference is that the context of the prompt has changed from “write code” to “optimize code”, which affects the correlations involved. There is no “write optimal code” because it’s trained on everything and the kitchen sink, so you’ll get correlations from good code, newbie coders, lesson examples of bad ways to do things (especially if it’s presented in a “discovery” format where a prof intended to talk about why this slide is bad but didn’t include that on the slide itself).

I bet if these get used that soon after there will be new extreme maintenance videos that make those cell towers look like nothing. Probably some guy hanging from a powered cable climbing device, showing the things on the ground getting smaller and smaller, occasionally taking a puff from an asthma inhaler because they were told an oxygen tank would cause weight issues (it’s actually about financial issues), until enough people die that they realize it’s cheaper to pay for oxygen than training new workers.

Depending on how the turbine is set up, it could generate AC power instead of DC. I believe they even have several options on how to do this.