Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Beliefs that are true for regular software but false when applied to AI

boydkane.com

Post reply on HN
Print view View original

371–380 of 461 posts

Re: Beliefs that are true for regular software but false when applied to AI

#371

Post by dweinus »

drsupergud wrote: post #11

> bugs are usually caused by problems in the data used to train an AI This also is a misunderstanding. The LLM can be fine, the training and data can be fine, but because the LLMs we use are non-deterministic (at least in regard to their being intentional attempts at entropy to avoid always failing certain scenarios) current algorithms are inherently by-design not going to always answer every question correctly that…

Fully agree. Also inherent to the design is distillation and interpolation...meaning that even with perfect data and governing so that outputs are deterministic, the outputs will still be an imperfect distillation of the data, interpolated into a response to the prompt. That is a "bug" by design

Re: Beliefs that are true for regular software but false when applied to AI

#372

Post by wrs »

mr_toad wrote: post #345
wrs wrote: post #36

My current method for trying to break through this misconception is informing people that nobody knows how AI works. Literally. Nobody knows. (Note that knowing how to make something is not the same as knowing how it works. Take humans as an obvious example.)

I think you’re confusing knowing how a system works with being able to predict how that system will perform. In a non-linear system the former is often easier than the latter. For example we know how planets “work” from the laws of motion. But planetary orbits involving > 2 bodies are non-linear, and predicting their motion far into the future is surprisingly difficult. Neural networks are the same. They’re actually…

Claude just analyzed the emotional content of a piece of music for me -- quite accurately -- by just looking at an uploaded PDF of the score. How does that work? "It's a nonlinear system" or "it's a bunch of matrix multiplication" is in no useful way an explanation. That's way down at the bottom of an explanatory abstraction hierarchy that we have only begun to make tools to begin to explore. It's like asking how humans work and getting the answer "it's just undergraduate chemistry!"

The study of LLMs is much closer to biology than engineering.

Re: Beliefs that are true for regular software but false when applied to AI

#373

Post by kbelder »

danaris wrote: post #334
silvestrov wrote: post #313

Earlier quoted context omitted.

I feel too many humanities teachers are like your brother. Why use 10 words when you could do 1000. Why use headings or lists, when the whole story could be written in a single paragraph spanning 3 pages.

I mean...this depends very heavily on what the purpose of the writing is. If it's to succinctly communicate key facts, then you write it quickly. - Discovered that Bilbo's old ring is, in fact, the One Ring of Power. - Took it on a journey southward to Mordor. - Experienced a bunch of hardship along the way, and nearly failed at the end, but with Sméagol's contribution, successfully destroyed the Ring and defeated Sa…

Sure, but different people judge differently what should be told as a story.

"When's dinner?" "Well, I was at the store earlier, and... (paragraphs elided) ... and so, 7pm."

Re: Beliefs that are true for regular software but false when applied to AI

#374

Post by kbelder »

bombcar wrote: post #308
mikkupikku wrote: post #299

Earlier quoted context omitted.

Maybe they remembered that a lot of people aren't actually good writers. My brother will send 1000 word emails that meander through subjects like what he ate for breakfast to eventually get to the point of scheduling a meeting about negotiating a time for help with moving a sofa. Mind you, I see him several times a week so he's not lonely, this is just the way he writes. Then he complains endlessly about his coworker…

There’s a thread here that could be pulled - something about using AI to turn everyone into exactly who you want to communicate with in the way you want. Probably a sci-fi story about it, if not, it should be written.

And AR glasses to modify appearance of everyone you see, in all sorts of ways. Inevitable nightmare, I expect.

Re: Beliefs that are true for regular software but false when applied to AI

#375

Post by danaris »

kbelder wrote: post #373
danaris wrote: post #334

Earlier quoted context omitted.

I mean...this depends very heavily on what the purpose of the writing is. If it's to succinctly communicate key facts, then you write it quickly. - Discovered that Bilbo's old ring is, in fact, the One Ring of Power. - Took it on a journey southward to Mordor. - Experienced a bunch of hardship along the way, and nearly failed at the end, but with Sméagol's contribution, successfully destroyed the Ring and defeated Sa…

Sure, but different people judge differently what should be told as a story. "When's dinner?" "Well, I was at the store earlier, and... (paragraphs elided) ... and so, 7pm."

Now, that's very true! But it's a far cry from implying that all or most humanities teachers are all about writing florid essays when 3 bullet points will do.

Re: Beliefs that are true for regular software but false when applied to AI

#376

Post by ako »

mattmanser wrote: post #293
ako wrote: post #273

Earlier quoted context omitted.

There's no magic involved, the LLM creators can go anywhere and rebuild an LLM with pretty much the same outcome, if they have the same training data. With unlimited time you could even reproduce the output of an LLM manually as it is just a lot of mathematics. Including reasoning, as that is mostly adding words in the context that will steer the word predication to include reasoning. As this is a useful LLM behavior…

Yes, but we can't inspect, reproduce or explain the emergent property independently. We can't pick out the "math reasoning" part or the "programming" part, or inspect how it's working, or selectively change any of it. You can't turn any dials or twiddle any knobs. You can't replace one part with another, or pick out components. You can't peek inside and say: "hey it's got an irrational preference for naming its varia…

So just because all the parts are collaborating to get the desired outcome, and specific aspects of that outcome cannot be contributed to specific parts of the llm you think we don’t understand LLMs? With mixture of expert systems we’re introducing dedicated subsystems into the llm responsible for specific aspect of the llm, so we’re partially moving in that direction.

But overall in my opinion if devs are able to rebuild it from scratch with a predefined outcome, and even know how to improve the system to improve certain aspects of it, we do understand how it works.

Re: Beliefs that are true for regular software but false when applied to AI

#377

Post by ako »

wrs wrote: post #36

My current method for trying to break through this misconception is informing people that nobody knows how AI works. Literally. Nobody knows. (Note that knowing how to make something is not the same as knowing how it works. Take humans as an obvious example.)

We don’t know how to build humans from scratch (other than letting nature do it) so that’s not really a relevant example. We only know how to fix certain aspects of defect humans, and how to minimize the chance of future defects, but that’s far from building a human from scratch.

Re: Beliefs that are true for regular software but false when applied to AI

#378

Post by casey2 »

I've never seen an essay intentionally miss the point so hard. Currently people use these systems to generate the same kind of artifacts that traditionally would be written by a human. Since there is such a clear delineation I seen no good reason to make the distinction.

Likewise a person you hire "could" take over the country and start a genocide, but it's rightfully low on your priority list because it's so unlikely that it's effectively impossible. Now an AI being rude or very unhelpul/harmful to your customer is a more pressing concern. And you don't have that confidence with most people either which is why we go through hiring processes.

The statics here are key and AI companies are geniuses at lying with statistics. I could shuffle a dictionary and outputting a random word each time and answer any hard problem. The entire point of AI is that you can do MUCH better than "random". Can anyone tell me which algorithm (this or chatgpt) has a higher likelihood of producing a proof of the RH after n tokens? No, they can't. But chatgpt can generate things in human timescale that look more like proofs than my bruteforce approach so people (investors) give it the benefit of the doubt even if it's not earned and could well be LESS capable than bruteforce as strange as it sounds.

Re: Beliefs that are true for regular software but false when applied to AI

#379

Post by ako »

wrs wrote: post #372
mr_toad wrote: post #345

Earlier quoted context omitted.

I think you’re confusing knowing how a system works with being able to predict how that system will perform. In a non-linear system the former is often easier than the latter. For example we know how planets “work” from the laws of motion. But planetary orbits involving > 2 bodies are non-linear, and predicting their motion far into the future is surprisingly difficult. Neural networks are the same. They’re actually…

Claude just analyzed the emotional content of a piece of music for me -- quite accurately -- by just looking at an uploaded PDF of the score. How does that work? "It's a nonlinear system" or "it's a bunch of matrix multiplication" is in no useful way an explanation. That's way down at the bottom of an explanatory abstraction hierarchy that we have only begun to make tools to begin to explore. It's like asking how hum…

Did it show you the reasoning? Did it recognize the notes, the scale, tempo and determine the emotion effect of these or did it use some other reasoning?

Re: Beliefs that are true for regular software but false when applied to AI

#380

Post by huhkerrf »

immibis wrote: post #282
huhkerrf wrote: post #245

Earlier quoted context omitted.

This is a pretty damning example of backwards product thinking. How often, truly, does this happen?

Never heard of terms of service?

Are you regularly getting emails with terms of service? You're, like, doubly proving my point.
Post reply on HN