The estimate is 16 to 42 million birds die each year from hitting glass. On the low end, each building (of which there are 15.5 million in Canada approximately [source]) on average one bird dies due to hitting glass per building. That is not quite in the realm of 'piles of birds around every low rise building', given that they do not occur all at once (it is an annual number after all), and the bodies do decompose, get eaten, or get cleaned up.
An LLM is not inherently non-deterministic though - if you don't randomly sample and instead have a fixed rule (which is what the recent fingerprint embedding approach does), if applied in all cases the output is deterministic, as the neural net at its core is deterministic function. A lot of the randomness beyond that is due to optimizations [source].
LLMs are however unreliable at 'compiling'. Whether or not it will be able to complete the requested task (translate human language into code) correctly it not guaranteed - at least nowhere near the compilers we use.
There collaborative filtering algorithms do tend to have a popularity bias. The other downside is that these algorithms also don't help new music and musicians get found.
Also, if it went the other way, there would be gripes nonetheless: if she didn't endorse you would have gotten "So much for vote blue no matter who" or something similar.
The outcome has been determined: El-Sayed the nominee of the Democratic party, and rather than than fracturing the party and giving the Republicans (and Trump) the win, they chose to get behind the (progressive) candidate that was elected in the primary, rather than pulling a Cuomo.
It doesn't have to be on purpose. Web crawlers like Google's specifically identify themselves as a bot. A headless browser just acts like any normal browser, and uses the same user agent as one, just controlled by code rather than a keyboard and mouse.
Classic bots (e.g. standard web crawlers, as used by search engines, that report their both nature inside their user agent) or the modern LLM-controlled bots that control a full modern browser that also runs javascript?
Because I can imagine they account for the first category, while missing the latter, which have become increasingly more prominent.
I don't think making the battery itself waterproof is necessarily the solution here though: you would need to make the battery user replaceable without sacrificing the waterproofing of the entire device.
There are ways of accomplishing this in devices though: action cameras generally have replaceable batteries with an extra waterproofing gasket for the battery tray. Those are even designed for frequent-ish swaps (though, outside of water).
Privilege detection can be detected (i.e. by having a certain application run as root), if you have a piece of software try out a bunch of things (fuzzer, or more recently an LLM-in-a-loop) until this detection is triggered - evidence of actual success, even when things like LLMs are prone to 'hallucinations' (or lies) - you've got evidence of an exploit.
We have reached a point where this methodology is applied to a bunch of codebases, with some technological advances making it more capable and faster, hence the larger number of exploits being found.
There is a replace filter on lemmy that replaces words in a configurable list with removed, though. I would argue there is some kind of algorithm for this present albeit visible rather than the invisible hiding of entire posts.
Another indication that interacting with others from outside your bubble is the best vaccine against right wing rethoric regarding immigration: in rural areas and suburbs such interactions tend to be much rarer than within the city...
It just so happens that many video codecs are based on image formats, so ffmpeg already has a lot of the complex machinery to do so available to also implement these image formats - internally it can just handle it as a single frame of video with specialized formats for that.
Imagemagick (and other tools) also work, but why use multiple pieces of software if what you already have is adequate? ImageMagick is also software, and can also have vurnabilities.
That's fair. I actually don't think we disagree that much - I just think I have trouble conveying what I am trying to say. Whenever someone talks about 'shallow statistical predictions', I think about older techniques like Statistical Machine Translation which even had trouble with things like word order, LLMs handle text on a higher level of abstraction (which I described as a form of textual understanding) - and hence handle things like word order better - but are still inherently statistical predictors. The model stores info about how words interact and relate to one another, but it does not 'understand' what the words actually (physically?) represent beyond these interactions nor does it 'understand' what it is doing. Albeit, those interactions are modeled well enough to give a convincing replica of doing so.
That is no more understanding the problem than my spelling or grammar checking understands the comment I’m writing.
My general point would be that even a grammar checking can have some form of understanding of the text, no matter how shallow. The checker probably has a rule for when a is used versus an, if this rule generalizes across new words that rule that were previously unseen is a form of 'understanding' of the language being used in my view, despite being overly simplistic, while rote memorization - having a list of words that are followed by an - may not be.
LLMs are a weird case, because their internal representations for many concepts generalize even across new settings / inputs - in that sense the model has a form of understanding of what is being given, while for many other concepts the patterns break down even in the simplest of cases. It may 'know' that the preceding text is the writing of an essay - and it should autocomplete accounting for that -as is defined in its weights, but this understanding is shallow - it does not know why it knows, or how it knows - it cannot self reflect as it does not see and understand its own internal workings, or account for that. Yet the internal representations represent a form of text understanding that can be useful nonetheless - it is a language model after all.
My comment was intended to show this duality, hence the duality between the two paragraphs.
And this is the crux of my beef, I think, because stochastic pattern matching is not understanding, it’s a mathematical representation of how the model processes your input tokens. The fact that it has to start over every time you provide it input, and uses the previous input/output tokens as context is why this is not ‘understanding’, it’s just fancy token prediction that gives a middling-to-passable facsimile to intelligence and understanding things.
The problems you note in your second paragraph fundamentally undermine the argument that there is any form of understanding to the AI, because those are basic mistakes that a trivial understanding of the problem would prevent.
I am not entirely grasping the point you are trying to make here. I am certainly not arguing that it is conscious, self-aware, or in any way not a mechanical procedure that is being performed (I would not argue for that!). My key point is that it is not a simple black or white it understands / it does not understand - it may have internal representations that relate many concepts together, allowing it to draw upon these links when generating text, giving it a certain semantic understanding of the language and text it is using, while simultaneously not having a bit of self-awareness.
These models do have some form of understanding though. There are features for bugs and typos, and general features that map descriptions and pieces of code. It understands the code in so far it helps with next token prediction.
The bigger problem is that these language models are inherently unreliable and stochastic in how they generate. You request a feature - and it destroys something else in the process - because a single incorrect prediction caused it to diverge and skip a portion of the original code. You request a small modification - and it decides to restart from scratch - because random sampling made a different way of doing something more likely rather than what was there already. Errors compound and the model has no way of fixing or correcting them.
Funnily enough, a deviation of +- 1.5mm is allowed according to the ISO standard, so it is technically allowed to be much larger than that.
The estimate is 16 to 42 million birds die each year from hitting glass. On the low end, each building (of which there are 15.5 million in Canada approximately [source]) on average one bird dies due to hitting glass per building. That is not quite in the realm of 'piles of birds around every low rise building', given that they do not occur all at once (it is an annual number after all), and the bodies do decompose, get eaten, or get cleaned up.
An LLM is not inherently non-deterministic though - if you don't randomly sample and instead have a fixed rule (which is what the recent fingerprint embedding approach does), if applied in all cases the output is deterministic, as the neural net at its core is deterministic function. A lot of the randomness beyond that is due to optimizations [source].
LLMs are however unreliable at 'compiling'. Whether or not it will be able to complete the requested task (translate human language into code) correctly it not guaranteed - at least nowhere near the compilers we use.
There collaborative filtering algorithms do tend to have a popularity bias. The other downside is that these algorithms also don't help new music and musicians get found.
Your instance (lemmy.ml) has a word filter - words in the filter get turned into removed.
Also, if it went the other way, there would be gripes nonetheless: if she didn't endorse you would have gotten "So much for vote blue no matter who" or something similar.
The outcome has been determined: El-Sayed the nominee of the Democratic party, and rather than than fracturing the party and giving the Republicans (and Trump) the win, they chose to get behind the (progressive) candidate that was elected in the primary, rather than pulling a Cuomo.
Less tennis and more the anti-golf homeowners federation whose windows have gotten smashed in a couple times too much.
It doesn't have to be on purpose. Web crawlers like Google's specifically identify themselves as a bot. A headless browser just acts like any normal browser, and uses the same user agent as one, just controlled by code rather than a keyboard and mouse.
Classic bots (e.g. standard web crawlers, as used by search engines, that report their both nature inside their user agent) or the modern LLM-controlled bots that control a full modern browser that also runs javascript?
Because I can imagine they account for the first category, while missing the latter, which have become increasingly more prominent.
I don't think making the battery itself waterproof is necessarily the solution here though: you would need to make the battery user replaceable without sacrificing the waterproofing of the entire device.
There are ways of accomplishing this in devices though: action cameras generally have replaceable batteries with an extra waterproofing gasket for the battery tray. Those are even designed for frequent-ish swaps (though, outside of water).
I have seen others too, I think that the difference lies in that privilege escalation is just more critical and hence prominent.
Privilege detection can be detected (i.e. by having a certain application run as root), if you have a piece of software try out a bunch of things (fuzzer, or more recently an LLM-in-a-loop) until this detection is triggered - evidence of actual success, even when things like LLMs are prone to 'hallucinations' (or lies) - you've got evidence of an exploit.
We have reached a point where this methodology is applied to a bunch of codebases, with some technological advances making it more capable and faster, hence the larger number of exploits being found.
Unlikely, since DJI Mavic drones have a 500m (or 1640ft) maximum altitude limit in software, and usually lower due to legal limits. [source]
That is, unless the neighbor modded the software to go above the legal limits set.
There is a replace filter on lemmy that replaces words in a configurable list with removed, though. I would argue there is some kind of algorithm for this present albeit visible rather than the invisible hiding of entire posts.
Another indication that interacting with others from outside your bubble is the best vaccine against right wing rethoric regarding immigration: in rural areas and suburbs such interactions tend to be much rarer than within the city...
It just so happens that many video codecs are based on image formats, so ffmpeg already has a lot of the complex machinery to do so available to also implement these image formats - internally it can just handle it as a single frame of video with specialized formats for that.
Imagemagick (and other tools) also work, but why use multiple pieces of software if what you already have is adequate? ImageMagick is also software, and can also have vurnabilities.
You would be inventing some style of UUID. Include a timestamp in front, so that it is sortable and you have Snowflake.
That's fair. I actually don't think we disagree that much - I just think I have trouble conveying what I am trying to say. Whenever someone talks about 'shallow statistical predictions', I think about older techniques like Statistical Machine Translation which even had trouble with things like word order, LLMs handle text on a higher level of abstraction (which I described as a form of textual understanding) - and hence handle things like word order better - but are still inherently statistical predictors. The model stores info about how words interact and relate to one another, but it does not 'understand' what the words actually (physically?) represent beyond these interactions nor does it 'understand' what it is doing. Albeit, those interactions are modeled well enough to give a convincing replica of doing so.
My general point would be that even a grammar checking can have some form of understanding of the text, no matter how shallow. The checker probably has a rule for when
ais used versusan, if this rule generalizes across new words that rule that were previously unseen is a form of 'understanding' of the language being used in my view, despite being overly simplistic, while rote memorization - having a list of words that are followed by an - may not be.LLMs are a weird case, because their internal representations for many concepts generalize even across new settings / inputs - in that sense the model has a form of understanding of what is being given, while for many other concepts the patterns break down even in the simplest of cases. It may 'know' that the preceding text is the writing of an essay - and it should autocomplete accounting for that -as is defined in its weights, but this understanding is shallow - it does not know why it knows, or how it knows - it cannot self reflect as it does not see and understand its own internal workings, or account for that. Yet the internal representations represent a form of text understanding that can be useful nonetheless - it is a language model after all.
My comment was intended to show this duality, hence the duality between the two paragraphs.
I am not entirely grasping the point you are trying to make here. I am certainly not arguing that it is conscious, self-aware, or in any way not a mechanical procedure that is being performed (I would not argue for that!). My key point is that it is not a simple black or white it understands / it does not understand - it may have internal representations that relate many concepts together, allowing it to draw upon these links when generating text, giving it a certain semantic understanding of the language and text it is using, while simultaneously not having a bit of self-awareness.
These models do have some form of understanding though. There are features for bugs and typos, and general features that map descriptions and pieces of code. It understands the code in so far it helps with next token prediction.
The bigger problem is that these language models are inherently unreliable and stochastic in how they generate. You request a feature - and it destroys something else in the process - because a single incorrect prediction caused it to diverge and skip a portion of the original code. You request a small modification - and it decides to restart from scratch - because random sampling made a different way of doing something more likely rather than what was there already. Errors compound and the model has no way of fixing or correcting them.