What most people call “algorithms” are not algorithms, artificial intelligences have replaced math formulas to back the recommendations your phone makes to you. Welcome to techno-feudalism!
Algorithms are discrete operations done in a sequence, they are typically blind to ideas or personal opinions and are used for data processing, filtering, or calculations. Facial recognition algorithms are typically what you would call a complex algorithm as it uses conditions based on what the camera captures to make decisions as to where the face of said person is, if it is a person, are they smiling? What are their physical traits? Do they have white, brown, or dark skin, to adjust the contrast level for a good photo.
Artificial intelligence models are not math driven. They are language models, trained through machine learning and operating on a transformer architecture. Not only do they not replicate the reasoning a human (they predict the probability of action based on statistics of choices of others) and are most definitely going to be used to influence choices and opinions, not in an educational manner but in coercive and controlling manner. And it will become very hard to understand what AI produced propaganda will become. Even scarier though to me, is that it will be even harder to understand the effects of AI use, since all political actions have a direct effects which are often easy to predict, but they also cause these “ripple effects” which are the more dangerous side of politics and these ripple effects caused by AI use in politics will be impossible to predict for me, you and the AIs due to their early stage.
it dude? lol. i make robotics. including computer vision and ml. worry not. i’ll think of you every time an algorithm gets too many layers and starts to smell like an ai.
Computer vision and machine learning has little to do with robotics, but alright since you seem to work with know something about thermodynamics and are not a sys_admin like all the desk jockeys, i ll be nice and call you the E word.
But robotics itself is such a small niche that random people invoke because they see it in movies that I am skeptical to believe it as a systems and automation engineer, that s basically as close as 95% of engineer will ever get to robotics unless you re in China (mechatronics is not robotics btw) or work for Kuka/ABB…
So we went from “i make robotics” to figuring out you re a CNC operator typing in G-code…? listen man, pure theoretical engineering and using mastercam or whatever MCAD/CAD software to cut up some chunk of 6061 aluminium or soft steel is not the same, and I say this as someone who has done both.
I know it is but I do not want to call it an algorithm? I’m asking as someone who doesn t work in it but pretty close. As I see it, AI is a set of functions turned into tensor operations in a dimensional space (basically my university matrices class) and the parameters weights are adjusted based on collected data (human data as i understand in this case it could be reddit comment libraries or books or Meta data ).
That last bit is why I don’t confuse it with an algorithm since then it would be a heuristic. Algos are predictible, like running a graphic simulation that uses a physical and logic model, yeah id agree that s all a system of algorithms. Heuristic in terms of structure and method may look like algorithms but they are “rule of thumb” approximations for specific problems where we cannot build a math system (like an algo). At this point it is the human element and adaptation of the AI to you that would make it a heuristic.
What most people call “algorithms” are not algorithms, artificial intelligences have replaced math formulas to back the recommendations your phone makes to you. Welcome to techno-feudalism!
that’s literally algorithms lol.
Algorithms are discrete operations done in a sequence, they are typically blind to ideas or personal opinions and are used for data processing, filtering, or calculations. Facial recognition algorithms are typically what you would call a complex algorithm as it uses conditions based on what the camera captures to make decisions as to where the face of said person is, if it is a person, are they smiling? What are their physical traits? Do they have white, brown, or dark skin, to adjust the contrast level for a good photo.
Artificial intelligence models are not math driven. They are language models, trained through machine learning and operating on a transformer architecture. Not only do they not replicate the reasoning a human (they predict the probability of action based on statistics of choices of others) and are most definitely going to be used to influence choices and opinions, not in an educational manner but in coercive and controlling manner. And it will become very hard to understand what AI produced propaganda will become. Even scarier though to me, is that it will be even harder to understand the effects of AI use, since all political actions have a direct effects which are often easy to predict, but they also cause these “ripple effects” which are the more dangerous side of politics and these ripple effects caused by AI use in politics will be impossible to predict for me, you and the AIs due to their early stage.
Algorithms are based on logic whereas LLMs are based on learning. LLMs have a layer of semantic understanding, which makes them more powerful.
laughs in sw engineer.
Meanwhile I, systems engineer wonders why the IT dude called himself an “engineer”
it dude? lol. i make robotics. including computer vision and ml. worry not. i’ll think of you every time an algorithm gets too many layers and starts to smell like an ai.
Computer vision and machine learning has little to do with robotics, but alright since you seem to work with know something about thermodynamics and are not a sys_admin like all the desk jockeys, i ll be nice and call you the E word.
But robotics itself is such a small niche that random people invoke because they see it in movies that I am skeptical to believe it as a systems and automation engineer, that s basically as close as 95% of engineer will ever get to robotics unless you re in China (mechatronics is not robotics btw) or work for Kuka/ABB…
lol. say that to my 6 axis machine that picks and pours and places.
bad bot. try again.
So we went from “i make robotics” to figuring out you re a CNC operator typing in G-code…? listen man, pure theoretical engineering and using mastercam or whatever MCAD/CAD software to cut up some chunk of 6061 aluminium or soft steel is not the same, and I say this as someone who has done both.
Artificial intelligence itself is a series of math formulas on very big vectors though.
I know it is but I do not want to call it an algorithm? I’m asking as someone who doesn t work in it but pretty close. As I see it, AI is a set of functions turned into tensor operations in a dimensional space (basically my university matrices class) and the parameters weights are adjusted based on collected data (human data as i understand in this case it could be reddit comment libraries or books or Meta data ).
That last bit is why I don’t confuse it with an algorithm since then it would be a heuristic. Algos are predictible, like running a graphic simulation that uses a physical and logic model, yeah id agree that s all a system of algorithms. Heuristic in terms of structure and method may look like algorithms but they are “rule of thumb” approximations for specific problems where we cannot build a math system (like an algo). At this point it is the human element and adaptation of the AI to you that would make it a heuristic.
Makes sense?
Shh!