Algorhithms

[RemyZee]

Could someone please explain algorithms in plain language? Why they have implications. Because if every decision a machine makes draws from a user's personal preferences and patterns, what does that say about people's ability to make independent choices and how will it effect things and impact people's ability to discriminate between alternatives? If that makes sense.

 

[Outdated]

Their so called algorhythms based on social and racial profiling don't work. They're good enough to funnel advertising to a more targeted audience for people that prefer not to do any of their own thinking and will do whatever the man on the screen tells them to. In my opinion that's over 50% of the population.

When it comes to trying to research specific topics on the net life is much more difficult, algorhythm driven results will just keep throwing commercialised and populist crap at you and you have to wade through mountains of irrelevant rubbish to find what you're looking for.

 

[saturnv]

An algorithm is a sequence of steps done to accomplish a task or calculate something. A classic example computer science students encounter is making a peanut butter sandwich. The instructor tells the student to say the steps. Open the jar of peanut butter, pick up a knife with your hand, scoop some peanut butter with a knife, etc. No matter what level of detail the students say there will be missing details in the algorithm that a person making a sandwich would fill in without consideration. For example, in the steps I gave, what if the peanut butter jar is not on the counter? How do I pick up a knife? What are hands? How do I scoop peanut butter? Ok, I scooped all the peanut butter at once and it fell on the counter. And so on. Computers only do exactly what they are programmed to do and it is up to the programmer to ensure that the steps work in the context the program runs in.

The algorithms people talk about with regard to social media involve programs that take as input all the information that has been collected about a person and try to calculate what things to show that person to manipulate them. The manipulation is usually geared towards making the person engage with the system as much as possible, eg continue scrolling or angrily respond with comments. Another example is the targeted advertising that Outdated mentioned.

Systems that only show you the information that you have requested are likely to be better for you. Imagine seeing content only from the people you follow or threads listed chronologically under topics like they are here. There might be manipulative things posted, but the system itself is not designed for that purpose.

People still have the ability to make independent choices if they attempt to limit how much information about them is available and avoid systems that are designed in the way I described. Knowledge about the way these things work might help as well, but I think it is best to avoid them altogether. In my opinion, it should be illegal to compile information in this manner and/or use it to manipulate people, but almost nothing has been done about it.

 

[Judge]

Upon seeing this thread, the first thing I did was to search for algorithmic bias in Duck Duck Go.

"Algorithmic bias describes systematic and repeatable harmful tendency in a computerized sociotechnical system to create "unfair" outcomes, such as "privileging" one category over another in ways different from the intended function of the algorithm. Bias can emerge from many factors, including but not limited to the design of the algorithm or the unintended or unanticipated use or decisions relating to the way data is coded, collected, selected or used to train the algorithm. "

 

[Shamar]

Could someone please explain algorithms in plain language? Why they have implications. Because if every decision a machine makes draws from a user's personal preferences and patterns, what does that say about people's ability to make independent choices and how will it effect things and impact people's ability to discriminate between alternatives? If that makes sense.

Algorithm is a term more often misused than used properly, generally by persons attempting to make themselves appear smarter in front of other less than well informed people in order to not say anything actually useful; see algorithmic, algorithming, algorithmism, algorithmises, algorithmized, algorithmation, most of which you will not find in any dictionary. Used in a similar manner to paradigm, such as paradigmattic, paradigming, paradigmized, paradigmed, and others, also not in any dictionary.

Algorithm is more properly defined as a process or set of rules to be followed in calculations or other problem-solving operations. When applied to AI, the algorithms are usually biased by the programmer or trainer, and reflect those biases. I question the general reliability of AI because of this. There can be no general intelligence algorithm, at best it can only mimic intelligence.

 

[Neonatal RRT]

Could someone please explain algorithms in plain language? Why they have implications. Because if every decision a machine makes draws from a user's personal preferences and patterns, what does that say about people's ability to make independent choices and how will it effect things and impact people's ability to discriminate between alternatives? If that makes sense.

In healthcare, we deal with algorithms at work daily. All of our differential diagnoses, all of our clinical practices during respiratory and cardiac arrests...are all based upon these "decision and intervention trees". If this, then that. It is an "agreed upon" (not always) way to do things, for consistency. It is especially helpful when you have a team of people at the bedside, because if there is "a way" to do things, and everyone knows, then the team can work together more cohesively and anticipate future steps.

 

[TBRS1]

When applied to nonspecific questions (not about a specific person in a specific set of circumstances with a clear go/no go outcome), an algorithm is a complex calculation in which the variables are invisible to those who are affected by it.

When the algorithm is used by an AI, the variables are invisible to those who use it as well (i.e.: nobody can trace the outcome backwards to find out how it was arrived at, therefore, the outcome must be taken on trust, and can not be tested except through application. This does not allow for comparison to other possibilities, making it impossible to know if THAT choice was REALLY the BEST choice).

The classic example of this takes place in Hitchhikers Guide to The Galaxy, where a computer is asked : What is the meaning of life? The computer responds with "42."

 

[Outdated]

Upon seeing this thread, the first thing I did was to search for algorithmic bias in Duck Duck Go.

They do the same thing though.

I've complained to duckduckgo a couple of times about it but I still keep using them because they don't suck anywhere near as badly as google. An old fashioned straightforward "string search" was far more practical.

For example, whenever I was looking for information on a topic I could exclude results from my search by putting a minus symbol in front of them:

autism -facebook -youtube -twitter

 

[SusanLR]

This is an explanation I like:
An algorithm is a procedure used for solving a problem or performing a computation. Algorithms act as an exact list of instructions that conduct specified actions step by step in either hardware- or software-based routines. Algorithms are widely used throughout all areas of IT.

The classic example of this takes place in Hitchhikers Guide to The Galaxy, where a computer is asked : What is the meaning of life? The computer responds with "42."

This is certainly a classic example. There is no list of instructions or step by step actions to answer the question. The answer of "42" makes as much sense as the meaning of life.

 

[Crossbreed]

How AIs, like ChatGPT, Learn​

 

[WhitewaterWoman]

Years ago when I started using Pandora, it was a little more transparent. Pandora asked what kind of music I liked and encouraged me to give lots of examples. After doing that task, Pandora told me what they (the algorithm) found about about what I liked in music. It was a list of about seven things, most of which I no longer remember, but one of them was that I liked music with some syncopation. Other things might have been like, minor key, folk style, etc.

When I listen to Pandora, it picks out for me music that it thinks I will like. Then I can use the up or down thumb to indicate whether I agree with their choices. My choices on each song are used by Pandora (the algorithm) to decided what to play next for me.

I liked this service. I still do. It broadens my field of listening experience and I have found a number of artists that I had no idea even existed.

So, unlike Spotify which plays only music I have said I want played, Pandora plays a wide range of music adapted to my taste.

Every now and then it tries to strike out in a new direction for me. If I don't like it, I just use the thumbs down and after a few tries in that direction, it gives up and accepts I don't like that music. This used to be more of a problem than it is now. At some point, a few years, there was a big dump of new music I liked and it has been much better since then.