ICE lawyers citing non-existent cases

[jacksonlamb]

Apparently, neither do you.

Read the article.

You are trolling, as usual, without a clue. Who is telling you to be stupid?

 

[airplanemechanic]

You are trolling, as usual, without a clue. Who is telling you to be stupid?

You are a fucktard idiot. You are shown a study proving that AI is bias and you then attack me.

Means I'm right. You can't prove otherwise.

Thanks for playing. Go **** yourself, jackasslamb.

 

[jacksonlamb]

You are a fucktard idiot. You are shown a study proving that AI is bias and you then attack me.

Means I'm right. You can't prove otherwise.

Thanks for playing. Go **** yourself, jackasslamb.

You are a witless Trumpcuck! You are almost never right, bud.

 

[NewsVine_Mariyam]

You DO realize AI is written by people, right? It's been WELL documented that popular AI platforms have a left wing slant.

Popular AI Models Show Partisan Bias When Asked to Talk Politics

In a recent study, both Republicans and Democrats perceived left-leaning bias in LLMs’ discussion of contentious topics.

www.gsb.stanford.edu

"AI is written by people?" That is not actually an accurate statement.

Saying AI is “written to be woke” is like saying spellcheck is written to have opinions. It’s not—it learns patterns and then has limits on what it’s allowed to output.

The system is engineered by people, yes, but the behavior it exhibits comes from the archival data that it trains on, not someone scripting viewpoints.

 

[MostBoringManInTheWorld]

You are a fucktard idiot. You are shown a study proving that AI is bias and you then attack me.

Means I'm right. You can't prove otherwise.

Thanks for playing. Go **** yourself, jackasslamb.

TRIGGERED

 

[frigidweirdo]

G-d bless any honourable ICE agent.

Yeah, any "honourable" ICE agent, just got to find one.

 

[airplanemechanic]

"AI is written by people?" That is not actually an accurate statement.

Saying AI is “written to be woke” is like saying spellcheck is written to have opinions. It’s not—it learns patterns and then has limits on what it’s allowed to output.

The system is engineered by people, yes, but the behavior it exhibits comes from the archival data that it trains on, not someone scripting viewpoints.

AI is a program, programs are written by computer programmers.

Don't even begin to play with me.

 

[airplanemechanic]

TRIGGERED

No, I just never seem to fully grasp the stupidity of the left. Every day they surprise me with just a little bit more stupidity than the day before.

 

[airplanemechanic]

You are a witless Trumpcuck! You are almost never right, bud.

Fuckoff newbie. I'm always right, ************. Nobody has proven wrong on here yet, certainly not you, jackasslamb.

 

[jacksonlamb]

Fuckoff newbie. I'm always right, ************. Nobody has proven wrong on here yet, certainly not you, jackasslamb.

I just did, and others have today. You are but a jumped-up Trump cuck, not a jot more.

 

[jacksonlamb]

AI is a program, programs are written by computer programmers.

Don't even begin to play with me.

Not to be woke, though. What an immature, stupid cuck thing to say.

 

[Dragonlady]

G-d bless any honourable ICE agent.

Well that's going to be a short list. Anyone joining ICE right now is doing so to abuse racial minorities.

 

[NewsVine_Mariyam]

AI is a program, programs are written by computer programmers.

Don't even begin to play with me.

Nobody is playing with you. I know how computer programs work and are written because this is my career field. I've been writing computer code as a software developer and engineer for decades. Conventional software application development consists of a series of IF-THEN-ELSE statements in order to produce a "pre-programmed" result. That's not how AI works.

The simplest example I can give is if you took all of your email and dumped it into a data file where each data element was parsed into fields, such as Sender, Receiver, Subject, etc. You set aside say 25% of your records and then add a label to each of the remaining 75% of the email records indicating if the email is considered SPAM or not. Once all of the records are labeled, you create a machine learning model that makes trains on your 75% dataset by processing each record and attempting to "learn" what makes an email spam or not and that prediction, made by the model gets recorded in the label field as either 0 for not spam or 1 for spam (this assignment is arbitrary).

At the end of the training process, you run some mathematical functions to see how well your model performed meaning how well was it able to correctly predict whether something is spam or not. These functions measure accuracy, precision, recall, etc.

You can adjust your model, tweaking the various features (fields) and other components to try to make it as precise as possible and once you're satisfied with the results THEN you test it on the 25% of the dataset that you held back to see how well it does on "unseen" data meaning data that is not labeled. In other words, there is no label to give it a hint on how the email should be classified therefore it must rely on its previous "training" on the 75% of the data in order to attempt to properly predict the right answer. Surely you can see how this differs from a series of IF-THEN-ELSE statement directing output or action based on which logical branch your code traversed.