Slashdot Mirror
A New Bill Would Force Companies To Check Their Algorithms For Bias (theverge.com)
An anonymous reader quotes a report from The Verge: U.S. lawmakers have introduced a bill that would require large companies to audit machine learning-powered systems -- like facial recognition or ad targeting algorithms -- for bias. The Algorithmic Accountability Act is sponsored by Senators Cory Booker (D-NJ) and Ron Wyden (D-OR), with a House equivalent sponsored by Rep. Yvette Clarke (D-NY). If passed, it would ask the Federal Trade Commission to create rules for evaluating "highly sensitive" automated systems. Companies would have to assess whether the algorithms powering these tools are biased or discriminatory, as well as whether they pose a privacy or security risk to consumers.
The Algorithmic Accountability Act is aimed at major companies with access to large amounts of information. It would apply to companies that make over $50 million per year, hold information on at least 1 million people or devices, or primarily act as data brokers that buy and sell consumer data. These companies would have to evaluate a broad range of algorithms -- including anything that affects consumers' legal rights, attempts to predict and analyze their behavior, involves large amounts of sensitive data, or "systematically monitors a large, publicly accessible physical place." That would theoretically cover a huge swath of the tech economy, and if a report turns up major risks of discrimination, privacy problems, or other issues, the company is supposed to address them within a timely manner.
The Algorithmic Accountability Act is aimed at major companies with access to large amounts of information. It would apply to companies that make over $50 million per year, hold information on at least 1 million people or devices, or primarily act as data brokers that buy and sell consumer data. These companies would have to evaluate a broad range of algorithms -- including anything that affects consumers' legal rights, attempts to predict and analyze their behavior, involves large amounts of sensitive data, or "systematically monitors a large, publicly accessible physical place." That would theoretically cover a huge swath of the tech economy, and if a report turns up major risks of discrimination, privacy problems, or other issues, the company is supposed to address them within a timely manner.
What is bias? Does "bias" mean "not a white male?"
In Asia, AI training data is almost exclusively Asian. That means results will skew Asian. Is that evidence of algorithmic bias? How would you go about determining that?
means what you think it means.
Dear lords as a cybersecurity/IA professional reading this thing makes my head hurt from the buzzword bingo and absolutely worthless definitions included within.
"taking into account the novelty of the technology used" - what? WTF does that mean? Is that a legal phrase or the verbal diarrhea of a staffer that thinks this sounds cool but is worthless from a legislative, and more importantly judicial, perspective? The corporate lawyers are going to run rings around this B.S.
You can't legislate them out of existence.
Bias is favoring one thing over another. Which is what you want certain algorithms to do. I want Youtube to find stuff I like. I want Google to find pages that are relevant to me.
Not sure how you are going to tease out the "good" bias from the "bad" bias, though. To extend your example, if 90% of the people in Hong Kong looking for a famous concert pianist are trying to find Lang Lang, who is hugely popular there, he's going to come up pretty fast when looking for concert pianists in general, which is what you want. It means the algorithm is being biased against Helene Grimaud, which is fine, because she isn't what most people are looking for in Hong Kong. That doesn't mean she doesn't come up at all, it just means she's ranked lower in the search results.
My Other Computer Is A Data General Nova III.
Let's get some hard data showing the bias that is present in censorship. The US is more conservative than liberal:
https://news.gallup.com/poll/2...
Without algorithmic bias online media would lean conservative for the simple reason that the US has more conservatives than liberals. Yet somehow online platforms (Reddit, Facebook etc.) tilt overwhelmingly liberal.
This can only be a result of bias that has been put into algorithms and sanctioned.
I am currently reading "The Sum of Small Things." In the first chapter, the idea of different racial groups having different demand levels is shown through the data.
There are valid reasons for that difference in demand. However, to pretend it isn't there is to try to live in denial.
In the case, in the book, the increased demand by Blacks for conspicuous consumption goods, ceteris paribus, is based on the belief that many Blacks find it necessary, but often not the result of conscious decision making, to carry visible markers of the middle class because it is not assumed. Now, we can reject this conclusion. However, to reject the discussion because we reject the data gets us no closer to truth. instead, it moves us away from truth.
Say I want to look for hospital staff/garbage disposal representative, I could look up at people who apply and people who did it succesfull in the past to find the people who are most qualified.
If I find that one gender is more represented than the other, this is bias, right? When I adjust this, I will then misrepresent the percentage of people who apply.
Don't fight for your country, if your country does not fight for you.
Q: What do you call a Black test-tube baby.
A: Janitor In A Drum.
What this bill proposes is to replace the real, actual bias with another, artificial bias that is more desirable / politically correct / whatever.
Not saying this is a bad thing - combating centuries-old entrenched preconceived ideas is probably a good thing more often than not - but please stop saying we're *removing* bias.
"A door is what a dog is perpetually on the wrong side of" - Ogden Nash
Detects real crime by real people in inner city areas.
Data tracks to decades of FBI stats.
Its not bias to have a computer create a database of crime.
To detect the use of shared/fake ID by illegal immigrants.
Domestic spying is now "Benign Information Gathering"
This strikes me as another attempt by clueless pols to legislate fairness. Not that legislating fairness is wrong, however in this case, it is more "Well golly, tech companies can do anything, we'll make them do this!!!" Ya, and they wouldn't find a way to game and unworkable mandate, eh?
Giant companies like Microsoft, Twitter, Facebook, Google Search and YouTube all have been claiming they can't can't be bias because moderation, trends, and recommendations are not done manually but with automated systems. Finally Congress is learning just being an automated algorithm doesn't make something impartial. Look at Microsoft's twitter bot that was made to be racist after just one day because it's algorithm responded to inputs. Not just that. It these algorithms often take into account flags by users which pretty much means if a bunch of users flag posts that have a specific word in them, the algorithm will start deciding that the word is inflammatory even if it is mundane. Say a lot of people starts flagging a word or hash tag like "#learntocode" then that word can cause you to become banned or get moderated.. that actually happened with the mundane phrase "#learntocode". Data science uses training data that often contains factors like race, sex, income, and education level that when included cause an algorithm to train to moderate or recommend people differently based on your assigned group. This in itself is not that horrible but it does create a divide among groups and could even be used to treat people differently based on a protected class which is a big no-no. It's about time regulation catches up and recognizes algorithms can and often are bias.
It would be nearly impossible with AI. With learning AI currently, while they can learn, we don't know exactly how they came to those conclusions in many cases. One would have to spend considerable time and effort designing tests to determine bias, as simply checking code or algorithms wouldn't be possible.
If we're using neutral data as an input and the system comes to it's own conclusions...doesn't that say something about the data set? Shouldn't we try to understand why the algorithm came to that conclusion instead of immediately jumping to "check your privilege" ?
Mod me down with all of your hatred and your journey towards the dark side will be complete!
... is it biased?
E.g. an algorithm that is supposed to recognize humans will probably do worse for humans in a Mickey Mouse costume. Should it now be trained to recognize those dressed up as MM equally well, although it is a very rare case? Or should it be trained so it deals best with those situations that it will probably encounter more often, and that are thus more relevant.
I.e. should the algorithm be trained with a representative sample (for the country it is to be used in), or should every ethnicity, every height, every kind of clothing in any combination be equally weighted, including bolivian basketball players in burkas?
"By the way if anyone here is in advertising or marketing... kill yourself." -- Bill Hicks
The vast majority of users serviced by AI systems have expectations by which they judge the service. When humans serve those users, good customer service usually dictates meeting those expectations. The expectations are largely driven by the microcultural background of the customer. At other times, they are expressed in the phrasing of the question, especially in context with the microcultural background. When a human has a great sense of a customer's expectations and utilizes it to meet those expectations more than other humans, they are considered talented at serving customers. This talented human service person is not expressing bias, they are responding to suspected bias in order to better meet customers' expectations.
I've seen many cases where people have lambasted a system for bias when they asked a leading question and successfully led the system. That is a user bias, not a system bias, unless the system gives the same answer when the question is asked in a manner or from an individual (assuming the system is good enough to take that into account) that leans the other way - it is good customer service.
There absolutely is bad bias in systems. In general, systems will improve in helping their users if bad bias is removed. So companies usually want to remove it when they recognize it. For example, facial recognition systems are improved when they are able to recognize faces from all racial backgrounds well. They are harder to create, not because of the training set, but because the job is harder to perform. Most people are far more capable of recognizing faces of people from their cultures than others because there are real differences in the parameters that are best used to determine uniqueness.
But AI systems cannot match humans in providing service if they are banned from recognizing human bias and utilizing that to provide expected responses.
Laws tend to be a broad sword. It is very likely that a law of this nature will not walk the fine lines that it would have to in order to serve us well.
Watch Weapons of Math Destruction by Cathy O'Neil to see how algorithms have bias, and the results can also be used in various ways. If this law addresses some of that, then it is a positive change.
2bits.com, Inc: Drupal, WordPress, and LAMP performance tuning.
This may be an unpopular opinion, but I'm not sure where bias even enters into it.
Isn't the point of any algorithm to make a choice? Like "this face matches sample A to a larger degree than it matches any other sample", in the case of a facial recognition algorithm? If so, then shouldn't the one and only criteria be "does this algorithm, as it is programmed, return the most correct answer with the highest probability and lowest probability of false positives?"
Now, if the data/choices the algorithm uses/makes causes a HUMAN to act in a discriminatory manner, shouldn't the bias be considered on the part of that human?
Example: Algorithm says "based on my data, Asians are 80% less likely to buy anything in this store than other demographics are", and human decides to bar Asians from their store or to have their sales staff pay less attention to Asian customers who enter the store, I blame the human that made that decision.
Counterexample: Algorithm says "based on my data, Asians are 80% less likely to buy anything in this store than other demographics are", and there's an automated system that uses this analysis as a basis to bar Asians from the store, or treat them differently once they're in the store, then that's a systematic issue and is indeed a problem with the automated system - but not necessarily with the original analysis algorithm.
function GetBias(time,bias)
{
return (time / ((((1.0/bias) - 2.0)*(1.0 - time))+1.0));
}
Which has nothing to do with bias. Bias, in this context, is unwarranted assumptions. Men are on average stronger and taller than women, but a system which, say, ranks potential firefighter applicants using their gender as a factor instead of looking at their performance in the actual job is biased.
Sorry, but you haven't been listening to the left if you think the test for bias is about assumptions. Equal outcomes is very strongly being pushed as the measure for bias.
Do you have more males than females going into trades? That must mean a bias against females exists in the trades.
Do you have more Asians getting into STEM? That must mean a bias in favor of Asians in STEM.
Language is being redefined and weaponized to push people's agendas(I know, it always has). Today we have the definitions of equality, racism, bias, violence, assault and others being changed to better fit agendas. Racism being the grossest example because of it's importance and power. I grew up understanding racism to be discrimination based on race. Today though the push is on to redefine racism to be a combination of discrimination AND power. This turning the convenient trick then that 'whites' have all the power, so now only they can be racist, by definition.
the congress of the United States can not work on any law or regulation until the 12 appropriation bills that make up the budget of the United States of America are passed by congress and signed by the president.
;)
FYI the US government rarely does a budget anymore they are to busy doing useless political investigations and passing things that are just a waste of tax payers time and money.
One plus is the useless ness of government is bi partisan. The US government is now made up of DEMs and GOPers who think their government job is to do the bidding of their parties
Maybe administration/congress/president and their staffs should not get any pay checks either until they DO THEIR MAIN JOB!!!! a budget!!
Just my 2 cents
Does Slashdot have algorithmic moderation? If they can do that then why can't they support "smart" punctuation?
I think the senators should just skip to the chase and give us the output that the algorithm should produce. Problem solved.
I'll bet if you completely exclude specific criteria from being a factor for consideration because of some undesired bias that might occur around such information, the resulting decisions may still show bias.
File under 'M' for 'Manic ranting'
Will someone please punch these stupid n iggers in the face? fucking faggots
I hope George Washington's descendants beat this shit out of these semen lovers
Urban legend, I'm afraid.
File under 'M' for 'Manic ranting'
Same for almost anything. Skin colour rarely matters, and given enough more direct data on factors that do matter, skin colour will have no predictive value, so the algorithm will ignore it.
Exactly the opposite will happen. .let's pick example that won't brush POC people wrong as it disadvantages men, car insurance.
E.g
You let AI know the gender - it will figure men are more likely to get into crash.
If you stop feeding gender to AI, it will figure people named John are more likely to be causing trouble, than people named Julie.
You will have to play a long cat and mouse game cutting off information sources for the AI to a point of it becoming useless.
Just think about the whole "discrimination" issue.
Black and Latinos are poorer than White/Jewish/Asian americans.
Hence AI, created to optimize the strategy, figured it makes more sense to target the latter with ads.
This is an attempt to enforce equality of outcome
your thin skin doesn't make me a troll