“Artificial intelligence… could fundamentally change the world while contributing to greater racial bias and exclusion,” Heather Kelly reports for CNN. “Much of the focus on any downsides of artificial intelligence has been on things like crashing self-driving cars and the rise of machines that kill… But many of the researchers behind this technology say it could pose a greater threat to society by adversely impacting the poor, the disenfranchised, and people of color.”
“One part of AI is machine learning, in which a system analyzes massive amounts of data to make decisions and recognize patterns on its own. And that data must be carefully considered so that it doesn’t reflect or contribute to existing biases,” Kelly reports. “A recent study by the M.I.T. Media Lab found facial recognition software has trouble identifying women of color. Tests by The Washington Post found that accents often trip up smart speakers like Alexa. And an investigation by ProPublica revealed that software used to sentence criminals is biased against black Americans. Addressing these issues will grow increasingly urgent as things like facial recognition software become more prevalent in law enforcement, border security, and even hiring.”
“Many of those who gathered at last week’s discussion, ‘AI Summit – Designing a Future for All,’ said new industry standards, a code of conduct, greater diversity among the engineers and computer scientists developing AI, and even regulation would go a long way toward minimizing these biases,” Kelly reports. “Technical approaches can help, too. The Fairness Tool, developed by Accenture, scours data sets to find any biases and correct problematic models.”
Read more in the full article here.
MacDailyNews Take: Microsoft’s Tay chatbot was the canary in the coalmine.
SEE ALSO:
Apple tells U.S. Senate Democrat Al Franken they trained iPhone X’s Face ID system to not be racist, ageist, or sexist – October 17, 2017
Microsoft’s ‘Tay’ chatbot turns into a foul-mouthed racist nazi – March 25, 2016
“One part of AI is machine learning, in which a system analyzes massive amounts of data to make decisions and recognize patterns on its own. And that data must be carefully considered so that it doesn’t reflect or contribute to existing biases,”
So conclusions based on massive amounts of data shouldn’t be trusted? Political correctness over fact based analysis?
Not if the “massive amounts of data” is coming exclusively from upper-middle class white men with college educations and the “received accent” for native-born English speakers. The data will train the software to regard other groups as not just a departure from the statistical median of the control group, but as abnormal.
The federal court sentencing software that suggested longer sentences for African-American defendants shows just how dangerous that can be. Women who have a heart attack have been frequently misdiagnosed because the diagnostic criteria were based on “massive amounts of data” from older white men. People have died because of exactly the problem identified here.
Personally, I’d like to see something closer to raw data. If indeed the software suggested longer terms for persons essentially identical save only for their race, then that’s a concern. But if race was only incidental, and the black person had a longer or more serious criminal history, somehow I’d tend to discount racial bias at that point.
(And yes, I’ve seen this kind of thing before where “racial bias” was claimed when there were obviously other reasons for the differing treatment of persons.)
“Not if the “massive amounts of data” is coming exclusively from upper-middle class white men”
Wow, just wow!
That is the most RACIST words you have ever posted. Then you repeated your racism again!
“departure from the statistical median” Your opinion was automatically disqualified because of racism and spare us the FAKE CONSERVATIVE lying posts. You are a PC liberal activist and everyone knows it …
I would have thought that ANYBODY would understand that if you train a machine learning program to distinguish between faces using thousands of white male examples, it is going to be terrible at telling one black woman from another. That isn’t racist. It is inevitable.
Likewise if you train voice recognition software using thousands of voices that sound like Los Angeles TV announcers. It will be terrible with ethnic accents. Not racist, because it will be just as bad with regional white accents.
If face recognition and voice commands are important parts of the user interface, non-standard faces and accents will be walled off on the wrong side of the digital divide.
I guess some people think that’s just fine. Tough if you grew up in West Virginia, I guess.
“I would have thought that ANYBODY would understand that if you train a machine learning program to distinguish between faces using thousands of white male examples”
No Mr. Condescending you were caught twice making racist comments. Show us the proof that “thousands of white male examples” excluding people of different races in testing. I’ll save you the trouble hapless liberal — it does not exist. Trying to walk back racist comments is all you have left. You FAILED and not fooling anyone …
Ar you just stupid or always insulted?
If you READ what Tx is writing, you would see no racism and another valid point having to do with what a group can wrongfully suffer if the data comes from a group ill-representing the group at hand. In this case, women with heart-problems. Oh wait, now Tx is racist and I am a male chauvinistic pig. Please just keep hating and stop writing here. And buy a Windows PC. Nice to empty out the hate in your heart. Windows is used to the hate, see….macaron.
I don’t know what you are and don’t care. USER made racists comments. To even entertain the idea that one group would design tech to favor them and exclude others in a liberal company where they value diversity is the most ridiculous tech accusation I have ever read. Technology is color blind and I sure it was thoroughly tested …
You want evidence, GeoB? Scroll up about ten inches. That is exactly what the linked article at the head of this thread is about. If you didn’t read it or don’t believe the sources it cites, don’t blame me.
Your comment is a pretty good example of a major problem in America today. Lots of people are convinced that the world is a zero-sum game where one party will inevitably rise if another party falls. If so, there is no need to do or say anything positive to build yourself up. That might require deep thinking or honest work. No, all that is necessary is to tear someone else down. GeoB, I have never seen you make a single argument or offer a single scrap of evidence FOR any of the positions you have taken at MDN. All you ever do, and always do, is engage in name-calling and insults AGAINST those with whom you disagree.
The same tendency is seen in the notion that all we have to do in order to make America great is to tear down every other nation. We cannot prosper unless they suffer. Domestically, Christians, whites, males and the native born don’t need to earn their status in society. All they have to do to win is tear down non-Christians, people of color, women, and immigrants. Their loss is somehow our gain.
So, as long as facial recognition gets you into your iPhone and Siri recognizes your voice, that’s all that matters in a zero-sum world. If other folks can’t use the technology, that is a win for you.
“GeoB, I have never seen you make a single argument or offer a single scrap of evidence FOR any of the positions you have taken at MDN. All you ever do, and always do, is engage in name-calling and insults AGAINST those with whom you disagree.”
The absolute statement racist and FAKE conservative suffers from hubris delusion.
“If other folks can’t use the technology, that is a win for you.”
Yet another clueless accusation with ZERO basis in FACT. Yes, you are pathetic and deflective…
Of course not! LOL! Anyone can pull “massive amounts of data” from the internet right now and build a horrible machine learning engine. Sounds like it’s more about “get your facts straight”. If you have a machine learning engine that looks at a picture of a white person and gives the result “trailer park” yeah, I’d say your data is screwed in some serious way.
Correct. Because when those large datasets are cherry picked to be rich profitable people, or affluent social groups only, then everyone outside the gated communities will be left in the cold with antiquated tech while premium services will be offered only to the insiders. Same as it always was.
Pretty sure this article is racist…
Fake.
The only thing hurting people of color these days, is the never ending mindless dribble from leftist morons insisting that anyone with darker pigment, should be considered a special case for society to cope with. These people are the most racist individuals in the political spectrum of western civilization and too damn stupid to see it.
This is why the Democratic Party is the single most racist institution in America past and present. From slavery to the terrorism of the Klan to the apartheid of Jim Crow to the modern racism of good intentions and lowered expectations, they remain the party from which racist ideology springs forth. The foundation upon which they base their conclusions and politics about “people of color” are all based on the PRESUMPTION OF INFERIORITY.
This is not hate or simple minded tribal separatism. This is the very definition of racism. The is the racism of PROGRESSIVES. This is the racism that came from the early progressive intellectuals, and it lingers and persists like a bad odor that has seeped into the fabric of leftist thinking. This is the racism of universities. This is the racism of white liberals. This is the racism that creates phrases like “The Negro Problem.” And yes… this is the racism that inspired and influenced Hitler himself.
This is why blacks are finally beginning to leave the party in droves and why the party is so determined to import illegal immigrants to take their places. Some one has to fill black slots in the voting booths.
The only difference between the Klan of the past and the Democrats of today is the big yellow happy face on their sheets and hoods. Yes, as they like to say, “they’ve changed.” They don’t look at us with bigoted hate and fear, they look upon us with pure racist condescension and pity.
The leftist mindset of sublime absurdity never ceases to disappoint. The ideology of collectivism and categorization of people based on barn yard level prejudices, still flourishes within their ranks. Everyone and everything is racist to them, except themselves.
Now artificial intelligence is racist. Why? Because data analysis is prone to objective truth.
Objective truth. I.e. KRYPTONITE to the left wing.
As our world shifts ever more to intelligent automation, it will be fascinating to see how these people force their racist views of others into algorithms, neural networks, and rule based decision systems.
“… by adversely impacting the poor, the disenfranchised, and people of color.”
But what about THE CHILDREN !!! Isn’t this usually the biggest fear tactic for any of this liberal, hysterical bullshit? Oh the humanity!
Machine learning is the branch of artificial intelligence concerned with giving machines the ability to learn and adapt based on what they have learned without being explicitly programmed.
Machine learning. The idea must be terrifying to some. A machine that learns and is not afraid to say what it has learned. Heck it might say anything! It doesn’t fear begin called racist!
It might say poverty is the result of criminal behavior, for instance. A machine might say, a 2 parent home with a mother and father is optimal for raising children.
Yeah, they’re going to have to nip the AI shit in the bud. Quickly.
“…New technology brings great promise, and as many problems. Smartphones put access to infinite knowledge in our pockets, but led to the rise of tech addiction. The social media platforms that connected billions of people were turned against democracy…”
Oh puh-lease. Smart phones were “turned against democracy.” They really, really, really do not get it. People didn’t vote against Hillary because of silly fake news. They voted against Hillary because they really did not want HER for president. Not because she is a woman, not because we hate women, not because of Gucifer 2.0, but simply, purely because we did not want HER for President, and by extension we are sick of them.
“Every social scientist knows that variables are interrelated,” said. “In the US for example, zip code [is] highly related to income, highly related to race. Profession [is] highly related to gender. Whether or not that’s the world you want to be in that is the world we are in.”
Run for the hills AI pioneers… the dweebs are coming for you.
3 million more people would disagree but let’s not let reality get in the way of your delusions.
“Objective truth. I.e. KRYPTONITE to the left wing.
As our world shifts ever more to intelligent automation, it will be fascinating to see how these people force their racist views of others into algorithms, neural networks, and rule based decision systems.”
Great point. Yes, “objective truth” is anathema to the Democratic Party. Just look to rigged algorithms on Facebook and Twitter. I am hopeful that won’t fly with AI and a reckoning is coming.
Your comments regarding minorities on the Democrat plantation as you pointed out are falling apart by the day. Just “walk away’ ..,
LOL.