Generalaztions: Occurs over time, Can be positive or negative, Leads to change and Can be planned or spontaneous Student-led Research Results We decided to interview Bobby Azarian who is a cognitive neuroscientist and is a science writer who has had works published in The Atlantic,The New York Times, BBC, Scientific American, Slate, The Daily Beast, and the Huffington Post. Question: Answer: As artificial intelligence develops over time, is there a capability that they will inherit qualities like racism or discrimination based on race? What changes should be made to ensure that A.I. can adapt over time to be humane, objective, and morally or ethically sound? Does A.I. inherit qualities through its environment or its internal programming? “To address your first question about whether AI will inherit our racial biases and prejudices, the answer is, they will by default if we don't address this issue directly. Essentially, if you recognize that advanced AI often learns from "reading" tons and tons of human generated content, like that which is on the Internet, you will see that it is inevitable. Since the Internet contains documents and articles written by humans, who have inherent biases, they tend to inherit these biases, and this shows when they are making judgments or associations. I wrote an article just weeks ago about this exact issue, called, "Study finds AI systems exhibit human-like prejudices". Unfortunately our biases are embedded in our language and cultural history. Special care must be taken by the programmers of these systems to ensure that these biases are corrected for when the AI system is making decisions.” Is the capability of A.I. consistent across cultures around the world or do cultural or racial factors influence how an A.I. thinks and acts? “Until we have Strong AI, which should have morals and ethics that emerge naturally, we will have to be very meticulous in the way we write programs to ensure ethical and moral behavior. We won't be really creating moral machines, just machines that follow our instructions which ideally have built in rules that result in the appropriate moral and ethical decisions. For example, if a machine is used as a weapon on the battlefield, it will not be an intentional agent, it will blindly follow it's programming, so the responsibility of it's moral behaviors fall on the humans writing its software. A robot soldier should be programmed to distinguish between a civilian and a threat, for example, and to treat them appropriately. But these decisions are a result of its programming — it has no true intent. To ensure that moral and ethical behaving AI software is written, there should be non-partisan and diverse ethics committees in place that oversee these issues. Complex moral dilemmas that will affect us will inevitably arise. It is not as easy as making machines that will know how to act humanely or morally in every new situation, unfortunately. Since they aren't truly aware or self-aware, they are just making decisions based on predetermined algorithms, most new cases will have to be planned for in advance, when the program is being written. Until we know how to make AI where true intention and subjectivity emerges, the ethics and morals must be programmed as computational rules. “ “You ask if AI inherits properties through the environment or its internal programming, which is an astute question. It is a combination of the software's code and the information it is processing from its environment. Input is converted to output via specific algorithms, so the computer learns from the environment but its "learning program" is created in advance. What's neat about modern AI is that the neural networks they operate according to are modeled after the brain, and therefore have "plasticity". AI learns as humans learn in that associations are reinforced by what it learns from input from the external environment. In this way, it can 'learn' our negative stereotypes. But it does not understand or experience anything subjectively.” Changes over time - A.I has changed from the past and it will continue to improve in the future.
Ethics- A.I can bring both good things and bad things to occur
Contributions- AI can really change the way of life for many people
Language of the Discipline- AI is the theory and development of machines that can do human tasks without the help of humans
REFERENCES PAGE Lynn, S. (2017, March 16). Artificial Intelligence as a Weapon for Hate and Racism. Retrieved from http://www.blackenterprise.com/technology/artificial-intelligence-weapon-hate-racism/ Resnick, B. (n.d.). How artificial intelligence learns to be racist. Retrieved April 17, 2017, from https://www.vox.com/science-and-health/2017/4/17/15322378/how-artificial-intelligence-learns-how-to-be-racist Levin, S. (2016, September 8). A beauty contest was judged by AI and the robots didn’t like dark skin. Retrieved from https://www.theguardian.com/technology/2016/sep/08/artificial-intelligence-beauty-contest-doesnt-like-black-people Devlin, H. (2016, December 19) Discrimination by algorithm: scientists devise test to detect AI bias. Retrieved from https://www.theguardian.com/technology/2016/dec/19/discrimination-by-algorithm-scientists-devise-test-to-detect-ai-bias Ending question - Where do I fit in? What can I contribute as a future computer engineer to ensure that AI doesn’t continue human faults of bias and discrimination?
0 Comments
Leave a Reply. |
PDF Of TLP Essay.Archives
May 2017
Categories |