Digital Age Flashcards
what is digital surveillance used for
counter terrorism, propaganda, counterespionage, data on politics, policing, capitalism, public health, data tracking and harvesting (attention economy)
moral objections of digital surveillance
it deprives people’s privacy, as they either harms people’s interest or violate one’s moral right to privacy
what are the two theories of privacy
control account and access account
what is the control account
if another person has possession of your information they breach your privacy
what is access account
“information must be accessed to have an actual violation of privacy…”
what benefits does algorithmic-decision making have
improve accuracy if predictions, minimise human error, minimise human labour required, minimise cost of service, minimise morally unjustified bias
what are the issues with machine learning algorithms
secrecy, filter bubbles and bigoted data
why is algorithms secrecy a problem
although is secrecy is due to proprietary reasons and as because these processes are too complex for human comprehension. it reduces transparency, making humans lack trust and ability to understand decisions. Leaving them lacking autonomy and understanding.
how can machine learning algorithms be bigoted (proxy problem)
these run off of previous data, which is often discriminatory and despite input data having no such characteristics (e.g race, gender, sexuality, etc.), other data can have proxy characteristics which it can draw conclusions from (proxy problem)
what are filter bubbles
how search engines tailor data to an individuals interests
what are the issues with filter bubbles
they isolate individuals within certain information, cut off opposing evidence, creating a false impression of what other’s believe overall creating (political) polarisation
what is the solution and rebuttals to the proxy problem
curate training data (can sacrifice accuracy) or address social problems & injustices (requires a lot of time and effort)
Difference between narrow AI and AGI
narrow AI is domain specific meaning it only performs specific tasks, Artificial General Intelligence is capable of learning across many domains of ordinary human beings with minimum of human intelligence
what is superintelligent AI
across most domains it learns and executes cognitive tasks greater than most intelligent humans
reasons for superintelligent AI
Breakthroughs in science, medicine, and engineering
Cheaper and more reliable labour for manufacturing, agriculture and human service
Entertainment
Concerns of superintelligent AI
mass unemployment
doomsday scenarios
playing god
moral concerns
reasons why it could be wrong for people to think robots are like humans
unhealthy, inefficient, misassign responsibility to robots
what is the intelligence explosion (closely grey goo problem)
if AI begin creating AI which are better than themselves, AI can easily take over the world with their superior intellect and capabilities
another cause of doomsday scenarios other than the intelligence explosion
persevere instantiation; fulfils goals through undesired means
why may the intelligence explosion occur
either AGI is brought to us from another location or we invent AGI that change their goals
issue of programming morality into superintelligent AI
the laws may not be precise enough, can be worked around
issue of capacity control methods for superintelligent AI
they can manipulate naive people into helping them
ways to avoid dystopia (intelligence explosion)
have select motivations and moral guidelines, limit their controlling methods
what is described when referring to violent video games and implications to gaming
it includes violent, realistic, immoral crimes. Is it wrong or morally problematic, therefore should there be a ban or at least restrictions
