Videos Flashcards
The Weaning Period in Self-Driving Car Introduction
the period of time when cars will gradually become more autonomous and intelligent with an increasingly number of features being added to the cars, making them increasingly smart
Sudden Human Input Need Problem
when the car has been driving itself but sudddenly requires human input to avoid a bad outcome, such as a crash;
the human is unlikely to be able to react appropriately within the few seconds that he/she has to react
Under and Over estimation Autonomous Cars’ Capabilities
It is likely to be common among people both to overestimate and underestimate the capabilities of self-driving cars.
In the first case, overestimation of the car’s capabilities may lead to users trusting the car when they should not, and thus getting into potentially serious accidents.
In the second case, understimation may lead to users taking control of the car in situations that they perceive as dangerous (if possible) and save their lives or worsen the situation.
When will be ready for self-driving cars?
When society, technology and individuals will all be ready. These are the three important aspects.
Percentage of Employed Adults In Last Century
has risen despite automation becoming increasingly prevalent
Why are there still so many jobs?
- O-Ring Principle:
- concept introduced in the context of understanding the Challenger space shuttle disaster in 1986
- serves as a metaphor for explaining how the success or failure of complex systems can depend on the reliability of even the smallest and seemingly insignificant components
- in the case of the Challenger disaster, the O-ring seals, which were designed to prevent the leakage of hot gases from the solid rocket boosters, failed due to cold temperatures on the day of the launch
- the more accurately parts of a task are carried out, the more importance there is on the remaning parts to be carried out effectively too
PEOPLE ARE THE O-RINGS. Since every other, automised part of the tasks works great, there is a lot of importance on us to work well too.
- Never Get Enough Principle:
- regardless of how much we advance technologically, we always want more
- that is, innovation creates more innovation, and thus technological inventions create more and more jobs and industries
- invention is the mother of necessity
- the more we create, the more we want
- material abundance has never eliminated perceived scarcity
What’s important in terms of automation?
Not the losing of overall jobs but:
How we use the resources that we gain from automation - Norway vs Saudi Arabia example in regards to oil
The quality of the new jobs being created (many new low-level jobs are being created at the expense of mid-level jobs) and how people are prepared for those new jobs (also many high-level jobs are being created for which people need to be prepared)
What’s the risk of job polarisation?
Society will become far more divided by the removal of the middle class, with a group of many high-skilled high-paid individuals and another of low-skilled low-paid individuals making up most of society.
A solution to this polarisation of society in terms of skill and thus living standards
Education - investing more in education to give all people the opportunity to compete for the high-level jobs
Investment in automisation of low-level jobs
What does everything that we have achieved as a species depend on?
On some few minor changes that our brain has undergone since we we were apes
Why does AI have the potential to overtake human intelligence?
no physical limitations in terms of size
information can be transferred much quicker electronically than biologically
technological growth can occur much quicker than evolution
What could be the preferences of an superintelligent AI?
a superintelligence is able to compute things exceptionally quickly, but there is no inherent link between a superintelligence existing and it having preferences or goals like humans do
superintelligence will like only want to solve a problem that is provided, but not have any other preferences; humans must thus ensure that all of their interests are incorporated into the problem that the AI is given, as otherwise the superintelligence may act in ways that humans do not want it to
goals must be perfectly specified
How to ensure that a superintelligence does not act against humanity?
We should not attempt to keep it locked in a bottle and prevent its ability to harm humanity
Rather, we should ensure that the AI is developed in such a way that it shares our values and would not harm us even if it could.
To train this AI so that it protects human interests, we should use another AI that over time learns what humans value and do not, and feed this information into our AI.
Solution Before or After Developing AI
BEFORE, but may not be entirely possible since the problem may need to be fully defined for the solution to be carried out
The Media Equation
findings in the field of human-computer interaction, particularly in the context of how people tend to apply social and interpersonal rules and expectations to interactions with computers and media
individuals tend to treat computers and media as if they were social beings, attributing human-like qualities and social responses to them
people often respond to technology, whether it’s a computer interface, a robot, or virtual characters, with behaviors and expectations similar to how they would interact with other humans
