Task 1 Flashcards

1
Q

cognitive science

A

suggests that we have mental procedures that operate on mental representations to produce action and thought

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Artificial Intelligence (AI)

A

intelligence exhibited by machines / softwares

-> act in world by perceiving environment, interpret data reason on knowledge and decide on action

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

strong AI

A

creating a human-like or non-human-like conscious computer (not established yet)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

weak Ai

A

subfields like search engines -> seem to be intelligent but are actually not

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

technical singularity

A

if AI is sufficiently intelligient it might be able to improve on its own which could eventually lead do an exponential increase and drastic suppression of humans

singularity principle : point where technological evolution is so fast that humans cnanot predict or understand what will happen: blackbox AI

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

cognitive modelling

A

development of psychological theories that can be translated into computer programs (comparing in order to imrpove)

most important advantage: computers force you to concisely describe all terms important to your theory (computer needs transparency)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

there are differnet descriptional and explanatory levels when attempting to understand human behavior

A

1- computational level
2- algorithmic/representational level
3-physical level

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

computational level

A

focus lies on the goal of a process (WHAT DOES THE SYSTEM DO)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

algorithmic/representational level

A

How can the process be executed, how input and output should be represented, and what algorithm can transform input into output. (What steps does the system go through?).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

the physical level

A

representation and algorithm are physically realised (In what way are the steps the system goes through implemented?)
➔ This is the level at which neurobiologists describe the world.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

what is the big advantage of Ai

A

prediction

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

machine learning and prediction

A

computer learn from past expereiences and predicitions -> they are able to form correlations
- become cheaper and easier accessible

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

who judges what outcome is best?

A

with learning more accurate predictions can be made

in cases where whole decisions can be clearly defined with an alogrithm (like autonomous driving) coputers are expected to replace humans

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

employing prediction machines // Super Clinican

A
  • need for machines that not only make predicitions but also make decisions based on them

Super clinician : AIs could make a super clinician by providing a simulated practtitioner with capabilities byond humans (eg advanced sensory techniques)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Prediction ≠ automation

A

tasks are more than just predicitions, they also require data collection, judgment and action which are NOT part of machine learning
- currently technology is improving quickly in prediction

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

judgement

A

machines will overcome humans when it comes to predicition but not when it comes to judgement

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

managing may require a new set of talents and expertise

A

Many manager tasks are predictive like hiring an employer. As AI become more advanced, this part of the tasks will become less important for humans, while the judgement part about managing tasks will gain importance.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

prediction

A

the ability to take information you have and genrate information you did not previously have

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

predicitions of Ai can be dividied into four types

A

1 timelines and outcomes
2 scenarios
3 plans
4 issues and metastatements

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

timelines and outcomes

A

predictions: predictions about when AI milestones will be achieved

21
Q

scenarios

A

conditional predictions, claiming that if the conditions of the scenario are met,
then certain types of outcomes will follow

22
Q

plans

A

special type of conditional prediction; if someone decides to
implement a specific plan, then they will be successful in achieving a particular goal. (i.e.: We
can build an AI by scanning a human brain and simulating the scan on a computer)

23
Q

issues and metastatments

A

relevant problems with approaches to AI and metastatements about the field (i.e.: AIs will have certain potentially dangerous behaviours)

24
Q

there are six different prediction methods

ways of arrivinf at different predicitions

A
1 causal models
2 non causal models
3 the outside view 
4 philosophical arguments
5 expert authority
6 non expert authority
25
Q

causal models

A

when given facts about a situation, a conclusion is reached about what the ultimate state will be
-> however lack understanding of exact underlying causes

26
Q

non causal models

A

even without understanding what influences what > one can extrapolate trends into the future

27
Q

the outside view

A

predicting by gathering specific examples that follow the same trend

28
Q

philosophical arguments

A

some are impossibility statements, some point out problems that
need to be resolved and highlight approaches to do so

29
Q

expert authority

A

predictions relying on the status of the predictor

30
Q

non expert authority

A

opinion of journalists, authors, …

31
Q

methods and tools to clarify assumptions

A

1 model testing and counterfactual resiliency

-> giving model a counterfactual resiliency check

2 more uncertainty

32
Q

empirical evidence

A

ehen seperating true and false predictions this must always be with the use of empirical evidence and the use of the scientific method

problem :
empiricial evidence is basically non-existent in AI field -> we have to put more trust into expert opinion

33
Q

expert opinion

A

reliance on expert opinion is unavailable in field of Ai -> therefor need to highlight where there is need to rely on expert opinion

34
Q

Grind versus insight

A

some claim that AI will result from grind (hard work and money), while others claim that it will result from insight (new unexpected ideas). We are good at predicting grind while predicting insight is much more difficult and risky.

35
Q

non expert opinion

A

there is no reason to trust it in the first case -> however can always refer to good predictions

36
Q

Maes Garreau Law

A

Ai experts predict Ai happening towards the end of their own lifetime
- no evidence

37
Q

twenty years to AI / Moores Law

A

AI is -/+ 10 to 15 years in future so that the predictot can gain credit for working on somethig that will be of relevance but without any possibility that the prediction could be shown to be false in their current career

-> Moores Law: says that the complexity in computer circuits doubles every two years

38
Q

greys legal model of decision making

A

law is rationally determinate: judges either decide cases deductively, by subsuming facts under formal legal rules or use more complex legal reasoning than deduction whenever legal rules are insufficient to warrant a particular outcome

39
Q

Leiter’s legal realists’ theorisation

A

judges primarily decide cases by responding to the stimulus of the facts of the case

40
Q

overestimating and underestimating of predicition

A

= mistaken predictions lead to fears of things that are not going to happen

  • effect of technology in short run is overestimated and in long run underestimated
41
Q

Imagining MAgic

A

Clarke’s three laws:

(1) When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong.
(2) The only way of discovering the limits of the possible is to venture a little way past them into the impossible.
(3) Any sufficiently advanced technology is indistinguishable from magic.

42
Q

Performance vs Competence

A

When people hear that some AI system has performed some task, they generalize from that performance to a competence that a person performing the same task could be expected to have.
➔ Apply that generalization to the robot/AI system

43
Q

Searles Chinese Room

A

thought experiment showing that a computer program demonstarting an input-output performance is NOT a strong AI -> simulated intelligence

44
Q

supervised learning method

A

system gets inout and output, hope it learns how to generalize from examples and how to interpret.

45
Q

deep learning systems

A

learning method based on several layers between input and output, allows to learn overall relation is successive steps -> inspired by neural systems

46
Q

CRUM

A

cenral hypothesis that thinking can be best understood in terms of representational structures and computatinal procedures

Mental Representations : Data structures
COmputational Procedures : Algorithms
Thinking : Running program

47
Q

unsupervised learning

A

gets input, involves analysis of unlabelled data, no correct answer given, learns by itself how to eg categorise -> based on pattern, structures and not preprograms

48
Q

foraging models

A

one success attracts others to follow same path and therefore also have success, the more joint, the more evidence for success

49
Q

espert systems

A

systems designed to incorporate knowledge and ability of a human expert in a domain these systems can identify patterns, trends and menaing from complex data