Philosophical Foundations Flashcards
AI is called strong, when it outperforms humans in intelligent tasks, e.g. chess.
False
What is the Turing Test based on?
The imitation game = TRUE
Turing Algorithm
Questions to a machine
Chess
What is a modern day version of the Turing Test?
the Mind-Body Problem
the Imitation Game
the Loebner Competition = TRUE
the OpenAI GPT3
What is TRUE about CAPTCHAs?
Are designed to prevent humans from accessing something in the web = FALSE
Weak AI cannot solve the task in most cases = TRUE
Are basically adapted Turing Tests = TRUE
Are easy tasks for humans = TRUE
A weak AI is an universally intelligent agent.
False
The general view is that there are differences between strong AI and weak AI. Which point is not one of them ?
- Strong AI: build a universally intelligent agent encompasses the full range of human cognitive abilities (false)
- Strong AI: build a universally intelligent agent, who can pass the Turing test (false)
- Weak AI: build agents that act rationally, accomplish specific problem solving or reasoning tasks (e.g., playing chess, mowing the lawn, solve a Sudoku puzzle, …) but are not universally intelligent. (false)
- Weak AI: build agents that always do the same specific things (e.g. Digital clocks tell the time, memory sticks stored data,…) TRUE
BE careful of the question itself. It has a negation!
The Chinese Room opposes the idea of strong AI.
True
A weak AI can only do ”weak” tasks that are not difficult but would annoy humans to do it.
False
Which statements are true about Searle’s Chinese Room:
Intentionality is Searle’s word for the difference between ”true” & ”simulated” intelligence = TRUE
The Chinese Room would NOT pass the Turing Test = FALSE
The man in the room knows Chinese = FALSE
Nothing in the room knows Chinese = TRUE
One of its purposes is to show that the Turing test is inadequate = TRUE
The Room shows that the Turing Test can be used to discover ”true” intelligence = FALSE
What is true about the person in the thought experiment ”Searle’s Chinese Room”?
The man speaks Chinese and English = FALSE
The man does not speak any language = FALSE
The man does not speak Chinese = ** TRUE**
The man speaks only Chinese = FALSE
Which sentences about the ”Chinese Room Argument” are correct?
However, the ”thing” in the room learned Chinese with English instructions on how to interpret Chinese symbols = FALSE
However, nothing in the room knows Chinese (the symbol manipulator just follows rules that are written in plain English). = TRUE
Just as the room does not know any Chinese, a computer that passes the Turing Test can not really think. The Turing Test cannot be used to discover ’true’ intelligence, it can only be used to discover ’simulated’ intelligence. = TRUE
The Chinese Room would pass a Turing Test (from the outside, you have to assume that there is a fluent Chinese speaker hidden in the room). = TRUE
The Chinese Room would never pass a Turing Test (from the outside, you have to assume that there is a fluent Chinese speaker hidden in the room). = FALSE
Just as the room does learned Chinese, a computer that passes the Turing Test can really think. The Turing Test can be used to discover ’true’ intelligence, and it can also be used to discover ’simulated’ intelligence. = FALSE
In 1970 Newell & Simon hypothesized: ”A physical symbol system has the necessary and sufficient
means for intelligent action.
True
Map the following components of Searle’s Chinese Room argument:
Rule Book = Program
English speaking man inside the room = CPU (brain)
Baskets with symbols = Memory
Letter slot of the door = Input/Output Units
In Searle’s Chinese Room Argument: Who does certainly not speak Chinese?
The person who executes the rules inside the room.
Rene Descartes proposed that animals are machines.
True
The Loebner Competition:
Modern version of the Turing Test
Multiple judges rank-order multiple humans and multiple computers programs from most likely to be human to least likely to be human
DOES NOT ADVANCE the state-of-the-art in AI
most programs apply cheap tricks like ELIZA
test scenarios are restrictive in order to be conclusice (ONE TOPIC only)
5time winner is MITSUKU/KUKI, which is based on ALice (successor of Eliza)
Since 2019:
no human competotors, only bots
public feedback instead of judges
Turing Test backwards
we give computer questions that can only be answered by humans. The reason behind is to ensure that WEB BOTS do not register a million free e-mail accounts etc.
WEAK AI
agents that act rationally
accomplish specific problem solving/reasoning task
NOT UNIVERSALLY INTELLIGENT
STRONG AI
universally intelligent agents
have a full range of human cognitive abilities
can pass the Turing Test
AI-complete problem:
informal/half-joking, denoting problems since it is believed that universal intelligence is necessary
Is the PSSH the heart of the strong AI?
True
Grandmother Cell (“physical symbol”)
Discusses whether the representation of complex objects in the brain is sparse (few neurons) or distributed (activation patterns of many neurons).
The AI debate
1) Herbert Simon: Dig. computers will be the world’s best chess champion
2) Hubert Dreyfus: Human intelligence is more than manipulation of symbols
3) John Searle: Opposed idea of a strong AI that machines can think; Chinese Room Experiment
Brain-in-a-vet Argument
Your brain is removed and put in a vat
all neural connections are replaced by bluetooth transmitters
Brain-Prosthesis Experiment
- assumably neurons can be replaced with electrical circuitry
- neuron by neuron the entire brain
- then the process is reverted
- after operation you are physically the same person
- nobody can tell from the outside whether you are self-consious or not
Monism/Materialism
mental states are brain states
“Brains cause Minds” by Serale
Dualism
mind and matter are separate
mental stuff are non-physical
formulated by Descartes (17th cent.)
MIND = Self, Personality, Soul
Physical Symbol Systems
It’s an idea about how computers (or even humans) might “think” or process information using symbols
cosists of a set of entities (symbols)
which are physical patters that occur as components of another type of entity called an expression (or symbol structure)
A symbol system possesses a number of simple processes that create, modify, copy and destoy symbols
Example: assembling ikea furniture
Turing started the philosophical debate by formulating 9 possible objections to intelligent machines:
- Theological
-> He states that if God wishes, he could give a soul to a machine and theological theories do not impress him - Head-in-the-sand
-> Not substatial to require refutaion (calling sth false) - Mathematical (Gödel’s, Lucas & Penrose’s Theorem)
-> Humans also make mistakes - Consciousness Objection
-> Cannot be sure that everyone has the ability to think & feel
-> The example here is writing a sonnet - Objections from various disabilities
-> Something is stated within X = {human characteristics, e.g., fall in love, make mistakes, tell right from wrong etc.}
-> Does not see an issue to implement them to machines - Lady Lovelace’s Objection
-> Computers can only do what they are programmed to
-> A mechanical chess-player cannot outplay its master, but can replicate the thoughts of its programmer - Continuity of the nervous system
-> A nervous system cannot be built a computer
-> But can be approximated well wnough
-> 2020 Nobel Prize Winner, Roger Penrose stated that consciousness is due to quantur gravity - Argument from informality of behaviour
-> If a human being had a set of rules to regulate its life, be would be no better than a machine
-> We follow laws of physics and there could be higher laws of behavious - Argument from Extra-Sensory Perception
-> Telepathy-proof room, otherwise the judge cannot distiguish due to ESP