Digitalization Examen: meerkeuze vragen, geen open vragen
Les 1: Digitalization and the challenges of AI
Stef Aupers
Three waves of digitalization
50 years: three converging waves of digitalization and their influence on
society
- Personal computer
- Internet and social media
- artificial intelligence
o AI: a social science perspective
First wave of digitalization: personal computer
IBM computers in the 1950s giant computers, the prediction was that
they needed 10 to 12 of these machines to make all kind of calculations
The democratic promise of the PC…
1950s: the hacker ethic (not hackers as we now them now, but people
who wanted to bring the power of the computer to the people)
1960s and 1970s -> hippies/ hackers in Silicon Valley
Goal: “bringing computer to the people”
1975: the first personal computer (apple)
1975 – 1985: the development, mass-production and commercialization
of the personal computer (the “apple computer”)
All the devices that we have now are based on the same principles
that they made during this time
,Second wave of digitalization – internet and social media
Internet = ecological system made out of interconnected computers
First used in libraries
Very imported developments in the process of digitilazation
The democratic promise of the internet
The promise that everyone could be connected through the internet
Web 1.0: 1990s – interconnected PCs, visit websites
Web 2.0: 2000s – social media platforms an User Generated Contend
(UGC)
with Facebook as the pioneer
Time magazine (2008) – Person of the year (YOU.) -> utopian climate:
everyone can be connected
Facebook (2004) -> blueprint social infrastructure internet
« making the world more open and connected” (mark Suckerberg, 2010)
From democratization to ‘survaillance capitalism’ (Zuboff?)
Third wave of digitalization = Artificial intelligence
Pioneers
John McCarthy -> mathematician/ scientist
Introduction concept DartMouth Conference 1955 – small/ modest
conference: first called it artificial intelligence (not a lot of money to
spend, not a lot of people, just talked a lot about the subject)
“…making a machine behave in ways that would be called intelligent if a
human were zo behaving”
,You can argue that machines are always an imitation of us, an extension to
help us do tasks/ work (e.g. cars to move from one place to another)
Here we have a new step:
machines are not only stronger and bigger in a physical sense but
also smarter and more intelligent in a cognitive sense
Basic forms of AI
Weak AI: Top Down AI:
AI that imitates our cognitive All instructions, no interpretations,
functions, but takes one specific it just does what it is supposed to
task. do e.g. in war, if you see someone
The computer is given one task who looks like this, shoot.
e.g. AI computer beated the best
chess player
Different from our brain, we can do
multiple things with our brain, this
cant be combined in one machine.
Strong AI: Bottom Up AI/ emergent AI:
The opportunity to speak to AI that is self-learning, evolutionary
understand sentences, make art, computing. We appropriate
play chess,… decisions, learn things an act
AI can do multiple tasks differently the next time. Acts
increasingly beter and beter
In 1950s they mostly used Weak AI and Top Down AI
Now AI is increasingly Bottom Up
The “holy grail” is Strong AI (utopia)
The classical philosophical debate on AI
Alan Turing (Mathematic: 1912-1954) VS John Searle (linguistic interested
in AI: 1932-…)
When can we say that the intelligence of the computer is more intelligent
the that of a human ?
When does a computer really has conscience?
The debate between these two people were
, Alan Turing
Mathematics, technology,…
The touring test: “How can AI be indeed be considered intelligent, and how
can we decide that?”
Experiment: one human being in front of a blinding
screen, on the other side of the screen there is a human,
and a computer. They start a conversation with the
computer or the human, if they don’t now after 8 min
whether they are taking to a computer or a human being,
the computer is considered intelligent.
The computer has to be programmed in a way that
when they don’t know the answers to a question they still can
answer with something smart.
It also works the other way around: we as humans are also just
machines, we use information to do better the next time.
John Searl
The Chinese Room experiment: “Is AI ‘really’ intelligent?”
Experiment: He wants to explain that computers can and
will never be intelligent
o The person in the room gets Chinese texts and uses it
to make new sentences, then gives it back to the
people outside of the box
o The person in the room has instructions, knows
exactly what to do, BUT has no idea what the text is
actually about.
Computers can do intelligent things, but aren’t intelligent
themselves, they cant actually understand what the information
means
Kevin Warwick (2011) - A third position in the debate
Wants to transform humas into cyborgs (half machine, half human)
Professor in UK, considers himself a cyborg
Writes books about AI
Les 1: Digitalization and the challenges of AI
Stef Aupers
Three waves of digitalization
50 years: three converging waves of digitalization and their influence on
society
- Personal computer
- Internet and social media
- artificial intelligence
o AI: a social science perspective
First wave of digitalization: personal computer
IBM computers in the 1950s giant computers, the prediction was that
they needed 10 to 12 of these machines to make all kind of calculations
The democratic promise of the PC…
1950s: the hacker ethic (not hackers as we now them now, but people
who wanted to bring the power of the computer to the people)
1960s and 1970s -> hippies/ hackers in Silicon Valley
Goal: “bringing computer to the people”
1975: the first personal computer (apple)
1975 – 1985: the development, mass-production and commercialization
of the personal computer (the “apple computer”)
All the devices that we have now are based on the same principles
that they made during this time
,Second wave of digitalization – internet and social media
Internet = ecological system made out of interconnected computers
First used in libraries
Very imported developments in the process of digitilazation
The democratic promise of the internet
The promise that everyone could be connected through the internet
Web 1.0: 1990s – interconnected PCs, visit websites
Web 2.0: 2000s – social media platforms an User Generated Contend
(UGC)
with Facebook as the pioneer
Time magazine (2008) – Person of the year (YOU.) -> utopian climate:
everyone can be connected
Facebook (2004) -> blueprint social infrastructure internet
« making the world more open and connected” (mark Suckerberg, 2010)
From democratization to ‘survaillance capitalism’ (Zuboff?)
Third wave of digitalization = Artificial intelligence
Pioneers
John McCarthy -> mathematician/ scientist
Introduction concept DartMouth Conference 1955 – small/ modest
conference: first called it artificial intelligence (not a lot of money to
spend, not a lot of people, just talked a lot about the subject)
“…making a machine behave in ways that would be called intelligent if a
human were zo behaving”
,You can argue that machines are always an imitation of us, an extension to
help us do tasks/ work (e.g. cars to move from one place to another)
Here we have a new step:
machines are not only stronger and bigger in a physical sense but
also smarter and more intelligent in a cognitive sense
Basic forms of AI
Weak AI: Top Down AI:
AI that imitates our cognitive All instructions, no interpretations,
functions, but takes one specific it just does what it is supposed to
task. do e.g. in war, if you see someone
The computer is given one task who looks like this, shoot.
e.g. AI computer beated the best
chess player
Different from our brain, we can do
multiple things with our brain, this
cant be combined in one machine.
Strong AI: Bottom Up AI/ emergent AI:
The opportunity to speak to AI that is self-learning, evolutionary
understand sentences, make art, computing. We appropriate
play chess,… decisions, learn things an act
AI can do multiple tasks differently the next time. Acts
increasingly beter and beter
In 1950s they mostly used Weak AI and Top Down AI
Now AI is increasingly Bottom Up
The “holy grail” is Strong AI (utopia)
The classical philosophical debate on AI
Alan Turing (Mathematic: 1912-1954) VS John Searle (linguistic interested
in AI: 1932-…)
When can we say that the intelligence of the computer is more intelligent
the that of a human ?
When does a computer really has conscience?
The debate between these two people were
, Alan Turing
Mathematics, technology,…
The touring test: “How can AI be indeed be considered intelligent, and how
can we decide that?”
Experiment: one human being in front of a blinding
screen, on the other side of the screen there is a human,
and a computer. They start a conversation with the
computer or the human, if they don’t now after 8 min
whether they are taking to a computer or a human being,
the computer is considered intelligent.
The computer has to be programmed in a way that
when they don’t know the answers to a question they still can
answer with something smart.
It also works the other way around: we as humans are also just
machines, we use information to do better the next time.
John Searl
The Chinese Room experiment: “Is AI ‘really’ intelligent?”
Experiment: He wants to explain that computers can and
will never be intelligent
o The person in the room gets Chinese texts and uses it
to make new sentences, then gives it back to the
people outside of the box
o The person in the room has instructions, knows
exactly what to do, BUT has no idea what the text is
actually about.
Computers can do intelligent things, but aren’t intelligent
themselves, they cant actually understand what the information
means
Kevin Warwick (2011) - A third position in the debate
Wants to transform humas into cyborgs (half machine, half human)
Professor in UK, considers himself a cyborg
Writes books about AI