/ My talk from the UXcamp Europe in Berlin. Please enjoy and feel free and don't hesitate to contact me if you have questions or want to talk about UX and AI
What is artificial intelligence, how do we create collaboration and what’s gonna happen in the future?
20. Machine Learning is a part of AI.
It’s the ability for an algorithm to
learn from prior data in order to
produce a behaviour.
ML is teaching machines to make
decisions in situations they have
never seen.
21. Deep Learning is a branch of ML.
Stands for a class of optimisation
methods of artificial neural networks.
AI
Machine
Learning
Deep
Learning
The goal is, to enrich and improve the
many layers of those networks.
22. A neural network is a collection of
connected simple units called
artificial neurons.
Artificial neural networks find
patterns in raw data by combining
multiple layers of artificial neurons.
23. Natural Language Processing is the
understanding, processing and
reproduction of natural language.
NLU is a huge priority and challenge
in AI research.
Because human communication is not
straightforward, but the key to
radical progress.
28. Narrow
Intelligence
General
Intelligence
Superartificial
Intelligence
“An intellect that is much smarter than the
best human brains in practically every field:
creativity, general wisdom and social skills.”
Nick Bostrom
When a machine can do things in a way that
is indistinguishable from human behaviour.
AI of today can do specific tasks:
(driving a car, translation or Netflix binge).
40. !AI has by now succeeded in doing essentially
everything that requires ‘thinking’, but has
failed to do most of what people and animals
do ‘without thinking.’ Donald Knuth
42. Surprisingly, despite AI’s breadth of impact, the types of
it being deployed are still extremely limited.
43. AI work requires carefully choosing an input and a response
and providing the necessary data to help the AI understand
the input to response relationship.
Choosing those things creatively has already
revolutionized many industries. It is poised
to revolutionize many more.
47. AI algorithms should make people’s jobs simpler,
easier, and more productive.
Status quo: most AI software tools are developed
by software engineers.
49. The large and at the same time easy UX challenge:
Create tools, which can be populated with
knowledge without someone with a PhD in
Machine Learning having to be in the room.
50. Look back: First commercial software
Image: Rexhep-bunjaku/Wikimedia Commons
51. User Experience will probably mark out the winners from the
losers in the race to commercial success for AI software firms.
Image: Rexhep-bunjaku/Wikimedia Commons
53. _BLANC
DISCLAIMER:
All design- and usability-paradigms are of course relevant
for AI tools too.
Image: LESLIE JONES COLLECTION/BOSTON PUBLIC LIBRARY
55. As artificial intelligence algorithms infiltrate the enterprise,
organizational learning matters as much as machine learning.
- AI -
- FRAGILE -
Image: focusmovers.com/
56. Prefer smart algorithms over well thought use cases is the
biggest mistake in many recent company AI initiatives.
Image: wikipedia.org
57. The hunt for better results is gonna shift from training the
algorithms to improve the use cases.
Image: No source
69. Question Answer
In the future, questions will be solved by returning questions…
…and the path to the correct answers includes a series of
answers I have to give first.
Question
Answer
Question
Answer
Question
Answer
Question
70. There will be a shift in expectation: some computers don’t
serve ‘dumb answers’ but ‘smart questions’.
So what are the rules and etiquette of computers
asking us questions and conversing with us?
71. In particular how users form a view about
whether the time investment in this kind of
interaction is likely to yield a positive reward.
What has learned about the usage and
acceptance of Wizards or decision trees
will help here.
72. Usern den erfolgreichen und einfachen Umgang
mit Algorithmen,Trainingsmodellen, Dateneingabe,
Darstellung, etc. ermöglichen.
If customers can sense or ‘smell’ a waste of time they will bail
out.The earliest moments of interaction with AI are likely to be
critical, and they probably need deliberate design.
Image: http://www.roadpickle.com/spam-museum-of-austin-mn/ - Steve Johnson
80. AI is already part of so many
applications,
we just don’t know it.
Image: Google
81. Similar to good design,
AI should be decent, but still
support the user in a best
possible way.
Image: Google
82. The large amount of data and the complexity in visualisation
is way to big to pass it unfiltered to the user.
Our job is, to enable the user to browse and
analyse large amounts of data.
Easy isn’t it?
96. ?
In 2013, researchers at Google set loose a
neural network on a corpus of 3 million
words taken from Google News texts.
The neural net’s goal was to look for
patterns in the way words appear next
to each other.
97. Words with similar meanings occupied similar
parts of this vector space.
They could represent these patterns using vectors in a
vector space with some 300 dimensions.
The relationships between words could be
captured by simple vector algebra:
“man : king :: woman : queen.”
“sister : woman :: brother : man,”
98. This data set is called Word2vec
and is hugely powerful.
Researchers use it to better
understand everything
from machine translation to
intelligent Web searching.
Image: http://mylearning9.com/?p=4
100. There is a problem with this database:
It is obviously sexist.
Image: http://flatironschurch.com/fi-messages/so-far-so-good-far-good-hope/
101. Examples:
Paris : France ::Tokyo : x x = Japan
father : doctor :: mother : x x = nurse
man : computer programmer :: woman : x x = homemaker
SAY WHAAAAAAT
102. Survey, whether these analogies
are appropriate or inappropriate.
Removed the warp with “hard
de-biasing”-process.
Searching the vector space for
word pairs that produce a similar
vector to “she: he”.
Correct the warp, but preserve
the overall structure of the space.
Image: Retronaut / Mashable
103. Computer programmer CV
That has important applications - One example: web search
Computer
programmer
#3
#1
#2
#4
104. The AI community itself has a homemade problem with this kind
of challenge, because most of them are still white, young and male.
Image: R.H. Fowler
109. We generally perceive other people to be
reasonably competent drivers. Mostly.
We understand why people behave the way they
do on an intuitive level, and feel like we can
predict how they will behave.
We don’t have this empathy for current AI.
110. Siri doesn’t make life-
changing decisions for you.
It’s okay if it isn’t really
clear how it comes to its
conclusions.
Image: Jim Merithew/Cult of Mac
111. BEEPBEEP
But interacting with a system
that makes an important
decision for you requires
much more than a few
buttons and a status indicator.
Where the magic
happens
Damn complex
input
Damn complex
output
112. Trust Empathy HumanTechnology
If the purpose of smart systems is to make sophisticated subtle
decisions, it is pointless if people can’t trust them to do so.
129. Think about use cases for the AI
Questions will be the new answers
Trust in machines is hard work
Fit your reasoning on your user
Don’t pass the complexity to the user
Usable without a PhD in Machine Learning
No bias. Seriously.This sucks
131. AI has the potential to reflect both the best
and the worst of humanity.
AI providing conversation
and comfort to the lonely
AI engaging in racial
discrimination.
132. The biggest harm that AI is likely to do to individuals in the
short term is job displacement, as the amount of work we can
automate with AI is vastly bigger than before.
It’s our turn to make sure we are building a
world in which every individual has an
opportunity to thrive.
133. ?
What about us?
Those that want to stay relevant in
their professions will need to focus on
skills and capabilities that artificial
intelligence has trouble replicating.
Understanding, motivating, and
interacting with human beings. Megan Beck &
Barry Libert
138. _BLANC
WHERE AI CAN HARM
“No set of individuals has control over advanced set of AI”
“AI on its own will not develop something bad, it’s just
the people using it in a way thats bad”
Elon Musk
Image: OpenAI