This year, A and I became the probably most used letters in the alphabet. Time to reflect upon the role we play as designers in an increasingly AI-driven landscape.
11. Question:
How do you envision AI shaping your work and the field of design in the long run?
I think AI could lead to entirely new
ways of designing that we can't even
conceptualize yet
9
I foresee AI tools maturing to a point
where they handle mundane tasks,
letting me focus entirely on creative
problem-solving
8
I believe that AI will fundamentally
reshape the design field, introducing
new roles and making some current
roles obsolete
4
I don't think AI will have more than a
marginal impact on my work or the
broader field of design even in the long
term
1
14. Foundation Value Creation
• SXSW 2023
• Everlearn x2
• AI committee formation
The aha moment
Our internal process
15. Foundation Value Creation
• SXSW 2023
• Everlearn x2
• AI committee formation
• Safe usage & Risk management
• Discuss in internal kick offs
• Paragraph in proposal
The aha moment
Guidelines & Policy
Our internal process
16. Foundation Value Creation
• SXSW 2023
• Everlearn x2
• AI committee formation
• Safe usage & Risk management
• Discuss in internal kick offs
• Paragraph in proposal
The aha moment
Guidelines & Policy
Understanding & Testing
• Understand risks
• Explore opportunities
• Upskilling internally
Our internal process
17. Foundation Value Creation
• SXSW 2023
• Everlearn x2
• AI committee formation
• Safe usage & Risk management
• Discuss in internal kick offs
• Paragraph in proposal
• Ensure everyone has the right
tools to work with (efficient &
safely)
• Reduce technical risks of being
dependent on immature tools
• Generative AI as a natural part of
our internal processes?
• AI Solutions as logic options
integrated in our deliverables?
The aha moment
Guidelines & Policy
Understanding & Testing
AI Implementation
• Understand risks
• Explore opportunities
• Upskilling internally
Our internal process
18. Foundation Value Creation
• SXSW 2023
• Everlearn x2
• AI committee formation
• Safe usage & Risk management
• Discuss in internal kick offs
• Paragraph in proposal
• Ensure everyone has the right
tools to work with (efficient &
safely)
• Reduce technical risks of being
dependent on immature tools
• Generative AI as a natural part of
our internal processes?
• AI Solutions as logic options
integrated in our deliverables?
The aha moment
Guidelines & Policy
Understanding & Testing
AI Implementation
• Understand risks
• Explore opportunities
• Upskilling internally
Our internal process
20. Generative AI exploration is accelerating, thanks to the
popularity of stable diffusion, midjourney, ChatGPT and
large language models. End-user organizations in most
industries aggressively experiment with generative AI.
Gartner VP Analyst Svetlana Sicular,
22. AI Guidelines
Transparency
We are open with our clients
about what tools we use and
when we use it in the process.
Privacy
We respect the privacy of both
our colleagues and clients. We
do not paste or type in any
names of individuals or
company names while using any
AI tools.
AI with clients
23. AI Guidelines
Human decision-making
We are aware that we have the
ability to override any type of
decision that AI tools have
made.
Exploring together
We have an open mindset
around the topic and are willing
to try to implement it in our
processes to better understand
where in the process it make
sense.
AI with colleagues
24. AI Guidelines
Mastery of Prompting
We make sure to stay
knowledgeable and learn the art
of prompting to receive relevant
and accurate responses
Content Quality & Safety
We think twice when we create
prompts to reduce the risk of
getting harmful, biased, or
undesirable content generated.
Prompt Engineering
25. Risks & Challenges
Loss of uniqueness
Generic input will most likely render generic
output, why we need to understand how to
prompt.
Ethics & Biases
Models are a product of the data they were
trained on, in many cases biased.
Hallucination
Models do not have a good or true sense of
reality and may create incorrect information.
Dependent on third-party providers
Becoming dependent on third-party providers
makes us vulnerable to sudden price changes
etc.
Copyright
AI-generated work are constructed from
existing material. Can we use it without
breaking copyright law?
Privacy & Security
How is information we input used? How do we
safeguard confidential data?
27. Two types of AI in our work
Tools
As a help to improve and optimize our own
processes
Products
When AI is part of the products and
services we create
28. Proof by Design
DESIGN PRACTICE
Insight
What is the problem?
Concept development and test
Prototype, test, learn
Development
Iterative process
29. Proof by Design
DESIGN PRACTICE
Insight
What is the problem?
Concept development and test
Prototype, test, learn
Development
Iterative process
Create interview
guide
Plan
workshop/sprint
Competitor analysis
Desk research
Ideation
Transcribe
interviews
Analyze research
30. Proof by Design
DESIGN PRACTICE
Insight
What is the problem?
Concept development and test
Prototype, test, learn
Development
Iterative process
Create concepts
for test
Analyze
user tests
Create test
manuscript
Create text/images to
prototypes
Transcribe
user tests
31. Proof by Design
DESIGN PRACTICE
Insight
What is the problem?
Concept development and test
Prototype, test, learn
Development
Iterative process
Optimize
code
Find bugs
in code
First draft
of code
33. How frequently do you partner with AI-tools in your work?
Weekly:
I engage with AI tools about once or
twice a week
9
Daily:
AI is an integral part of my workflow 8
Occasionally:
I've experimented but don't use AI
consistently
7
Monthly:
Once or twice a month, I'll use an AI tool 3
Never:
Haven't dabbled in AI tools yet 3
38. Risks & Challenges
Loss of uniqueness
Generic input will most likely render generic
output, why we need to understand how to
prompt.
Ethics & Biases
Models are a product of the data they were
trained on, in many cases biased.
Hallucination
Models do not have a good or true sense of
reality and may create incorrect information.
Dependent on third-party providers
Becoming dependent on third-party providers
makes us vulnerable to sudden price changes
etc.
Copyright
AI-generated work are constructed from
existing material. Can we use it without
breaking copyright law?
Privacy & Security
How is information we input used? How do we
safeguard confidential data?
39. Risks & Challenges
Loss of uniqueness
Generic input will most likely render generic
output, why we need to understand how to
prompt.
Ethics & Biases
Models are a product of the data they were
trained on, in many cases biased.
Hallucination
Models do not have a good or true sense of
reality and may create incorrect information.
Dependent on third-party providers
Becoming dependent on third-party providers
makes us vulnerable to sudden price changes
etc.
Copyright
AI-generated work are constructed from
existing material. Can we use it without
breaking copyright law?
Privacy & Security
How is information we input used? How do we
safeguard confidential data?
44. Risks & Challenges
Loss of uniqueness
Generic input will most likely render generic
output, why we need to understand how to
prompt.
Ethics & Biases
Models are a product of the data they were
trained on, in many cases biased.
Hallucination
Models do not have a good or true sense of
reality and may create incorrect information.
Dependent on third-party providers
Becoming dependent on third-party providers
makes us vulnerable to sudden price changes
etc.
Copyright
AI-generated work are constructed from
existing material. Can we use it without
breaking copyright law?
Privacy & Security
How is information we input used? How do we
safeguard confidential data?
45. Risks & Challenges
Loss of uniqueness
Generic input will most likely render generic
output, why we need to understand how to
prompt.
Ethics & Biases
Models are a product of the data they were
trained on, in many cases biased.
Hallucination
Models do not have a good or true sense of
reality and may create incorrect information.
Dependent on third-party providers
Becoming dependent on third-party providers
makes us vulnerable to sudden price changes
etc.
Copyright
AI-generated work are constructed from
existing material. Can we use it without
breaking copyright law?
Privacy & Security
How is information we input used? How do we
safeguard confidential data?
48. Risks & Challenges
Loss of uniqueness
Generic input will most likely render generic
output, why we need to understand how to
prompt.
Ethics & Biases
Models are a product of the data they were
trained on, in many cases biased.
Hallucination
Models do not have a good or true sense of
reality and may create incorrect information.
Dependent on third-party providers
Becoming dependent on third-party providers
makes us vulnerable to sudden price changes
etc.
Copyright
AI-generated work are constructed from
existing material. Can we use it without
breaking copyright law?
Privacy & Security
How is information we input used? How do we
safeguard confidential data?
49. Risks & Challenges
Loss of uniqueness
Generic input will most likely render generic
output, why we need to understand how to
prompt.
Ethics & Biases
Models are a product of the data they were
trained on, in many cases biased.
Hallucination
Models do not have a good or true sense of
reality and may create incorrect information.
Dependent on third-party providers
Becoming dependent on third-party providers
makes us vulnerable to sudden price changes
etc.
Copyright
AI-generated work are constructed from
existing material. Can we use it without
breaking copyright law?
Privacy & Security
How is information we input used? How do we
safeguard confidential data?
52. Case:
Insurance bot
User problem:
It is difficult for the users to sort in
the information on the website to
figure out if they are covered or not
55. Risks & Challenges
Loss of uniqueness
Generic input will most likely render generic
output, why we need to understand how to
prompt.
Ethics & Biases
Models are a product of the data they were
trained on, in many cases biased.
Hallucination
Models do not have a good or true sense of
reality and may create incorrect information.
Dependent on third-party providers
Becoming dependent on third-party providers
makes us vulnerable to sudden price changes
etc.
Copyright
AI-generated work are constructed from
existing material. Can we use it without
breaking copyright law?
Privacy & Security
How is information we input used? How do we
safeguard confidential data?
56. Risks & Challenges
Loss of uniqueness
Generic input will most likely render generic
output, why we need to understand how to
prompt.
Ethics & Biases
Models are a product of the data they were
trained on, in many cases biased.
Hallucination
Models do not have a good or true sense of
reality and may create incorrect information.
Dependent on third-party providers
Becoming dependent on third-party providers
makes us vulnerable to sudden price changes
etc.
Copyright
AI-generated work are constructed from
existing material. Can we use it without
breaking copyright law?
Privacy & Security
How is information we input used? How do we
safeguard confidential data?
64. Top 3 inspiration
1.Best podcast
AI revolutionen - (in Danish) host: Mikkel Rosenvold & Anders Bæk-
Møller
2.Best talk
Design in Tech 2023: Design and Artificial Intelligence - John
Meada
3.Best AI design ressource
AI meets design toolkit - Tools to help at each step of the design
thinking process to design with and for AI
65. What do you think about
bringing AI into the design
process?