Archive by Author | Laura M

Work up! x HDW : AI and ethics

The Work up! event on AI was held during Helsinki design week at Bio Rex. The event had two main speakers and two panel discussions centered around their performances. I was excited to learn more on how AI is perceived by professionals and what the big questions surrounding it’s use are at the moment.

AI as a challenger for the current work life

Minna Mustakallio gave a speech on how AI is going to challenge the current field of work. Developing organizations with AI is seldom a plug and play kind of a thing, so every organization should know what they are about and where AI could be applied to bring value. It’s very important to know what you are trying to make better with the help of AI and if it’s the right way to go.

dddd15

Themes that should be considered when organizations start onboarding AI according to Minna include

  1. Encounters : knowledge, understanding, shared visions and even conflicts create something new. Are there some parts of encounters that can be made better with AI or is there something we want to protect, something that happens between people.
  2. Changing roles: what does it require from a human/employee when you’re making decisions with AI. Is the goal to reduce or increase autonomy for a single person. Should people focus on their strengths and leave the mundane tasks to AI? Should an equal amount of effort be put to understanding human and the context AI would be used in as is put to finding opportunities to implement AI?
  3. Ethics and responsibility: Who is responsible for the decisions AI makes? How transparent is algorithm-assisted decision making?
  4. People & purpose: What are we trying to make better with AI? What does the best work-life look like in Finland 2022? What is the purpose we’re fulfilling?

dddd17

Mindfulness and using algorithms for the right purposes can bring a lot of good. Implementing AI is about co-creation. AI is a part of global digitalisation and will inevitably change the way we work.

Panel discussion about AI x Work

In the panel discussion Minna alongside with Jaana Leikas (VTT), Jaakko Särelä (Reaktor) and Petri Lattu (Nordkapp) dug in a bit deeper on the questions surrounding AI. Asking questions in fact was one thing the panelists were hoping to encourage. We are in the position right now, where we need to ask as many questions as we can to really determine where the future is going with AI.

dddd18

The panelists discussed control. How much do algorithms already steer our lives and how will the development go in the future? How much of our decisions are already based on some type of algorithms?

It was suggested by Petri that we are already algorithm augmented because we use AI in our everyday lives without paying attention to it. Humans, social media and all of the connected technologies form a cybernetic collective. AI is not some separate entity. It’s “under the hood” in many things we already use.

One problem with AI is that data is always warped in some way. If you write “hän on sihteeri” or “hän on johtaja” to Google translator and translate to english it’s easy to see what happens. This is just one example that’s easy to test. AI also speaks with the voice of a young woman (Alexa, Siri..). Jaana suggested that this is due to the fact that a female voice is easier to listen to. Things like these should still be taken into consideration.

Minna brought up that instead of asking what will the future organizations look like, we should as what do we want the future organizations to be like. Cause when you place an assumption it many times becomes reality. Can AI help us have more spare time and more meaningful work lives in the future?

The methods and processes of technological advancements should also be looked from a new angle when creating AI-systems. Development should be done in multidisciplinary and diverse teams and ethical questions should be kept in mind through the process. Education is the key to good AI implementation. It was also suggested within the panelists that in the future it might benefit the development to make a motivation map for the service as well as the user.

The questions the panelists wanted to leave the crowd with were:

  1. Will AI make communication between people easier?
  2. Is AI fair?
  3. How can I be a part of the development so that AI would benefit me in my everyday work?
  4. Why? Why do we need AI? Why are we doing this?

Ethical questions and AI

After the panel it was time for the days second speaker Maija-Riitta Ollila to take the stage and lift the veil on ethical questions in AI design.

dddd19

AI is a reflection of human life. Ethics is still human-centric as opposed to machine-centric. Maybe we should shift our sights to life-centric or earth-centric view? If the earth is not liveable there is no work life.

Maija-Riitta also pointed out that algorithms are always warped. Individuals have cognitive bias, data is warped, organizations and societies have prejudice and discrimination.

AI strengthens trends in societies. Which trends do we want to strengthen? Who or what is AI making powerful? Do individuals in work life feel empowered by AI?

Amazon, Alibaba and google have acquired a lion’s share of business opportunities surrounding AI. Where is platform economy going? Big data and big brother meet in the middle: trust good, control better? It’s time to start planning from an ethics point of view.

AI is changing things – or are we the ones driving change? Only agents can change things. Phenomenons aren’t actors, actors create and modify phenomenons. AI can’t be held responsible for the decisions it makes. There’s always a human behind that’s responsible (responsibility = response ability).

dddd20

Panel discussion on ethics and AI

The panel discussion was instantly all over the ethics discussion. Ethical principes like avoiding accidents and doing good aren’t enough – we should think about the context behind AI to really get down to something concrete.

Petri was again talking about the idea of what’s fair in terms of AI. Jurisdiction, upbringing and agreements all have an affect on our future. Transparency in societal decision making is an important theme. Jaakko suggested that everyone should take the elements of AI course and also read the book Rauhan kone by Timo Honkela.

Again the discussion went to the bias in data. This was a common link through the whole day. As humans we should think what discrimination means. AI algorithms bring existing problems to light when they use all the data they can. What aspects of data are relevant and which are a source for discrimination (when is it relevant to consider a person’s age or gender?).

What decisions do we want AI to make concerning our lives? What values does AI base it’s decisions on? Who decides on the values? Regulation is important.

When used right AI can help make the peaks and valleys lower and help us look further into the future. This could potentially bring more safety. AI can also help spot challenges in

Maija-Riitta also wanted to challenge investors to think whether they should invest in increasing consumption or in environmental technology making a tie back to her speech about earth-centered thinking and ethics.

Reflections

All in all I really enjoyed the event. I came in hoping that the discussion would reach a more concrete level with more examples regarding solutions and implementation. Maybe the panelists could have challenged each other more on the subject, now it was pretty much down to the presenter to lead the conversation.

On the other hand I was very happy that the crowd was included in the discussion through “twitter board” and polls that the crowd was asked to answer during the breaks in discussion. There were some questions or comments from the crowd, but the Twitter board seemed to have many interesting comments that could have spurred the conversation on.

The bias in data and algorithms was a topic that seemed to surface through the day. I think it’s a great topic and the event with it’s many great experts would have been a good chance to dig a little deeper. Same goes with the concrete codes of conduct from an ethical perspective – I hope the conversation between the panelists didn’t end when the “bell rang”.

We talk a lot about co-creation and multidisciplinary teams when it comes to developing solutions for the future. I think our experts in tiny Finland have great knowledge and should bump brains more often. How about a Super Hackathon on AI for the best of the best in Finland? Or a societal “co-working space” that would gather the experts that work around these subjects in the same space every now and again. Are we really coming together on this – we as people? Outside of these events that pose certain limitations.

We’re off to a start. Let’s make it a great one.

Author

Laura Manninen

 

 

Data-Driven Design Day 2018

Data-driven design day was held as part of Helsinki Design Week this year. I first heard about it when I attended the DASH Design + AI event back in June. Lassi A. Liikkanen was there talking about how AI impacts interaction design and giving a short introduction to AI and ML. At the Data-Driven Design Day the focus was not on Lassi though who was hosting the event, but in an array of interesting speakers.

Design is Emotional Functional Feasible and Sustainable

Tom Nickels from Avaus started the day by talking about how design should be first and foremost emotional, but also functional, feasible and sustainable.

dddd001

Tom talked a lot about futureproofing and about what’s on the horizon for design. His speech really resonated with me in terms of content. Futureproofing is in the intersection of emotions and AI, where these to intertwine.

dddd1.jpg

Social context and empathy, ethics and social responsibility are important subjects that rise in importance when we develop technology, design and business further.

Tom talked about three horizons. The first horizon is insight-driven design that’s already within our grasp: The digital twin created for physical products and for consumers/groups of people. It’s about designing for humans while integrating a feedback loop that keeps the iteration going (Attributes – Customer/Digital Twin – Behaviors and outcome).

Horizon 2 is generative design. This is the upcoming big change for actual automated design & content. Algorithms are producing parts of the design. Personalizing the service experience with customer information tucked in the background paradigms (the ChAIr project). Designing for digital assistants which will be a major interface change in production of services. Designing digital personalities and marketing in the age of Alexa which acts as a filter. This horizon is still covered partly in clouds, but it’s in our foreseeable future.

dddd2

Horizon 3 is emotionally adaptive design. Adapting design using AI to the emotional stage  of customers. It’s the shift from AI to (A)EI which stands for (artificial) emotional intelligence – the capacity to recognize the emotions of oneself and others. Read emotion – Emotional target – Adapt service output. There’s still a lot of work to be done for this horizon to became everyday design, but it’s out there.

Back to top with a top class customer experience

After Tom Ilari Pohjola and Elina Martikainen talked about their experience in app design to gain top class digital customer experience. 85% of bookings and thus sales are done online nowadays and Aurinkomatkat wanted to ensure seamless customer experience in digital touchpoints as customer moves through the stages of dreaming, planning and booking a holiday.

dddd3

Their app concentrated on a few key features during the stages of preparing for the holiday and the customer experience during the holiday. Testing was done in the actual locations and they used both guides, staff and customers as part of the development and iteration process. Customer needs and challenges were the number one guideline for their work. They reminded that data should drive decisions, not the highest paid persons opinion.

Designing the future of Urban Mobility

Apaar Tuli & Brylie Oxley from MaaS Global were next with their speech about the future of Urban Mobility and Whim.

dddd5

Their idea is to move from ownership of transportation towards access to mobility. Instead of focusing on how to design a better car (which will still balloon the amount of vehicles on our roads to 2.1 billion by 2050) we should be focusing on how to design a better city and encourage public transport, active modes in the city where trips are shorter (ie. public bicycles) and sharing of vehicles – MaaS is like having a skeleton key to the city. Inspiring stuff worth thinking about.

Pull down that dashboard: Are you really data-driven?

After MaaS Jan Hiekkaranta from Fourkind gave the audience a wake up call by challenging what data-driven really means.

dddd8

Everyone is data-driven nowadays: we have some data to support our views. The problem is that it’s difficult. Easily accessible doesn’t automatically equal good if data means visualizations and presets of dashboards to get data that supports your opinion.

dddd10

Jan urged everyone to embrace failures and learn from them as well as demanding more and making data as accessible as possible.

What data is meaningful then? Creating the metric first before you name it (made for masses metrics are not relevant most of the time). Stick to your data and know its limits. What’s your default option? What decision would you make if you had no data and why? Data inspired vs Data-driven : prove your ideas wrong and learn from it.

dddd9

Bringing data to life: Principles for leveraging machine power for human good

Vilma Sirainen & Jay Kaufmann from Zalando turned the focus back to utilizing the collected data.

Jay pointed out that data is the stuff below the surface that we found on. In the user experience pyramid joy is on top, usability in the middle and usefulness at the bottom. To make sure we are building useful designs we need to make sure that they are based on actual data and customer needs.

dddd11

Jay suggested a bunch of pairings to awake thought like organic compounds vs digital compounds – nature doesn’t have abrupt endings, it weaves seamless connections: what connections are meaningful here? How does this design adapt to any environment or device.

Dynamic vs static – does user input show immediate effect, are rewards clearly visible etc. and Human vs Machine – can the system interpret emotional states, what tone is appropriate etc.

After Jay Vilma gave a glimpse to their case study which is an algorithmic fashion companion. The development started by understanding their customers needs through customer portraits like the need for validation vs the need for inspiration. They did AB-testing by bidding the algorithm against an actual fashion advisor on suggesting outfits around anchor-items.

dddd12

Zalando’s goal is to create auto-related content for each customer. Content that is automated but personalized to the highest degree.

Tallink Silja digital journey: smarter design decisions through data

Next up Matias Pietilä from Qvik, who had been a big part of the previously heard Aurinkomatkat app development process gave a little insight on how to make smarter design decisions through data.

dddd13

Matias assured the audience that even if you don’t have the best data tools and practises at place you can still get some nice results by utilizing the data you have and remembering that qualitative data is data too.

His  lessons learned were:

  1. It’s about attitude, not about tools or process
  2. Don’t be afraid of sunken costs
  3. Smart default values don’t require AI
  4. Sometimes you learn by accident

It all starts with solving the correct and actual problems.

AI for news media

Last speech I had the pleasure to hear came from Jarno Koponen at Yle about AI for news media.

dddd14

Jarno is the product lead for Yle NewsWatch. The value proposition is : interesting news come to you, get the news alerts that matter to you – on topics that you find interesting, experience and interact, wherever you are (by allowing location tracking), the way that works to you (mobile, apple watch etc.).

Jarno was especially focused on the battle for the lock screen on mobile devices which he believes to be the new news feed as well as a new algorithmic layer for personalization.

dddd21

NewsWatch is developing a digital assistant called Voitto.

Voitto is a digital assistant that learns from you and with you all the time, not just in the app but on your lock screen too. Notification on the lock screen -> user can tell directly to Voitto if the content is interesting.

Figuring out the KPIs for a smart assistant is the next challenge: Content + AI + UX = Impact. What is the KPI for truth?

Reflections

This day gave me a bunch of new information and a lot of repetition on previous learnings as well – which is always good. I’m no data scientist so I was glad none of the presentations were too complicated or filled with industry terms for me to follow.

I especially enjoyed Tom’s talk as well as Jarno’s because I’m big into emotions and the humane/ethical side of designing for the future. Cheers DDDD2018 and kudos to Lassi for this informative event.

Next year maybe have polls for participants during the event online or think up some other cool ways to encourage questions and conversation, this year it was very “finnish” 😉 (myself included).

Author

Laura Manninen