Rewriting the Rules of Intuitive AI Design

00:00:10: Hello, I'm your host Michael Doudin and here with me today is Christian Kuhn to discuss UX & AI design.

00:00:17: Christian thank you for joining me Today.

00:00:20: Thank You To Heaven Me

00:00:21: Absolutely so.

00:00:22: we had a great time of few weeks ago hanging out at Wabanali And what was your big takeaway from that event?

00:00:29: A lot of things happened.

00:00:29: there's a Lot Of Great Conversations

00:00:32: Mmm.

00:00:34: So the Webinale in Berlin, I mean it's an old institution isn't there?

00:00:38: For me personally.

00:00:39: It was a very special.

00:00:41: Maybe now that was my fifteenth anniversary.

00:00:44: Wow, amazing.

00:00:45: fifteen years ago at the same hotel I performed one of my first conference talks and I Was remembering all that stage fever And which was coming along.

00:00:58: so It's just a great conference in the center of Berlin.

00:01:03: There is always very good quality of creating content, also speakers there and workshops here And for me it's so easy to get into this conversation with meaningful exchange of knowledge.

00:01:20: I mean i staying all week to dive really deep really use all the opportunities to go out for, yeah.

00:01:30: For Currywurst or a beer with some other speakers and to exchange just you know what we know?

00:01:37: And I brought home a lot of great memories, great conversations...I learned a lot!

00:01:46: It's fair to say that the webinar this year was a little bit smaller as i expected but Very on the point and

00:01:56: yeah, I love that.

00:01:56: Yeah.

00:01:57: Yeah

00:01:58: II always appreciate conferences where have a good meaningful conversations during the event after the event between events absolutely

00:02:08: Sure.

00:02:09: So tell me a little bit about your UX journey over the years And how you're now leveraging that expertise to build AI capabilities.

00:02:20: Okay, great question.

00:02:22: So I'm feeling old if you have to tell these stories again.

00:02:30: however.

00:02:31: well yeah i've working since over twenty years with user experience design With research and testing of software applications And in this these twenty years span Of applying use experience design on research was accompanied by all these technical innovations we had in that time frame, like the change from graphical user interfaces to touch interfaces.

00:03:01: So the iPhone revolution is called it.

00:03:03: even so Apple didn't invent a touch interface

00:03:05: right?

00:03:06: So two thousand seven was very important year for the industry or software and web design industry because we have to think of new paradigms.

00:03:16: how do you design and code software and websites which are working on these so-called smartphones, this new device category.

00:03:27: And I'm remembering a lot – comparing the time we're right now with that day's from two thousand seven to two thousand let us say twelve when the first iPad was also launched.

00:03:40: then the Jesus pet called back then.

00:03:44: So a lot of expectations have been about that device category as well, right?

00:03:49: Um so we didn't have this technology.

00:03:54: We did know how to do these things you need to do.

00:03:57: and who didn't know How design the things we had designed?

00:04:02: it was actually try an error.

00:04:04: Exploration experiments.

00:04:07: Every software company, every hardware company came up with all new rules.

00:04:13: How to interact with these devices was touch right?

00:04:16: We had very weird gestures like four finger gestures on the displays and most of them didn't survive.

00:04:26: I mean today we have swipe scroll pinch expand maybe double tap press long pressure.

00:04:35: I think that it's right, but two thousand eight and beyond.

00:04:42: we had a catalog of let's say thirty or more right

00:04:47: through the last.

00:04:48: yeah exactly Yeah.

00:04:51: And then we didn't also didn't have to the frameworks and technology actually produce what needed to be produced with help of Apple.

00:05:01: that Flash might be not the solution.

00:05:05: And they killed it, right?

00:05:07: By just closing out.

00:05:10: but HTML five big promises cascading style she's he says three, right.

00:05:17: response of design was invented multimodal multi device interface designed was invented

00:05:24: responsive versus adaptive design

00:05:27: Yeah, right.

00:05:28: Atomic design as well.

00:05:31: And many of these things we still use and there are so our.

00:05:37: We created fundaments for the work we're doing today Right?

00:05:40: To answer your original question is you know?

00:05:42: um We are now on that hype Of AI Is finally here to stay.

00:05:50: I want to emphasize That this is For me The second type.

00:05:55: We had already the hype, two thousand seventeen eighteen nineteen.

00:05:59: Yes it was getting quite silent around this topic in twenty nineteen and then the pandemic situation happened.

00:06:09: And then Pam!

00:06:10: Chess GDP moment.

00:06:12: that was a breakthrough for whole hype or after-hype development.

00:06:19: So, but here we are.

00:06:20: We're talking.

00:06:21: so I was working with AI systems with machine learning system with intelligence systems chat boards and voice voice driven voice interfaces in such a thing.

00:06:33: i was working these things already over ten years ago.

00:06:35: yeah

00:06:37: With and be collected a lot of experience about what the capabilities about how great it is to use this Machine Learning based Systems How hard it is to use, to talk to a machine via Alexa API or Google voice skills and so forth.

00:06:56: But we did it!

00:06:57: We tried it out...we pushed through the markets and learned about these things.

00:07:02: Variables man Internet of Things not only fitness trackers I mean including goggles and smart glasses ...We didn't call them smart glasses.

00:07:13: because then whatever AR, VR, XR.

00:07:20: You know?

00:07:22: Wish and pro... I mean it couldn't fly just too heavy to expensive very bad user experience for them.

00:07:34: great specific application of these technology in very specifically users cases but not ready for the end consumer market.

00:07:42: so In that sense is also super interesting to see how smart glasses are rolling out again and the second trial, let's say.

00:07:49: But to come back to you Richie... So my point is I saw it already that for a third time we are just having huge change of technology And we are adapting to it.

00:08:10: We professionals, we practitioners designers developers and so forth.

00:08:18: We have to ask ourselves a question, how is technology changing user behaviors and expectations?

00:08:23: And the chat GDP moment is one of these moments where expectations and requirements of users massively are shifting.

00:08:32: Massively changing.

00:08:33: Yes The iPhone that touched revolution was another Of these you know tipping points.

00:08:41: let's say Right Yeah.

00:08:44: So there's a lot of hype about AI.

00:08:47: What I see is how people struggling to use it.

00:08:51: When you observe People, How they're using AI products and features or if You even get so far To conduct a proper user research the user tests right there nothing fancy Nothing Fancy.

00:09:05: Just just having structured interview Observes.

00:09:08: the user ask some structure questions Analyze that new.

00:09:11: The outcome Is It's A very miserable User Experience.

00:09:17: And yeah, that's a point.

00:09:19: It is interesting depending upon the person their experience there technical proficiency and what they're trying to accomplish.

00:09:31: They have very different expectations of what the technology will do There expecting anything from it should work like Google search or Star Trek computer and give me absolutely everything, including a cup of tea when I ask it to.

00:09:48: And anywhere in between right?

00:09:51: The reality is not quite that at this moment.

00:09:58: Yeah Absolutely Not!

00:10:06: It's remarkable how many Alexas are sold but still used in private households.

00:10:11: Even so, the user experience wasn't never that super great.

00:10:14: I mean there is some very specific use cases where it's very helpful but a big promise was never fulfilled and it's remarkable to see actually trust of users or end consumers pretty much undermined.

00:10:33: So today most people are not even willing on their own intention, the capabilities of a voice user interface.

00:10:45: Because they had made so many bad experiences before in the last ten years or more right?

00:10:53: But that point is AI driven voice interfaces are much better nowadays.

00:11:05: They're incredibly good!

00:11:08: And you get surprised if use them, what they can do for you.

00:11:12: So my job now together I guess with marketing people is to convince consumers.

00:11:18: sure try it out again.

00:11:20: yeah that makes sense.

00:11:21: and Yeah!

00:11:22: That's the same as chatbots.

00:11:25: i mean nobody likes to term chatbot anymore right?

00:11:28: Now we call it agentics experience or agentex interface plainly like, oh another shitty chatbot.

00:11:40: you know.

00:11:41: they see the offer of their synchronization interface on a website or in certain app and they decline it.

00:11:50: They say no I don't want to try out.

00:11:52: i had met so many bad experiences with that.

00:11:56: So to convince them That today the experience might be very helpful is another challenge.

00:12:06: it will need time that people are adapting back to, okay yeah give it another chance.

00:12:13: That's really be a hard part I think

00:12:15: for me.

00:12:17: from the user experience perspective there is two things that i would expect to see and i'm curious if you actually see this in the first one because there are situations, like if I'm driving a car or busy with my hands.

00:12:40: Or in the middle of cooking

00:12:42: maybe

00:12:44: can't use text interface but talk to computer.

00:12:47: and then other one situation is based upon your age when this stuff came out.

00:12:53: so i find that kids much more likely using it than I am right?

00:12:59: Yeah!

00:13:00: If you're seeing those same types situation land or demographic?

00:13:06: Yeah, I can confirm that.

00:13:08: It's definitely depending on the age and then previous experience of user.

00:13:12: right but it is really like yeah think about yourself you have made such a bad experience with driving in certain car so don't trust that brand anymore.

00:13:23: That kind of car any more will be very hard to convince.

00:13:28: you buy another car for their brand.

00:13:30: Yeah, it's more like that.

00:13:34: It also you know what is funny?

00:13:38: You can observe every day on the streets people having very sophisticated modern cars with very sophisticated entertainment or interconnectivity systems.

00:13:48: The dashboard everything has super digitalized and they connect your phone easily.

00:13:55: we think of these system.

00:13:57: so phone-free experience while driving, talking to your phone having access to everything on it.

00:14:06: But if you just stand on a road and observe the traffic in a city You will see that A lot of people have their phones still at hand.

00:14:19: So why is this?

00:14:21: Why are they not connecting their phone with car using cars as an interface?

00:14:27: Why are they using the phone at the ear while driving, which is super dangerous?

00:14:31: The way many countries forbidden and you know.

00:14:36: It's almost always one of two answers right.

00:14:38: it's either complete lack of trust or its difficult or inconvenient to do for some reason.

00:14:45: yeah more difficult or more inconvenient than the other alternative.

00:14:49: Yeah And I think this we can use as a metaphor on adapting too many other use cases where we as practitioners who are building these kind of software products expecting that the user will adapt to it because now he can.

00:15:06: Now, be giving him this software or the interface for.

00:15:09: but they don't just ignore it.

00:15:13: They're keeping very old habits and not willing.

00:15:19: They're falling back to what is working for them best, even and they are not willing to experiment into learn.

00:15:30: So the scaffolding process of onboarding people in new ways to interact with technology it's often very underrated.

00:15:45: What you can observe right now is that many companies just bolding an input prompt field on a application with the expectation, that users will start to use that interface.

00:15:56: To interact via conversional interface means prompting writing with their AI capabilities and reality is they don't like it!

00:16:12: The reality is if you're making me write a prompt, that's actually closer to computer programming than it is to the conversation.

00:16:22: Right?

00:16:22: That's not a great question for us people!

00:16:26: If he decides to offer an input field or a prompt fields as they use interface then your choosing the worst and hardest-to-use interface will be used up.

00:16:38: the content of load is just tremendous.

00:16:43: And I know you've talked a lot about prompting, and want to take a slight detour though through an article we wrote last week, click-spin wait right?

00:16:54: Where you talk about these slot machine interface in generative AI which i think it's great article.

00:17:02: but what was interesting with context?

00:17:05: that developers or engineers, as soon as some user interface for a specific task becomes in the slightest way popular regardless of how bad it is.

00:17:19: I feel like other... The next person in line's like yeah we'll just do that because it already exists.

00:17:25: but what are your thoughts on that?

00:17:26: and maybe tell us little bit more from that article?

00:17:29: What you're thoughts are about that interface?

00:17:32: Okay let's break a leg colleagues, the developers.

00:17:38: First of all I don't assume that anyone is doing trying to create a bad user interface on purpose?

00:17:44: Oh no!

00:17:45: Of course not...

00:17:48: If we have to blame someone then we have our whole industry and are too much driven by wrong metrics like retention or how many new clients customers can use create in a certain time span.

00:18:07: How many people can you seduce into your software to subscribe for, or such as thing?

00:18:15: I think it's coming from that we orientate ourselves too much on successful competitors or something which we would like to be, you've probably heard that phrase already.

00:18:32: I'd say it's working in Netflix and Spotify... Or I want to have it as easy as Twitter!

00:18:43: These comparisons are often made right?

00:18:49: We try just copy-and-paste and a successful user interface, an interaction principles to our product.

00:18:59: And the hope that people will in the same way like it and have it as easy-to-use right?

00:19:08: That is big mistake!

00:19:11: The Right Way Is To Conduct Proper User Research And To Learn From Your Clients From Your Target Group.

00:19:22: What Do They Need?

00:19:23: what do they need?

00:19:25: What does it want, and that is an ideation creativity.

00:19:35: Where can we win them over?

00:19:36: right?

00:19:38: I think this is the error in most product teams.

00:19:42: We are not starting with a user.

00:19:44: We're starting with metrics.

00:19:55: Going back to the prompt style interface for a moment, you've done a lot of talking about some.

00:20:03: your conference presentations.

00:20:04: For example are don't make me prompt right?

00:20:07: What is the thinking of getting rid of prompts or a prompt free product design?

00:20:16: yeah so I think it's much harder to create a user interface with this actually not offering prompting as the first option.

00:20:25: It can be.

00:20:25: there is second or third option, you know?

00:20:28: But try to formulate prompts into buttons So that the fastest and most easy and brutally precise way to interact with.

00:20:42: listen computer system is a button Right one click on tap.

00:20:48: It's massive memory.

00:20:50: There is no really learning curve necessary, there is no cognitive load of how to formulate what I want.

00:20:59: it's just okay.

00:21:00: this Is What i can do?

00:21:01: Do it right?

00:21:03: Send the email write the text shorten The text create a picture.

00:21:09: and its really precise.

00:21:11: it's brutally compact doesn't give you any room for interpretation but most users appreciating that over formulating themselves what they need.

00:21:24: So the art here is actually to think as a system, can we try to assume and to predict?

00:21:31: What might be the next most relevant task for the user in offering that task formulated into a button?

00:21:41: so you have a button with a label But behind the buttons are actually a prompt.

00:21:46: Right, that prompt is prefabricated or on the fly in real time hyper individualized created for you and The user don't have to write a prompt.

00:21:55: It's just a bug.

00:21:57: And here as an update from just a couple of days ago In Google Studio AI studio You Have That Now As A Feature.

00:22:08: Oh nice

00:22:09: if you are creating this Google API Studio An Application This Studio Can Create These Kind Of predictive buttons for you.

00:22:19: That's awesome, I was waiting for it that someone is bringing them out as a developer feature and its now available!

00:22:29: I didn't try to also just read the news about this but i'm very excited to share with everyone here.

00:22:36: This one way another way we have other input fields like input types or modis where We can upload a picture, so vision AI.

00:22:50: We can use the camera we can use microphones.

00:22:54: there is so many ways to create input for our system.

00:22:57: it does not has to be written as text .We Can Use The System And The Capabilities Of AI Which Are Brilliant To Create Text Out of That Right?

00:23:07: So I can analyze the picture and write the prompt out.

00:23:10: Yes

00:23:11: Absolutely

00:23:14: Yeah.

00:23:15: Um i think It's Much More Work.

00:23:17: Yeah, of course it is.

00:23:18: Of course It Is like one of my the most used features on My phone for me right now is On Google Translate The image based translation.

00:23:31: Mmm...It's

00:23:32: brilliant!

00:23:36: And One of these hidden champions of AI?

00:23:39: Exactly and its been around For quite a long time And it doesn't require me to type anything.

00:23:45: It allows you to use the feature in real world context, and a way that is fast an efficient natural?

00:23:56: Yeah did your try out live translation already?

00:23:59: yes

00:24:01: yeah because its growing why we are using it week-to-week getting better different sometimes

00:24:13: back,

00:24:15: right?

00:24:16: Just getting worse again.

00:24:19: So there's a lot of stuff happening in the background.

00:24:21: yeah so I think this bubblefish vision you know that science fiction vision

00:24:27: um

00:24:28: that we were coming to a point where really can using our earphones for live translation

00:24:35: simultaneously

00:24:36: and very loud and confusing context man i think

00:24:41: Yeah, some of those things are incredibly good.

00:24:47: I remember the voice-to-text tools from the nineties and you had to spend hours training it on your specific voice... To hope for ninety percent accuracy!

00:25:02: And ninety percent acuracy sounds good until Exactly,

00:25:12: exactly.

00:25:12: We're

00:25:13: getting that high like ninety nine percent consistently well over translating in cross languages.

00:25:22: it's unbelievable.

00:25:23: how good is this?

00:25:25: Yeah!

00:25:26: It's amazing what you know.

00:25:29: and if I might the normal consumer out there not aware about these things I go online to do something.

00:25:44: But the marketing is promising so much, doesn't matter what kind of interface or product it is they're gluing AI labels on wash machines and fridges and promising whatever on TVs.

00:26:00: now with AI that does not tell end consumer anything about what you can do with that AI, and actually the benefit for him.

00:26:10: How is it helping him in his daily usage of a device or software?

00:26:16: And then on the other hand they're using users every day without really acknowledging it.

00:26:22: Spotify Netflix Google Maps translation out to correction and writing software doesn't matter which one.

00:26:32: And that is all AI driven and it's not even called AI anymore, which is correct

00:26:37: in my opinion.

00:26:38: Right?

00:26:39: It's technology...it will help us!

00:26:44: So I'm looking forward actually to stopping to call everything AI.

00:26:50: It's just tech.

00:26:52: Yeah yeah solve the human problem and you don't need to slam an AI label on

00:27:03: That article you mentioned, the click-spin and wait.

00:27:09: I got some not beef but some critics for it And i appreciating that It's a...I wrote an article in very let say entertaining way.

00:27:20: Yeah!

00:27:20: I wanted to shake up Some people thinking about hey what is happening here?

00:27:26: and maybe a couple of my comparisons or metaphors are using in that article, not hitting everyone's understanding.

00:27:36: However what I see if i look around is AI software, AI products and features are now using interaction patterns which we well know from social media.

00:27:54: And we have to ask ourselves, as practitioners who are producing websites and software why is that so?

00:28:01: Is it really needed or helping in any way the user.

00:28:06: That a software working with these variable disclosures of information there's plenty examples you can even AI which is non-deterministic software.

00:28:22: you know, it will be always having a different output.

00:28:24: That's correct and that's right is the core of AI.

00:28:28: however You can use AI in software without Rebuilding slot machine by forcing people to betting virtual money credits tokens whatsoever which is casino chips on an unpredictable outcome.

00:28:48: Which posts, around about twenty percent of all internet users will say their own addiction.

00:28:59: Right?

00:28:59: So it's not relevant for everyone but every one would get addicted by using an iSoftware which is designed in that way.

00:29:07: But there are many people who cannot insist on that urge and feeling to re-prompt, re-pronged, recreate another prompt because It follows exactly the same mechanics in the dopamine relationship, as a slot machine.

00:29:27: As gambling software that we learned is already implemented on social media products.

00:29:34: That's why many people are getting addicted to social media or having problems managing it.

00:29:41: We're at this point of our society where we try to hinder teenagers and get access which will not work because Tina just always find a way.

00:29:54: Right,

00:29:55: exactly!

00:29:57: We did

00:29:59: and

00:30:02: you know in the same way we are now implementing all these addictive patterns, addictive making patterns in professional software.

00:30:10: right encoding software and then designing software.

00:30:14: yeah You know?

00:30:15: And I have to ask why It's absolutely not necessary at all.

00:30:21: And it is just exposing some metrics, I believe the owners of that platforms want to present their managers like how long people are staying in this software and then using or how much tokens they have spent.

00:30:42: It doesn't feel right.

00:30:44: No, it doesn't.

00:30:46: there's definitely an ethical conversation to have around this sort of thing and There's another Taking this exact concept back a minute.

00:30:56: something else talked about like having your Amazon Echo or whatever you're Alexa in Your house.

00:31:05: why do people have them when the user experience hasn't been great?

00:31:08: They don't really necessarily use if they'll have them.

00:31:11: How much of that is tied to the personification?

00:31:18: Of The device because I'm able to communicate with it.

00:31:21: I'm unable to talk to it Because right there's some evidence That just having a conversation With, the device Changes how you feel about It regardless of being able To successfully do the thing we're trying to Do or not.

00:31:36: Is that something We need to be aware of?

00:31:39: oh yeah So we're having here synthetic empathy, synthetic performance of feelings.

00:31:50: We humans have one condition which is not in our favor as many that we are interpreting human-like behavior and feelings into things Most easily.

00:32:07: you can detect that on animals, on pets.

00:32:11: That we are interpreting in the behavior of our cat bird or dog... ...that they're feeling and thinking and behaving actually intentionally as humans would do which is totally wrong their animals And have totally different reasoning.

00:32:29: But be doing this also with technology.

00:32:32: The theory was stupid.

00:32:34: today.

00:32:35: Siri is not stupid, Siri is a person.

00:32:40: It's an interface and it was failing to understand you

00:32:45: correctly

00:32:47: And giving you feedback which isn't useful for you but you as human reacting on it like that Siri and it does not like me.

00:32:56: Today she has a bad day,

00:32:59: right?

00:32:59: And she's drama today.

00:33:02: Drama queen today!

00:33:03: She is not answering with me correctly... No no this all bullshit.

00:33:07: It's not right.

00:33:08: Synthetic personality is synthetic feelings, synthetic empathy.

00:33:14: just because we figured out that if you are using conversional interfaces in the way that its more human-like there is higher adaption rate higher, you know the people are staying much longer in their conversation and they're starting to communicate with a machine which is more human like as well.

00:33:33: It can lead that people falling in love with machines or in the persona of which software it's presenting.

00:33:41: And we have to ask ourselves if this direction we want follow.

00:33:46: Is it ethical?

00:33:47: Yeah, we're having now more and more these cases of AI psychosis.

00:33:59: People even without any precondition for mental health problems have developed psychosis by using intensively a chat board or a chat GDP similar.

00:34:16: yeah That's interesting and scary in the same way.

00:34:23: Yeah, how to detect that a person which is using your Cheshire DP or similar Is actually believing The machine too much And then regulating by saying you're having some safe guards inside of the app?

00:34:42: I mean, you know all these terrible stories where safeguards failed and AI bots told their users to kill themselves or motivated them go on with suicidal tendencies.

00:35:02: That's terrible!

00:35:03: It is a prime example of how technology is failing How it was not tested enough

00:35:10: Or

00:35:12: we didn't invent the tech to prevent such a thing.

00:35:17: I don't say that it's easy, just saying we have...I don't want to formulate in the wrong way but who is responsible for this?

00:35:33: Who has taken responsibility of these kind of fails?

00:35:36: We are working with clients and companies where when you're not delivering a proper tested and security-hardened software product, we're looting the job.

00:35:49: Right?

00:35:52: Yeah yeah that it.

00:35:55: so continuing this conversation maybe in a little bit different direction talk to me a bit about kind of your core expertise in hacks or human centered AI experiences.

00:36:13: I just need to pick up my water, just a second.

00:36:34: HAX, human AI experience exam.

00:36:41: Okay so HAX is an abbreviation coined by Microsoft.

00:36:48: the original HAX playbook workbook has... Eighteen principles.

00:36:59: Eighteen or was it over twenty?

00:37:00: I'm not sure here right now, sorry for that.

00:37:03: so a lot of principles how to create an human AI experience which is working with the humans too.

00:37:16: The core idea of HAXs To create software With the human in mind For the humans and help them to have better capabilities, right?

00:37:32: To accelerate the possibilities of the user by using AI as a co-creator.

00:37:41: So

00:37:43: what we see often in AI products and features is that... The software product is talking about capabilities.

00:37:53: What can I do?

00:37:56: Often not so much on what it cannot but But that is not really user centered.

00:38:06: It's AI centered, it talks about what the AI can do right?

00:38:11: The human-centered way to start with a user in mind and figure out where and how users use cases would be AI enhancing their users capabilities.

00:38:25: That was already the first principle, empathy first start with the user, not with capabilities.

00:38:32: So what I have done is...I've researched and created out of many resources from Microsoft, Apple Google GitHub and many scientific papers The six HAX principles as i call it these six principles are.

00:38:54: I recognized this pattern by analyzing all these materials that the six principles are in all of these resources available.

00:39:07: So, I did not reach...I didn't create them but just curated it then.

00:39:13: and the six principle is starting with user which is empathy first automation versus argumentation.

00:39:23: it's graceful failure, its mental model shaping.

00:39:28: It's getting let me go through my notes because I always forgot something is transparency and confidence a real control in editability?

00:39:41: yeah so we have here the sixth.

00:39:44: And The interesting part was when we researched and created all that materials for our online courses, that we tested these six principles as an audit tool on real-world projects in my team at Optimizer and Portal.

00:40:05: And what we learned is when you just leave out one of those principles in your audit or the software product isn't fully spilling successfully with this principle, then user experience is dropping massively!

00:40:23: So only if you address all the six principles, you can achieve a great user experience in other reasons.

00:40:31: Right?

00:40:32: It's powerful framework which we use for auditing software or produce create from stretcher software to follow them as guidelines.

00:40:45: In the next step, we have created some checklists and audit forms.

00:40:49: And I've really used it more as an audit tool.

00:40:52: now We're having a software tool for its tool to auditing all these applications So we could collect more evidence that the principles are working as framework.

00:41:05: I started then to offer full-day workshops And we conducted these workshops in Portugal, Denmark and Greece last May.

00:41:19: That was amazing because there I had a chance to work with senior developers designers product owners people who working big companies Big brands Some I can mention that was Mercedes IO.

00:41:35: There were PepsiCo some folks from the Greek energy company, which is providing everyone's energy there.

00:41:49: Telecom and so forth.

00:41:51: So in all these people could use my... These principles In our workshops And they confirmed that.

00:41:59: hey this is really helpful to understand what is missing What needs to be changed?

00:42:05: What does it need To make the software in a way That and interact with the AI in a better user experience, right?

00:42:14: In more sufficient and satisfying way.

00:42:18: Okay!

00:42:19: And then it started to fly pretty quickly by... I created an online course which i conducted now on the third edition.

00:42:30: most of us Brazilians and Portuguese.

00:42:33: Now there is new addition via Evendor in Denmark And there is an American knowledge provider I producing now on self-paced course, which will be finished in October this year.

00:42:50: So it's really getting out as a framework right?

00:42:54: Yeah i'm very happy about the development to be honest quite surprised that that it was resonating so strongly with the community, but chose where we are.

00:43:08: Again like two thousand seven eight were don't have tools yet to actually decide is there a good user interface for an AI product or not?

00:43:19: And this first attempt of a framework say okay here's something you can use and apply as a practitioner immediately if your feature is working well with the users.

00:43:31: and or what does it need to change, right?

00:43:35: And there's so much going on in the industry worldwide.

00:43:37: Right now everyone was working on these kind of topics.

00:43:41: I'm not the only one.

00:43:44: In Munich in October this year they will be the webinar web week and i am very happy that we have a full day workshop There.

00:43:52: Nice The HAX Workshop where Everyone is invited to try out that framework.

00:44:00: Hands-on, bring your laptop in good mood and then we will have a cool day.

00:44:07: I think We'll analyze some applications And create new interfaces based on what you've learned.

00:44:16: It's also fun.

00:44:18: My team has used an idea which isn't for me.

00:44:21: it was from Irrational Labs San Francisco A company specialized in behavioral science And they had made a webinar last year called the Seven Sins of AI Products.

00:44:34: Nice,

00:44:35: and that was already I think October or so last year... So long time before the Vatican was publishing anything regarding AI right?

00:44:45: The seven sins of AI product design.

00:44:51: we have used at Seven SINS as an audit tool as a dark pattern or deceptive pattern in the box because you know if your reframe principles, guidelines into certain sins it's much more catchy.

00:45:08: It is fun.

00:45:09: people might remembering that sin Rule number three.

00:45:16: And yeah, for some people it's just really easier to grasp the idea of the principle and apply as a seven-sins framework on their product is an audit tool.

00:45:27: let say we have made little website out where you can playful learn about every sin and what antidote was this thing.

00:45:38: so What You Can Do Is Making An Audit Of Your Product and you bring it to confession.

00:45:44: Nice!

00:45:45: You bring into confessions on what is the sin of your product, right?

00:45:50: And then can work from there... Yeah that's something we all do together in Munich.

00:45:56: I guess will be fun.

00:45:57: It was fun at Arsene in Denmark so i have to assume it'll still be fun

00:46:03: in October.

00:46:04: That should be good.

00:46:07: A framework like this We've needed for a long time, right?

00:46:13: Because you and I both know that there are concrete measurable components to design or to user experience.

00:46:23: But a lot of times the users or stakeholders don't realize that they're not... They don't know how to measure things And if we're very lucky it brings somebody in who knows what's going on somewhere but even then it's not necessarily documented in that way or the information isn't shared across their organization, and so you end up without framework to do testing of verification.

00:46:52: So now we're at this point where companies are trying to move faster.

00:46:57: they try automate things than ever thought about automating before And if don't have a solid codified framework for evaluation You're screwed like yeah, I'm right.

00:47:11: Yeah you're in a really big problem if you don't have a framework this.

00:47:14: So i'm not surprised that it's taking off Right now.

00:47:16: at this moment It seems like something is desperately in need?

00:47:22: Yeah, I agree The what but I learned the last two years Is That we are.

00:47:34: We Are Assuming that If your giving our users more Automatization that they will appreciate it.

00:47:45: I burned my hands as a German would say, I burnt my fingers.

00:47:49: so of course we tried out also to automate much more and now we have the capabilities with AI.

00:47:57: We didn't had these kind of capabilities just a couple years ago or not for that price tag is to be honest.

00:48:04: It was possible before but it was incredibly expensive.

00:48:09: Now, it's so easy to apply really cool technology and capabilities through any kind of

00:48:16: process.

00:48:20: It leads us often then also just say yeah let's automate that.

00:48:25: but again I'm guilty as well if you are not sitting down with the real client or end user going into a deep conversation about how much automatization they really would like

00:48:40: to see,

00:48:41: How Much Control They Really Would Like To Keep.

00:48:47: If you don't have that conversation then your end user is easily too disappointed and the trust will be undermined because he feels helpless.

00:49:02: if You Take Away The Control about the process from a human being that is you're making the machine superior to, through the humans decision-making.

00:49:15: So everyone knows the concept of human in the loop.

00:49:20: but what I learned was that people really appreciate if they can keep some control and decision- making approaches doing Often, and I use it also often as an example.

00:49:35: It's the so-called IKEA effect which is a bias that we as humans appreciating The outcome of any kind of work more As higher our efforts.

00:49:48: Our own effort was on the beginning right?

00:49:51: So if you help to build the furniture You're appreciating their furniture more.

00:49:57: That's a reason for IKEA effect.

00:49:59: It was discovered by Harvard Business Refuel, published already over ten years ago.

00:50:12: So if you are adding co-creation as in feature into your product even so that it could fully automated but if you keep on purpose the human in control to have some effort, it might be much more successful as an alternative.

00:50:35: That is one of the learnings we had made.

00:50:38: another learning that the biggest trust undermining aspect at this moment was lack of edit abilities.

00:50:49: Right, like added undo redo.

00:50:53: That's really missing in most products and people are not appreciating it to have an outcome which is not easily changed.

00:51:08: And if the only possibility change is create another prompt which is very hard to do because you have to be very, very precise about what will change.

00:51:18: Otherwise I might change much more as we want it changed.

00:51:22: so this is not control This again betting on an outcome.

00:51:32: So people are not appreciating that but they wanted something easy like a word

00:51:39: or

00:51:41: Yeah, email or whatever.

00:51:44: Full controllable but smart in the sense that hey I have here a perfect reply for you.

00:51:55: Do accept it or not?

00:51:57: Or take it as template and change on your own.

00:52:01: It's helpful because someone was analyzing an e-mail thinking about what might be the most possible answer to this E-mail And she has one click And it's done, right?

00:52:15: That is maybe working as an example for most use cases.

00:52:19: I think we can see the same in development.

00:52:22: I'm not a developer so just by conversations with other developers they told me some of them are not appreciating all the prompting and output that getting into their software.

00:52:41: Yeah,

00:52:44: I watched a speaker doing the live coding session at an event last year and i think he had to clear twenty different AI prompts suggesting things from his screen while he was trying to do something very simple.

00:53:00: And like that's too much whatever it is!

00:53:07: I mean, don't make me prompt right?

00:53:10: Prompting is not bad per se.

00:53:13: Absolutely no!

00:53:14: i'm a writer...I like to write so prompting should be fun for me.

00:53:18: the problem it's not that writing the prompts..the problem is to learn what does a good prompt needs to work for me.

00:53:26: So if i am writing in very sophisticated elaborated prompt thats quite alot of writing easily two diner for pages of writing.

00:53:38: And then, and that's the brilliance if you are able to create a very good prompt

00:53:43: yeah

00:53:43: You will get a great output

00:53:45: sure

00:53:46: right.

00:53:48: But that's not working for most people

00:53:51: Right?

00:53:51: That might work For some developers Not even for all developers or that might work for some writers not Even for all writers because I know writers who say hey but that's a fun enough writing

00:54:04: Right.

00:54:04: And I don't want to give it away, right?

00:54:10: So I think prompting should be kept as it was invented for engineers to train and correct large language models.

00:54:23: This is where prompt engineering comes from.

00:54:26: It never intended an interface with any consumers But now we have.

00:54:33: Now we have it.

00:54:34: And now some, I don't know why people think that's the way to go?

00:54:41: It will be not accepted right?

00:54:44: Is a little bit forced on us at the moment.

00:54:48: like so many things getting forced on US Like the token economy is getting force on us.

00:54:54: No one was asking The consumers clients and people who paying for all of that if they wanted Right.

00:55:06: It's a little bit tricky at this moment and it is easy probably to burn your fingers or tongue, who was saying too much about that?

00:55:16: but it feels like a bad scheme.

00:55:21: exploitation of end consumers.

00:55:26: I don't get for two decades, I pay my subscription months by month and they're getting a certain service rate.

00:55:34: And now I should change it overnight.

00:55:39: Right?

00:55:40: Yeah.

00:55:42: I'm making this comparison in my last article that the tokens are little bit like petrol digital petrol or fuel as Americans would say right.

00:55:55: so petrol is running out as a model, right?

00:56:02: We are electrifying the world.

00:56:06: So now we inventing digital fuel so we're forced to refill our machines week by week um...to be able actually do our job!

00:56:17: To have an income and make a. learning This how it feels like this I would like compare but i got also some Different opinions about that.

00:56:30: That's fair enough man, this is like open discussion.

00:56:32: We could

00:56:33: we can talk for another hour about the economics models of different business plan spits The pros and cons of them.

00:56:42: it does lead me into the next big question I have for you which Is what is the next?

00:56:46: Big thing in UX that you see happening over the next six to twelve months?

00:56:53: Okay so Okay, I'm not a fortune teller but just based on my experience of the past.

00:57:03: In the UX community industry we're having very, very lively discussion right now because our job is changing again.

00:57:14: What i observe?

00:57:15: Is that the discussion about how do you work and what kind tools are used?

00:57:26: We had this discussions many times before when we switched from Photoshop to Sketch or from sketch to Figma.

00:57:35: But there's tools.

00:57:37: Yeah, it was just tools and every new tool was bringing a lot of new exciting possibilities And in my opinion was improving the job our job as well.

00:57:52: There is a lot off figuring out right now going on how to use AI in combination with Canvas tools like Figma.

00:58:02: Sure, and the best way to make...to continue to make design.

00:58:11: what I'm missing is in the discussion The warning Like hey this all tools This is our technology.

00:58:22: It does not change the fact that we have to be..that we as designers have to know what to design and why to design it,

00:58:29: right?

00:58:30: And at the moment is all about capabilities.

00:58:32: Yes.

00:58:33: And

00:58:34: how I can use an MCP server in this?

00:58:37: on that too make my job more efficient or effective and so forth.

00:58:41: but yeah okay But Why do you want to design that feature for whom?

00:58:47: So i think i believe user research will have like a comeback big come back.

00:58:53: At the moment most companies are Cutting us off and that effort to research into interview people And two talking a tool to explore the needs of users.

00:59:06: It's all about capabilities is all about making things fast.

00:59:12: The People who having no design education, you know developer education starting to wipe code prototypes.

00:59:20: These are great prototypes very sophisticated high-end prototypes which our showing a lot of complex user journeys and this is amazing that we can do in these things.

00:59:32: We couldn't do that just one year ago, so awesome!

00:59:37: But the prototype was made to test with people.

00:59:41: A prototype's purpose is to evaluate the assumption if it works for my client.

00:59:48: And that evaluation step isn't happening.

00:59:53: The creating prototypes They deciding that this is good enough and they handed over to developers.

01:00:00: Please make it working And I believe It will come back like a boomerang.

01:00:09: The things will not work as expected, and then there were asked.

01:00:13: There was an experience researchers to investigate the real user needs here.

01:00:20: Then In the industry itself, there's a lot of discussion about synthetic users to try to use for exploration.

01:00:30: For testing and so forth.

01:00:34: There is a lot effort going on.

01:00:36: So far everything I read or heard in conferences was that it still far away from working well.

01:00:44: Okay That does not mean we will get them one year also If you can do synthetic user testing But at this moment, it's not working well.

01:00:57: When I talk to younger designers what is intriguing to me is that when you are a designer and want your design again now its all about vibe coding how we build things more much faster.

01:01:23: but where is the designer and the job right?

01:01:26: The designer has to be in the room with the developers, with some product owners who consult it.

01:01:32: To bring a different point of view on that table.

01:01:34: And the designer at the moment put like oh yeah!

01:01:37: That's the job whose dying out now... ...and will not work.

01:01:42: Developers are having no design education, designers have no developer education And both of them now can create very cool prototypes.

01:01:55: Yes, but that does not mean this is fulfilling the requirements off the product for security reasons or performance reason nor data privacy and so many other aspects we have to cover right?

01:02:13: Designers need space really draw being creative to investigate, to play around experiment creating variations of it and to formalize before we start a wipe code.

01:02:31: And that whole very important section as right now

01:02:38: put

01:02:38: by site which is not working in my opinion.

01:02:44: so I don't know if this happening as well in the area of development, but I could imagine that similar things happen there to close it and round up.

01:02:58: You know we are all people you'll meet.

01:03:01: yeah

01:03:02: We have over twenty years experience.

01:03:04: so when using these white coding tools or this AI driven tools?

01:03:17: That is usable.

01:03:19: We can also see the mistakes in the flaws and one percent of lying on other nations, or hey that makes no sense as context missing.

01:03:28: you know we can decide because we have twenty years experience.

01:03:34: a young designer does not has this experience.

01:03:37: how should they... How do we get

01:03:39: them

01:03:40: to the point where we need to discussion a little bit.

01:03:47: A

01:03:49: hundred percent agree with you on that?

01:03:51: In fact, I think that honestly... You think he just answered the last question.

01:03:56: I was gonna ask what's the number one takeaway?

01:03:57: you want to leave the audience today?

01:03:58: but i think maybe that's it.

01:04:02: no what is your take away

01:04:06: my takeaways

01:04:08: from this conversation?

01:04:09: when do you wanna leave everybody?

01:04:15: well um It's just another old wine and new pipes.

01:04:27: It's just technology, it is depending on us what we're doing with this.

01:04:33: And yes the impact of AI is massive!

01:04:37: It's

01:04:38: bigger as The Touch Revolution in this or that...it really is massive.

01:04:46: but for us how do you use technology?

01:04:49: We have to decide the rules.

01:04:51: There is a complete discussion missing about the ethical implications of it.

01:04:58: And that's shocking.

01:04:59: in my opinion, and I'm not a non-believer or an atheist but its so interesting to see that the Vatican on The Pope It's first authority who was making paper About hey we are humans!

01:05:14: We're beautiful and should handle this whole AI thing in our favor and not to exploit us ourselves.

01:05:23: So why is it coming from the Pope, I'm not from Silicon Valley?

01:05:29: We

01:05:29: don't have that discussion and i think this might take away.

01:05:34: Love it!

01:05:35: I agree with you completely on that.

01:05:39: Thank you so much.

01:05:39: It's good to hang out with you here online.

01:05:42: thank you for sharing your time.

01:05:46: Thank you very much for having me and I'm looking forward to see again.

01:05:50: Absolutely, all right?

01:05:52: You take

01:05:54: care.