I can introduce Linda Kronman now. While she's preparing for her presentation, I can give you a little bit of background. We've known each other for quite a while now. Linda is basically the half of an artist's collective called Kairos, together with Andreas Tingele, who's also around, I think. He was around earlier. But Linda will be the one giving the talk today. I think it's also been many, many years, decades probably, that you first participated at an AMRO festival. Was 2016 the first? Yeah, I think we participated in the earlier one like in the audience but that we actually were doing something was 2016. So I still remember maybe some of you also remember their project behind the smart world and others that followed. So I have to just quickly check the background of this talk that you are giving today. The title is Ideal Behavior Investigating AI Bias in the Employment Sector and you're doing this as part of a research that you conduct in a residency at a Helsinki-based space for media art and digital culture called mCult. And this is, again, part of the European Media Art Platform EMAP fellowship. So I hope I got this right, and if you're ready, then thank you for your talk. Thank you. Hello, everybody. Some of you know us quite well. We also lived in Linz for a while with Andreas. For those who don't know that well, we've been working together with Andreas as the collective Kairuus-kollektiivisessa vuonna 2010. Meidän taiteelliset harjoitukset ovat katsottava, miten teknologiat käytetään, mutta myös käyttää yhdessä elämässä kulttuurissa. Tämän suhteen myös taiteilijärjestelmämme ideoiden käsittely on menossa. in everyday life in creative ways. And that's also in this direction that our artwork ideal behavior is going. I'm very happy to have some legal expertise here because we are also looking at, like, a field, the employment sector, which is, of course, legally bounded. Is the microphone okay? The presentation is not. Okay. Okay, I saw it there. No, no. So, no. Niin, kuten mainitsit, teemme tätä mCultissa Helsingissä. Olen Helsingistä. Olemme yhteistyössä Andreasin kanssa, joka on eMAP-residentti. Olemme ensimmäiset taiteilijat Minna Tarkkan residenciin, joka on hyvin kiva puuvallinna Helsingissä. passed away after a short sickness and there has been a lot of restructuring in the organization but for us it's been a good place to be also because this is the first residency we are doing with family and that's also not always the easiest thing to puzzle your days or your schedules together. The ideal behavior builds actually upon earlier artwork which was also part of my PhD, called Suspicious Behavior, where we look at how data sets have been annotated and classified for surveillance. Ideal behavior looks at... When suspicious behavior is about this, like what is not normal, what is, it's ideal behavior is how we try to change ourselves to be the best of us ourselves when we are applying for a job. So it's very much addressing the ways AI affects the employment sector and specifically recruiting. We have chosen to take IKEA as our case study. So Ikea on yleensä tietty yritys, mutta se on myös aivan ideallinen kotimaana, joka on tullut globaalista ideaa, miten asettelua ja elämää voi tuottaa. to furnish your home and how to live. And they have also been one of the companies pioneering AI in job recruiting. So what are we doing in this project? similar tricks and AI tools to optimize our performance. We are applying for jobs at IKEA and using different AI tools to do this. this and we are also investigating how far we can get in the process of actually getting to talk to a human being in this recruiting process. So will we succeed in matching the matrix of an ideal candidate, the ideal of IKEA for the job, and can we get interviewed by a human? And it's actually like what's kind of behind there. Now, sorry. That's actually Andreas' phone. So we are asking if AI is making this recruiting process fairer. Is it de-biasing human prejudice as a lot of these companies who sell AI tools for recruiters for HR are claiming. So we also know that there's a lot of greenwashing happening with IKEA. Our hypothesis is that IKEA is putting out a brand image that they care about diversity and they want to have diverse people working for them. But is this actually happening? We also have forgotten that actually the founder of IKEA in the 90s, Ikea-järjestön alkuperäisestä. 90-luvulla oli näitä käsitteitä hänen oikeastaan olevan hyvin liittyviä Suomen nazipartiin liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä liittyviä journalists have been writing about, but with the kind of ideal Swedish values that IKEA has been pushing, this whole scandal was also kind of forgotten, even though a lot of Swedish journalists have been writing about this. And also there is like, they have been also doing tax invasion, which is the second scandal that kind of IKEA has been shadowed by. We have been looking at the application process at IKEA. This is very much still work in progress. We are still in the middle of doing the residency. What I'm talking about here today is the research that we have done until now, and what we are still gonna continue with the next couple of months. So we have been looking at the privacy notice, and from there we have been able to actually figure out like what kind of technologies, what kind of platforms IKEA is using in their application process. But it's also something that I didn't really know was that IKEA is actually a franchise so it's not one company there's like one company that owns the brand and then there's other companies that actually own other companies and this is the whole tax invasion ja tämä on koko talouskysymys, joka on myös kirjoitettu. Heillä on Ikea-järjestelmä. Tämä on oikeastaan yksilönsä hyökkäysohjelma. Ikea, joka omistaa markkinan, ja sitten on franchiseja. the IKEA that owns the brand, and then there are the franchises. So when we apply for a job at IKEA in Austria, then we are actually applying for a job at a company called Ink Holding. And this is kind of the biggest franchise owner. They own for example both in Finland and in Austria and in Europe most of the houses. It's a very complicated scheme and here is only parts of it. But in the privacy statement, we also found which platforms IKEA is using. So in IKEA Austria, or IKEA in general, Ikea käytettiin yleensä Avature-palvelun, joka on london-asemassa. Joka on järjestetty järjestöjä sivuilta ja Nytlantilla. data on servers in Germany and the Netherlands. And then we have actually in Austria we have a different company that comes in to do the recruiting, which is called Michael Page International Austria, also a UK based company. Ja sitten meillä on Higher View, joka tekee videoilmoituksia. Tämä on yhdysvalto, mutta dataa saavutettuja sivuilta on EU-palveluissa. are in EU. So this is kind of what we have been looking at, so that we can look more closely what these different companies are, how they are talking about AI and using AI in recruiting. in recruiting. All of these companies are very much talking about that AI will be the way of de-biasing human bias. So AI will really be like the fair way of employing people. You have to also like in the application process, you have to agree upon that you are getting, if you get to the next stage, that you are agreeing upon using this HireView platform. platform. So here we have also like kind of these claims that they want to reduce, they are like the technologies are reducing bias. This is the Avature platform which is used for the recruiting process. This is also most probably the platform where the CV is getting, when you upload your CV at the IKEA, to the IKEA interface. This is the technology that is used to do kind of a semantic search on your CV, which is then what the recruiter will see to a kind of interface like how many words and matching. And this is also technology that is powered by AI. What actually these technologies are doing concretely is quite hard to find. And this has to do with the transparency thing. It's very hard to find knowledge on what kind of data sets they were trained on or what do they reference to and so on. So we are still in this process of trying to dig deeper and connect the small bits together and to be actually confident enough to say that this is actually how it's working. to say that this is actually how it's working. Inka Group also says that it also only collects information about gender and age for statistics and legal reporting, so it's not data used in the assessment of your application, which implies that they also, the policymakers are very keen on doing this anonymization, because this is considered like kind of how technology can de-bias the recruiting process. the recruiting process. So AI hiring tools are basically marketed to be fairer by erasing gender, race, ethnicity and age from their hiring process. Instead these tools are using these standards that have already been used for a long time in recruiting, että nähdään koulutuksia ja henkilökohtaisia, eikä seuraavaa, kuten kenties ja aikaa. Yksi näistä on Big Five, jolla on erilaisia tavoitteita tehdä tätä. There are different ways of doing this. The history of this is from linguistics and psychology. But also there is now a lot of video, like kind of AI analysis on videos that can claim that they can do this kind of personality, like judge your personality according to your appearance and how you behave in front of a camera. appearance and how you behave in front of a camera. There's actually a really good journalist piece on this, where they were testing these different kind of tools and showing how just changing the background of the talker would make you more open, or if you put glasses on, or if you wear a scarf, how this changes. So it's really like kind of pointing out that at least on appearance, you can't do this. We have also been playing around with a tool. This is että ainakin näkökulmasta ei voi tehdä tätä. Olemme myös soittaneet näkökulmasta. Tämä on Oksfordin ryhmä, joka on kehittänyt tämän työn. Olemme myös yrittäneet tämän työn eri ryhmien kanssa, jotta näemme, miten tämä toimii ja miten se muuttuu. with different groups to see how this works, how it changes. Another way of doing this personality, typical way of doing this personality kind of checks is that you have questions and more and more there are these questions with just images? And then there's also like quite interesting social media threads on people posting these images and saying that this is ridiculous. Like, how should I interpretate these images? Am I more like this statue or this hand coming out? Like, what is this? How do I, what's kind of the rigorous research behind this that you can say that this actually maps to some kind of personality traits. But this is also taken very seriously from like developers of these tests. And they are then, new technologies are always implemented to this, so I mean first we were doing this with pencil and paper, these tests. Now they've been digitalized and now AI is coming into it. And I was actually talking for quite a long time with a friend of mine who is actually a headhunter, but she's also looking for another job. And she showed me actually a report in Finland by this company that uses AI to do this. Finlandin yrityksen, joka käyttää AI-tietoa. Hän teki tätä työskentelyyn. Erityinen asia oli se, että ettei yrittänyt kertoa, miten joku näyttää ja miten se toimii, mutta se oli käytettävissä esityksessä. Se myös vaikuttaa siitä, millainen tilanne näiden koulutuksen mukaan on käsittämällä itsesi. Onko se jotain, jota sinä vastaat kysymyksiin, evaluating your personality and is it something that you are answering to questions or is it something that they try to read from your appearance or your face. So there is a gradient in how to use these technologies, of course. how to use these technologies, of course. But then there is a really good article from Eleanor Drage and Kerry McCarrott, who are actually, if you want to listen to a really good podcast and just one podcast, I would recommend the Good Robot podcast where they talk about where technology means feminism. And that's from Eleanor Drage and Carrie Currett, this podcast. But there, they have this article where they argue that, or where they actually challenge these assumptions that AI is neutral or is able to de-bias human decision making at all. They are analyzing the personality instead to bypass race and gender is what they argue is kind of just a way of ignoring or actually misunderstanding what race and gender actually is, and other identity traits. And this is something that can't be stripped from us. So we have kind of this anonymity versus diversity. So a lot of companies want to have more diverse people coming, for example in technology, to work. But in the end, by anonymizing the process, they are kind of like shutting off the identity that we come all with different backgrounds into applying for jobs. So actually what these technologies are doing is exactly this, trying to form this ideal behavior, ideal person that the company wants to hire. And these decisions are still as biased as they are when a human being does it, even if AI does it. So this article has been very inspiring for us, Tämä artikla on ollut hyvin inspiroivaa, miten ajattelemme biasia työskentelyn ja työkalun osalta. Tämä on myös se, josta päädyin, kun aloitin ymmärtämohonkin järjestelmään, ja yritän ymmärtää sitä, mitä EU-Ai-järjestelmä tarkoittaa, koska työntekijöiden sektorin on kuitenkin käsitellyt suurta riskiä EU-järjestelmässä. Se pitäisi olla sääntöönotettava, se pitäisi olla tutkittava ja niin edelleen. Mutta on todella vaikeaa ymmärtää, miten kaikki näyttää. Ja se on vielä melko iso maa, mitä tuotteita voi laittaa. Mutta tämä on myös vähän vähentävä. HR-sektorit ovat myös todella tietoisia näistä asioista. aware of these things right now, so it's at least slowing down that they are not using just everything because AI is hyped so much. But then when you read kind of these reports or these white papers about different companies that are providing these AI tools for recruiting, then you get these claims that this does not mean that the inclusion of these predictors in AI assessments is not justified. It means only that the scientific research has yet to catch up with the applications of these technologies. So it's always like about more data, better models, more auditing, but it's still like we want to sell you, because of course these companies benefit from selling you the technologies. That's their business. että haluamme tuoda sinut, koska tietysti nämä yritykset hyödyttävät tekemään sinulle teknologioita, jotka ovat heidän yrityksensä. Mutta meidän tapa mennä tähän on se, että identiteetti ei ole kokoamassa asioita, joita voidaan vaihtaa, eikä se edesvaihda päätöksen tehtävää yksilöllisistä ihmisen of attributes, then can be switched off, and neither is de-biasing the decision made by individual human recruiters a way to fix systematic problems of discrimination in working culture. So we are not, we are actually quite like privileged. We are still in our 40s. We have both now a PhD. We have education. We have had all of these things. So when we are applying for the jobs, we are not the ones who are the most discriminated. And also these tools or these kind of taxonomic sorting that exists in our society, how we look at the ideal employee, we fit in that box. ja miten katsomme ideoita työnantajaa, niin me olemme seurassa. Mutta silti kun katsomme, kun katsomme artistin taustaa IKEA-töitä, meidän on tehtävä eteenpäin, että karjamme näyttäisi ideoita IKEA-töistä. make our career look ideal for IKEA. We are applying as ourselves with our own CVs. And then we are looking at a lot of influencers who are also gainers in this Wild West because there is loads of different tips, how to make your LinkedIn page look good, how to use chat GTP, how to prepare for interviews for different companies and so on. So we are then more looking at like what are we told to do in terms of getting this job. And there's also like, it's not just on the level of like kind of YouTube videos and how to do videos or TikTok videos. It's also like a lot of HR companies are also like giving these tips on how to apply for jobs. So this is one thing that we have tried. This is an AI generator, which you upload, I think, it was eight selfies of yourself, and then you can make professional photos of yourself. and then you can make professional photos of yourself. And you have unlimited ways of making professional photos of yourself. Then it's of course like, is this actually me? Or like, what is this? This is a very interesting question. Can I put this in my CV or not? I don't know. We also are looking into that you can also use like change your voice. If we are doing like AI interviews now, what what if I want to make my voice more like, change the personality of my voice, that it will be more ideal. Then, of course, chat GPT. And here we, for example, used AI to ask, Ja tässä esimerkiksi käytimme AI-tietoa kysymään, että mitkä ovat minun käsittelytietoni taiteilijoiden kanssa työskentelemisessä, koska se on paljon käsittelytietoja, erityisesti kun sinulla on PhD ja haluat tehdä työtä akademisessa. Se on käsittelytietoja. you have a PhD and want to have a job outside of academia, it's all about transferable skills. And what is also interesting that I found that there's a group who actually takes this framework of these personality tests and have made it on different large language models and actually figured out that if I use GPT-3, I'm very open and this also like kind of if they are using these technologies to figure out my personality from my motivation letter, then actually my personality is blended with the personality of chat GPT. And there's a. Ja siellä on paljon näistä tyyliä. Mutta vieläkin, vastaukset, joita olen saanut Ikeasta, ovat aina hyvin poliittisia. Ei, emme ole vielä saaneet tämän intervjuun. Mutta me olemme aina pyytäneet uudestaan, stage yet, but we are always asked to apply again, which we are doing. So all of this we are planning to work into a video work, which looks at this kind of how AI is both amplifying and reproducing ideal behavior, when both job seekers and the recruiters are increasingly automating using AI. So we have been filming at IKEA, and then the narrative is going to be more on the sound side. Filmoitetaan Ikea-kirjassa ja sitten narratiivinen osa on suunnattu. Toivomme myös käyttää Ikea-kirjoja, jotka rakentavat tämän asian. Kiitos. Thank you.