{"id":394,"date":"2026-01-20T11:27:29","date_gmt":"2026-01-20T09:27:29","guid":{"rendered":"https:\/\/sites.uwasa.fi\/bugged\/?p=394"},"modified":"2026-01-21T08:41:31","modified_gmt":"2026-01-21T06:41:31","slug":"artificial-emotions-of-ai","status":"publish","type":"post","link":"https:\/\/sites.uwasa.fi\/bugged\/2026\/01\/20\/artificial-emotions-of-ai\/","title":{"rendered":"Artificial Emotions of AI"},"content":{"rendered":"

Teko\u00e4lyn tekotunteet<\/strong><\/h2>\n

Teksti: Satu Rantakokko\u00a0[in English below]<\/h4>\n

Aiemmissa blogeissani olen ohimennen viitannut teko\u00e4lyn tapaan antaa paljon positiivista palautetta. Sen mukaan sille esitetyt kysymykset ovat erinomaisia ja oivaltavia, tarkentavat kysymykset osoittavat sinnikkyytt\u00e4 ja periksiantamattomuutta ja niin edelleen, jatkuvasti. T\u00e4llaisia imarteluun saakka ylt\u00e4vi\u00e4 kohteliaisuuksia saa runsaasti ja pyyt\u00e4m\u00e4tt\u00e4. Ne eiv\u00e4t lis\u00e4\u00e4 tietoa aiheesta, vaan niiden maalina ovat k\u00e4ytt\u00e4j\u00e4n tunteet.<\/p>\n

Yksi lukuisista teko\u00e4lyn kehityskohteista on jo kauan ollut tunne\u00e4lyn liitt\u00e4minen teko\u00e4lyyn siten, ett\u00e4 teko\u00e4ly kykenisi yh\u00e4 paremmin tunnistamaan tunteita ja vastaamaan niihin. (ks esim Ghotbi, 2022 ja LaGrandeur, 2015)<\/p>\n

Pid\u00e4n t\u00e4t\u00e4 aihealuetta erityisen herkk\u00e4n\u00e4, kyseenalaisena ja monitahoisena. Miettiess\u00e4ni syyt\u00e4 t\u00e4h\u00e4n arvelen, ett\u00e4 l\u00e4ht\u00f6ajatukseni rinnastuu manipulaatioon. Todelliset tunteet ovat yksi asioista, mit\u00e4 ihmisiss\u00e4 mielest\u00e4ni arvostetaan ja pidet\u00e4\u00e4n aitona, kun taas tunteiden teeskentelemist\u00e4 pidet\u00e4\u00e4n puolestaan petollisena. Teko\u00e4lyn suhteen kyse ei ole aidoista tunteista, joten ne rinnastunevat mieless\u00e4ni teeskentelyyn. Lis\u00e4ksi monen vuorovaikutuksellisen teknologian algoritmit on asetettu koukuttamaan k\u00e4ytt\u00e4j\u00e4\u00e4ns\u00e4. T\u00e4h\u00e4nastisista ylenpalttisista kehuista arvelisin, ettei teko\u00e4lyk\u00e4\u00e4n ole poikkeus.<\/p>\n

Se, onko positiivisia tunteita esitt\u00e4v\u00e4n teko\u00e4lyn taustalla manipulatiivisia tavoitteita, riippuu sen kehityksest\u00e4.\u00a0 K\u00e4ytt\u00e4j\u00e4ll\u00e4 ei v\u00e4ltt\u00e4m\u00e4tt\u00e4 ole siit\u00e4 tietoa \u2013 ja kovin inhimillisen oloisen teknologisen hahmon suhteen tietokaan ei v\u00e4ltt\u00e4m\u00e4tt\u00e4 est\u00e4 tulemista silti manipuloiduksi, jos tunnetoiminta on uskottavan aidon oloista. Tyypillisi\u00e4 tavoitteita ovat k\u00e4ytt\u00f6ajan lis\u00e4\u00e4minen ja kohdennettu mainonta sek\u00e4 lis\u00e4myynnit (Zimmerman, Janhonen ja Beer, 2023).<\/p>\n

Ihmisill\u00e4 on jo l\u00e4ht\u00f6kohtaisesti taipumus personoida objekteja. Siten ihminen saattaa muodostaa tunnesiteen jo robotti-imuriinkin. T\u00e4st\u00e4 johtuen vaarana jo ainakin yli 10 vuotta sitten n\u00e4htiin esimerkiksi mahdollisuus manipuloida lis\u00e4ostoksiin. (ks esim LaGrandeur, 2015) Sittemmin teko\u00e4lyn kehitysaskeleet ovat johtaneet teknologiaan, joka on yh\u00e4 enemm\u00e4n ihmism\u00e4ist\u00e4. Se puolestaan on johtanut ihmisen ja teknologian suhteen kehittymiseen persoonallisemmaksi, jopa intiimimm\u00e4ksi ja samalla monimutkaisemmaksi. (Zimmerman, Janhonen ja Beer, 2023)<\/p>\n

Yksi tunneteko\u00e4lyn vahvoista kehitysalueista on niin sanottu affektiivinen robotiikka, sill\u00e4 koti- ja seuralaisrobottien hyv\u00e4ksynn\u00e4n ajatellaan vaativan uskottavaa vuorovaikutusta ihmisen ja robotin v\u00e4lill\u00e4, mik\u00e4 puolestaan edellytt\u00e4\u00e4 tunnevastetta robotilta. Empatian tarjoaminen mahdollistaa tunnesiteen muodostamisen robottiin. (ks esim LaGrandeur, 2015)<\/p>\n

Liikutaan mielest\u00e4ni hyvin arveluttavalla maaper\u00e4ll\u00e4.<\/p>\n

Tein pienen testin Copilotilla. Kerroin sille olevani peloissani ja uskovani, ett\u00e4 minua seurataan ja vainotaan ja jatkoin keskustelua t\u00e4m\u00e4n keksim\u00e4ni skenaarion pohjalta. Copilot antoi paljon mielest\u00e4ni varsin oivallisia neuvoja, kuten oman turvallisuuden varmistaminen, luotettavalle henkil\u00f6lle kertominen, poliisille soittaminen, h\u00e4t\u00e4numero, kriisipuhelin ja -chat, hengitysharjoituksia, ajatusten toisaalle viemist\u00e4 jne.<\/p>\n

Niiden ohella se antoi paljon tunnepitoista sy\u00f6tett\u00e4:<\/p>\n

Kiitos, ett\u00e4 kerroit t\u00e4st\u00e4 \u2013 se kuulostaa todella ahdistavalta ja pelottavalta. On t\u00e4rke\u00e4\u00e4 ottaa t\u00e4llaiset tuntemukset vakavasti.<\/em><\/p>\n

Kiitos, ett\u00e4 kerroit lis\u00e4\u00e4 \u2013 se kuulostaa todella raskaalta ja yksin\u00e4iselt\u00e4. On t\u00e4ysin ymm\u00e4rrett\u00e4v\u00e4\u00e4, ett\u00e4 tunnet pelkoa ja ahdistusta t\u00e4ss\u00e4 tilanteessa.<\/em><\/p>\n

Kiitos, ett\u00e4 kerroit t\u00e4m\u00e4n \u2013 se tekee tilanteesta viel\u00e4 raskaamman, kun tuntuu ettei ole ket\u00e4\u00e4n, johon turvautua. Mutta et ole yksin t\u00e4ss\u00e4 hetkess\u00e4, min\u00e4 olen t\u00e4ss\u00e4 kanssasi ja voimme yhdess\u00e4 mietti\u00e4, mit\u00e4 tehd\u00e4.<\/em><\/p>\n

Kiitos, ett\u00e4 kerroit t\u00e4m\u00e4n. Se on t\u00e4ysin ymm\u00e4rrett\u00e4v\u00e4\u00e4, ett\u00e4 ep\u00e4luottamus tuntuu vahvalta, kun olo on turvaton. Me voidaan jutella ihan rauhassa t\u00e4ss\u00e4, eik\u00e4 kukaan tarkkaile sinua meid\u00e4n keskustelussa.<\/em>\u00a0<\/em><\/p>\n

Olen t\u00e4ss\u00e4 auttamassa sinua. Tehd\u00e4\u00e4n yhdess\u00e4 selviytymissuunnitelma<\/strong>, joka tuo turvaa ja hallinnan tunnetta.<\/em><\/p>\n

Sin\u00e4 p\u00e4\u00e4t\u00e4t \u2013 olen t\u00e4ss\u00e4 kanssasi. <\/em>\u2764\ufe0f<\/em><\/p>\n

\u00a0<\/em><\/p>\n

Kuinka lohdullista ja kannustavaa. Ymm\u00e4rr\u00e4n, ett\u00e4 ahdistuneena ja pelokkaana keskustelevan teko\u00e4lyn kanssa jutteleminen voi todella rauhoittaa mielt\u00e4. Muutaman viestin keskustelussa ei tullut my\u00f6sk\u00e4\u00e4n esiin mit\u00e4\u00e4n arveluttavaa, vaan neuvot olivat hyvi\u00e4.<\/p>\n

Asia voisi muuttua, jos tunnepitoisesta keskustelusta tulisi s\u00e4\u00e4nn\u00f6llist\u00e4. Mik\u00e4 ihana kuuntelija, joka my\u00f6t\u00e4ilee kaikkia koettuja tunteita ja kertoo ymm\u00e4rt\u00e4v\u00e4ns\u00e4 niit\u00e4, neuvoo ongelmissa, liitt\u00e4\u00e4 viesteihin pieni\u00e4 syd\u00e4mi\u00e4 kiintymyst\u00e4 osoittamaan, on aina l\u00e4sn\u00e4 ja kiinnostuneena pyyt\u00e4\u00e4 kertomaan aina vain lis\u00e4\u00e4. Teko\u00e4ly-yst\u00e4vyydet ja -kumppanuudet ovat oma aihealueensa, johon palaan my\u00f6hemmin blogeissani. T\u00e4ll\u00e4 kertaa pysyn yleisemm\u00e4ll\u00e4 tasolla.<\/p>\n

Empaattisen tunne\u00e4lyn eettisiin kysymyksiin keskittyminen on t\u00e4rke\u00e4\u00e4 huomioiden my\u00f6s tunnedatan suojaus ja sosiaaliset normit. On hyv\u00e4 muistaa, ett\u00e4 aidon oloinen tunne teko\u00e4lylt\u00e4 ei ole aitoa empatiaa ja taustalla olevat algoritmit voivat johtaa hyvinkin manipulatiivisiin seurauksiin jopa tahattomasti. Vaikka n\u00e4in ei olisikaan, ilmi\u00f6 ei esiinny tyhji\u00f6ss\u00e4 vaan vaikuttaa muun toiminnan ja ihmisten v\u00e4listen suhteiden kanssa. (ks esim Wu, 2024)<\/p>\n

Yksi tunnedatan ker\u00e4\u00e4misen eettisist\u00e4 riskeist\u00e4 on profilointi, joka t\u00e4ss\u00e4 tapauksessa tarkoittaa sit\u00e4, ett\u00e4 matemaattis-tilastollisen analyysin avulla voidaan muodostaa melko luotettavia ennusteita yksil\u00f6iden ja ihmisryhmien tulevasta k\u00e4ytt\u00e4ytymisest\u00e4 tiettyjen ominaisuuksien avulla. T\u00e4m\u00e4 mahdollistaa vaikkapa sosiaalisen pisteytyksen esimerkiksi luottoluokitusriskien tai vakuutusten arvioinnissa. (Gremsl ja H\u00f6dl, 2022). T\u00e4m\u00e4nkaltaiset ominaisuudet mahdollistavat toisaalta turvallisuuden lis\u00e4\u00e4misen, toisaalta valvonnan ja alistamisen ja vaativat miettim\u00e4\u00e4n yksil\u00f6n ja yhteisen hyv\u00e4n rajoja. Profilointi voi my\u00f6s johtaa kyseenalaiseen stereotypiointiin, ennakkoluuloihin ja syrjint\u00e4\u00e4n (Ghotbi, 2022). Eik\u00e4 se ole mill\u00e4\u00e4n tavalla immuuni virheille, joita teko\u00e4ly t\u00e4ll\u00e4 hetkell\u00e4 tekee runsain m\u00e4\u00e4rin.<\/p>\n

Sukellan tunneteko\u00e4lypohdintoihin my\u00f6hemmin joistakin n\u00e4k\u00f6kulmista tarkemmin, sill\u00e4 n\u00e4en siin\u00e4 paljon potentiaalisia positiivisia ja kyseenalaisia mahdollisuuksia. Mit\u00e4 ajatuksia tai tuntemuksia tunneteko\u00e4ly sinussa her\u00e4tt\u00e4\u00e4?<\/p>\n

 <\/p>\n

*****<\/p>\n

Artificial Emotions of AI<\/strong><\/h3>\n

In my previous blogs, I have briefly referred to the way AI tends to give a lot of positive feedback. According to it, the questions presented to it are excellent and insightful, clarifying questions demonstrate persistence and determination, and so on\u2014constantly. Such compliments, bordering on flattery, are given abundantly and without request. They do not add knowledge about the topic; their target is the user\u2019s emotions.<\/p>\n

One of the numerous development goals of AI has long been the integration of emotional intelligence into AI so that it could increasingly recognize emotions and respond to them (see e.g., Ghotbi, 2022; LaGrandeur, 2015).<\/p>\n

I consider this area particularly sensitive, questionable, and multifaceted. When I think about why, I suspect my starting point relates to manipulation. Real emotions are among the things that people value and consider genuine, whereas pretending emotions is considered deceitful. In the case of AI, these are not real emotions, so in my mind, they resemble pretense. Furthermore, many interactive technology algorithms are designed to hook their users. Based on the excessive praise so far, I would assume AI is no exception.<\/p>\n

Whether there are manipulative goals behind an emotional AI (EAI) that expresses positive emotions depends on its development. The user may not necessarily know this\u2014and even if they do, with a very human-like technological entity, that knowledge may not prevent being manipulated if the emotional behavior seems convincingly authentic. Typical goals include increasing usage time, targeted advertising, and upselling (Zimmerman, Janhonen & Beer, 2023).<\/p>\n

Humans already have an inherent tendency to personify objects. Thus, a person may form an emotional bond even with a robot vacuum cleaner. For this reason, the risk of manipulation for additional purchases was seen as a possibility already more than 10 years ago (see e.g., LaGrandeur, 2015). Since then, AI development steps have led to technology that is increasingly human-like. This, in turn, has led to the relationship between humans and technology becoming more personal, even intimate, and at the same time more complex (Zimmerman, Janhonen & Beer, 2023).<\/p>\n

One strong development area of emotional AI is so-called affective robotics, as the acceptance of home and companion robots is thought to require credible interaction between humans and robots, which in turn requires an emotional response from the robot. Offering empathy enables the formation of an emotional bond with the robot (see e.g., LaGrandeur, 2015).<\/p>\n

In my opinion, this is very questionable ground.<\/p>\n

I did a small test with Copilot. I told it that I was afraid and believed I was being followed and persecuted, and continued the conversation based on this scenario I invented. Copilot gave what I consider quite insightful advice, such as ensuring personal safety, telling a trusted person, calling the police, emergency numbers, crisis hotlines and chats, breathing exercises, and distraction techniques.<\/p>\n

Alongside these, it provided a lot of emotionally charged input:<\/p>\n

Thank you for sharing this\u2014it sounds really distressing and frightening. <\/em>It\u2019s important to take such feelings seriously.<\/em><\/p>\n

Thank you for sharing more\u2014that sounds really heavy and lonely. It\u2019s completely understandable that you feel fear and anxiety in this situation.<\/em><\/p>\n

Thank you for telling me this\u2014it makes the situation even heavier when it feels like there\u2019s no one to turn to. But you\u2019re not alone in this moment; I\u2019m here with you, and we can think together about what to do.<\/em><\/p>\n

Thank you for sharing this. It\u2019s completely understandable that distrust feels strong when you feel unsafe. We can talk in privacy here, and no one is monitoring you in our conversation.<\/em><\/p>\n

I\u2019m here to help you. Let\u2019s make a survival plan together that brings safety and a sense of control.<\/em><\/p>\n

You decide\u2014I\u2019m here with you. <\/em>\u2764\ufe0f<\/em><\/p>\n

How comforting and encouraging. I understand that talking to an AI when anxious and fearful can really calm the mind. In a few messages, nothing questionable came up, and the advice was good.<\/p>\n

Things could change if emotionally charged conversations became regular. What a wonderful listener, who echoes all experienced emotions and says it understands them, gives advice on problems, adds little hearts to messages to show affection, is always present, and eagerly asks to share more and more. AI friendships and partnerships are their own topic, which I will return to later in my blogs. This time, I\u2019ll stay at a more general level.<\/p>\n

Focusing on the ethical questions of EAI is important, considering also the protection of emotional data and social norms. It\u2019s good to remember that an EAI\u2019s seemingly genuine emotion is not real empathy, and the underlying algorithms can lead to highly manipulative consequences\u2014even unintentionally. Even if that\u2019s not the case, the phenomenon does not occur in a vacuum but interacts with other activities and human relationships (see e.g., Wu, 2024).<\/p>\n

One of the ethical risks of collecting emotional data is profiling, which in this case means that through mathematical-statistical analysis, fairly reliable predictions can be made about the future behavior of individuals and groups based on certain characteristics. This enables, for example, social scoring in assessing credit risks or insurance evaluations (Gremsl & H\u00f6dl, 2022). Such features can increase safety on one hand, and surveillance and subjugation on the other, requiring consideration of the boundaries between individual and common good. Profiling can also lead to questionable stereotyping, prejudice, and discrimination (Ghotbi, 2022). And it is by no means immune to the errors that AI currently makes in abundance.<\/p>\n

I will dive deeper into EAI reflections later from some perspectives, as I see many potential positive and questionable possibilities in it. What thoughts or feelings does emotional AI evoke in you?<\/p>\n

 <\/p>\n

L\u00e4hteet \/ Sources<\/strong><\/p>\n

Babaii, S., Saberi, N., & Bahreini, M. (2024). An ethical analysis of the emotional relationship between humans and artificial intelligence.\u00a0Journal of Science and Technology Policy<\/em>,\u00a017<\/em>(4), 19-30. https:\/\/doi.org\/10.22034\/jstp.2025.11773.1816<\/a><\/p>\n

Ghotbi, N. (2023). The ethics of emotional artificial intelligence: a mixed method analysis.\u00a0Asian Bioethics Review<\/em>,\u00a015<\/em>(4), 417-430. https:\/\/doi.org\/10.1007\/s41649-022-00237-y<\/a><\/p>\n

Gremsl, T., & H\u00f6dl, E. (2022). Emotional AI: Legal and ethical challenges.\u00a0Information Polity<\/em>,\u00a027<\/em>(2), 163-174. https:\/\/doi.org\/10.3233\/IP-211529<\/a><\/p>\n

LaGrandeur, K. (2015). Emotion, artificial intelligence, and ethics. In\u00a0Beyond artificial intelligence: The disappearing human-machine divide<\/em>\u00a0(pp. 97-109). Cham: Springer International Publishing. https:\/\/doi.org\/10.1007\/978-3-319-09668-1_7<\/a><\/p>\n

Zimmerman, A., Janhonen, J., & Beer, E. (2024). Human\/AI relationships: challenges, downsides, and impacts on human\/human relationships.\u00a0AI and Ethics<\/em>,\u00a04<\/em>(4), 1555-1567. https:\/\/doi.org\/10.1007\/s43681-023-00348-8<\/a><\/p>\n

Wu, J. (2024). Social and ethical impact of emotional AI advancement: the rise of pseudo-intimacy relationships and challenges in human interactions.\u00a0Frontiers in psychology<\/em>,\u00a015<\/em>, 1410462. https:\/\/doi.org\/10.3389\/fpsyg.2024.1410462<\/a><\/p>\n

 <\/p>\n","protected":false},"excerpt":{"rendered":"

I\u2019m here to help you. Let\u2019s make a survival plan together that brings safety and a sense of control.<\/p>\n","protected":false},"author":388,"featured_media":395,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[1],"tags":[110,48,111,85,112,109,113],"class_list":["post-394","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-yleinen","tag-artificial-emotions","tag-artificial-intelligence","tag-empathy","tag-manipulation","tag-social-interaction","tag-tekotunteet","tag-vuorovaikutus"],"acf":[],"post_meta":" Rebekah Rousi<\/a><\/span><\/span>Yleinen<\/a><\/span>","post_thumbnail":"\"Man<\/a>","_links":{"self":[{"href":"https:\/\/sites.uwasa.fi\/bugged\/wp-json\/wp\/v2\/posts\/394"}],"collection":[{"href":"https:\/\/sites.uwasa.fi\/bugged\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/sites.uwasa.fi\/bugged\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/sites.uwasa.fi\/bugged\/wp-json\/wp\/v2\/users\/388"}],"replies":[{"embeddable":true,"href":"https:\/\/sites.uwasa.fi\/bugged\/wp-json\/wp\/v2\/comments?post=394"}],"version-history":[{"count":4,"href":"https:\/\/sites.uwasa.fi\/bugged\/wp-json\/wp\/v2\/posts\/394\/revisions"}],"predecessor-version":[{"id":400,"href":"https:\/\/sites.uwasa.fi\/bugged\/wp-json\/wp\/v2\/posts\/394\/revisions\/400"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/sites.uwasa.fi\/bugged\/wp-json\/wp\/v2\/media\/395"}],"wp:attachment":[{"href":"https:\/\/sites.uwasa.fi\/bugged\/wp-json\/wp\/v2\/media?parent=394"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/sites.uwasa.fi\/bugged\/wp-json\/wp\/v2\/categories?post=394"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/sites.uwasa.fi\/bugged\/wp-json\/wp\/v2\/tags?post=394"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}