Razumijevanje umjetne inteligencije: Neki korelati stavova prema umjetnoj inteligenciji i umjetna inteligencija kao alat u procjeni ličnosti

• Post by: Psyche General
• 19.08.2024.
• Comments off

Sažetak

Razvoj umjetne inteligencije (engl. artificial intelligence, AI) doveo je do povećanja broja zadataka koji ona može obavljati. No uključuju li njezine sposobnosti i metodološku kompetenciju procjene ljudske ličnosti unutar petofaktorskog modela? Ovo istraživanje, uz anketu i upitnik proveden na uzorku od 423 osobe, imalo je četiri cilja. Primarni je bio ispitati faktorsku strukturu upitnika ličnosti koji je napravio algoritam umjetne inteligencije ChatGPT-a po petofaktorskom modelu ličnosti. Ispitana je i konvergentna valjanost
ChatGPT-ova upitnika usporedbom s upitnikom ličnosti koji je nastao na temelju pitanja iz IPIP-a. Istraživanje je ispitivalo relevantne osobine ličnosti povezane sa stavovima o umjetnoj inteligenciji. Konačno, ispitana je mogućnost razlikovanja autorstva čestica iz upitnika, odnosno mogu li sudionici razlučiti koje je čestice napisao čovjek, a koje umjetna inteligencija. Rezultati pokazuju na nezadovoljavajuću faktorsku valjanost IPIP i ChatGPT te umjerenu do visoku pozitivnu korelaciju pripadajućih faktora IPIP-ova i ChatGPT-ova upitnika, što ukazuje na zadovoljavajuću konvergentnu valjanost. Iako postojeća literatura upućuje na to da bi ekstraverzija i otvorenost mogle biti povezane s
prepoznavanjem korisnosti AI-a, a neuroticizam, ekstraverzija, ugodnost i savjesnost sa zanemarivanjem opasnosti AI-a, istraživanje ne pronalazi značajnost većine pojedinih prediktora, niti značajan učinak na razini bloka. Istraživanje pokazuje da autorstvo čestica točnije procjenjuju osobe ženskog roda, osobe više na neuroticizmu i osobe koje se više koriste tehnologijom u privatne svrhe.

Ključne riječi: ChatGPT, ličnost, stavovi prema umjetnoj inteligenciji, umjetna inteligencija

Abstract

The development of artificial intelligence (AI) has led to a sharp increase in the number of tasks it can do. Do its abilities include the methodological competence to assess personality under the five-factor model (FFM)? This research, a survey and questionnaire conducted on 423 persons, had four goals. The primary goal was to compare the factor validity of a personality measure created by the ChatGPT algorithm
under the theoretical framework of the FFM model with a personality measure created with the IPIP item base and to study the correlations between the two personality measures. Another was to examine the relevant personality traits correlated with attitudes towards AI, and to test the participants’ ability to differentiate human-created items from ChatGPT-created items. The factor validity of the two measures was not satisfactory, but the five corresponding factors from the two measures are correlated positively moderately to high. Although existing literature suggests extraversion and openness to experience to be correlated to recognizing the utility of AI, and neuroticism, extraversion, agreeableness and conscientiousness to be correlated with readiness to ignore potential dystopian dangers of AI, this research found almost no significant individual effects. There were also no effects of personality significant on a block scale. It has been found that gender, neuroticism and private technology use are correlated to the ability to differentiate human-created items from ChatGPT-created items. Women,
more neurotic participants, and those who used technology privately more often made more accurate judgements.

Keywords: artificial intelligence, attitudes towards AI, ChatGPT, personality