- 03/09/2024
- Category: Commentaries
Author: Ayom Mratita Purbandani
Editor: Hafiz Noer
Artificial intelligence (AI) has rapidly advanced in the last decade, integrating into more and more fields. The emergence of hysteria in response to AI is not unexpected, as it happens with any widely adopted technology. As a result of this hysteria, some people anticipate that AI will eventually replace human labor by emulating or even surpassing human intelligence (Brock, 2019). As a response to this concern, AI orientation has entered a new phase, focusing on a more humane approach. As a solution for designing technology that gives humans control over it, Human-Centered Artificial Intelligence (HCAI) has been proposed (Capel & Brereton, 2023).
The development of HCAI involves designing AI with human needs, values, and capabilities in mind (Shneiderman, 2020). AI should be aligned with humanistic values in order to produce “beneficial intelligence.” HCAI concludes that AI must remain relevant by aligning with humanistic values. Humanism as a concept, relies on humanist notions of ‘human nature’ as a basis for ethics, morals, and political status (Roden, 2015). This reliance on humanism, however, is ironic since HCAI fails to clarify what humanity represents. Humanity is problematic as it implies a binary, dominating relationship between humans and non-humans (Plumwood, 1993). This article will highlight the irony of the HCAI approach, which has introduced or reinforced a new “other” through the intersection of science, technology, and the humanities (STS).
What’s wrong with human-centered AI?
HCAI envisions a technological future that better meets human values and needs. Ben Shneiderman (2022: 4) asserts that HCAI offers superhuman capabilities, enhances human creativity, and enhances human performance and self-efficacy. In an example, he refers to digital cameras that clearly demonstrate human control, yet many AI features assist in setting the aperture, adjusting focus, and reducing vibration. With humanism as its guiding principle, HCAI emphasizes human agency.
HCAI’s goal of aligning AI with human needs seems visionary in today’s increasingly automated world. This, however, reflects an anthropocentric vision of human character. The separation between humans and everything else is deliberate, granting humans a special status and separating them from everything else (Plumwood, 1993). Hence, humans are framed as having special insights, the authority to command, lead, and control others. By performing HCAI, human goals are ultimately achieved through the instrumentalization of nonhumans (Ernst et al., 2019).
History shows that human separation from others has never been neutral. There has always been an associated matrix of dominance with this separation. Whether philosophically or politically, binary divisions such as mind/body, human/non-human, or human/machine always reflect hierarchical positions, with some considered superior and others inferior (Naess, 2008). These hierarchical positions are based on the concept of humanism, which defines intelligence as the determining factor. As a technology design based on humanism, HCAI emphasizes this determining factor and potentially harbors dominance tendencies. Donna Haraway (2016) uses the phrase “almost-minds” to describe how AI and machine technology are characterized. Using this phrase, she reminds us that the same characteristic (almost-minds) has historically been used to oppress others (women, indigenous people, people of color).
Human-centeredness perpetuates inequality. In this regard, HCAI falls short. As much as it attempts to formulate a more ‘ethical’ AI, HCAI’s focus on humanism fails to question the inherent logic of dominance. The instrumentalization of AI and the instrumentalization of techno-scientific regulation thus become normalized. The discourse was normalized to the point that the way HCAI reinforces human dominance over non-humans was not apparent. In contrast, AI is positioned as a technology able to deal with a variety of social problems. As a result, AI is never re-examined if it facilitates decisions that harm non-humans, such as in the extractive industries.
‘Beyond Human’ framework
Unresolved issues in HCAI compel us to seek better perspectives on intelligence to underpin technological advancement. By proposing a plural intelligence transcending human parameters, a posthumanist approach to AI might shift the command seat away from humans. Rather than be confined to human species, posthumanism advocates an imaginative connection among living beings (Banerji, 2016). As James Bridle points out in Animals, Plants, Machines: The Search for a Planetary Intelligence (2022), AI needs to be reconceptualized to express the movement of living beings. Therefore, AI should no longer be considered as a replacement or superior form of intelligence, but rather as a part of a broader ecological system.
Posthumanists challenge anthropocentric views by recognizing intelligence within and across species (Wilson, 2023). Thus, intelligence is not restricted to the minds of humans or machines, but is found in plants, animals, fungi, and larger ecological systems (Bridle, 2022). This approach advocates for a broader and more inclusive understanding of intelligence that acknowledges its existence in all forms of life and their interconnectedness. In order for design and technology to be more holistic, multi-sensory, and interconnected, posthumanism challenges the dominant view of AI. Posthumanist perspectives and values fundamentally challenge human perspectives in AI governance. A posthumanist approach helps us envision a world in which AI and other forms of posthuman intelligence evolve beyond human capabilities. Based on this perspective, human-centered values and control mechanisms are no longer relevant due to non-human governance (Roden, 2015). With such a shift, the concept of “human” would be redefined, moving away from biological and cognitive markers and towards a more non-anthropocentric view. As part of this design transition, not only engineers and scientists are encouraged to collaborate on AI development but also artists, ecologists, and indigenous groups (Lewis et al., 2018). As a result, we will be able to create technologies that are aligned with diverse life systems, leading to a greater understanding of plural intelligence.
