The human is not a means to an output. The human is the point.

In my previous post I wrote that if we define human value as whatever today’s system cannot do, we will be forced to retreat each time the capability boundary moves [1]. Since then I have discussed the question with friends, colleagues, and family, and one reply kept returning. In many activities, they said, the human is not a means to an output. The human is the point.

Chess is the cleanest example. Deep Blue defeated Kasparov in 1997 [2], and no human has been competitive with the strongest engines for two decades. Yet human chess did not shrink. Chess.com passed 100 million members in December 2022 and 250 million in February 2026 [3]. Engines serve as tools and commentators, but the games people watch are Carlsen against Nakamura, not one engine against another. In April 2025, humanoid robots ran the Beijing half-marathon alongside people. The fastest robot finished in 2:40:42, the fastest human in 1:02, and the robots were followed as a curiosity rather than as rivals [4]. Bernard Suits defined playing a game as the voluntary attempt to overcome unnecessary obstacles [5]. We watch because a person is attempting.

Live performance is similar. Baumol and Bowen observed in 1966 that performance resists productivity gains because the labor is the product [6]. Recorded music has become nearly free, and people still pay to be in the room. The preference concerns the source, not only the product. Listeners rated the same classical excerpts lower when told an AI had composed them [7], and images carrying an AI label were judged less profound and of lower worth than identical images labeled human-made [8]. Twenty professional comedians who tried language models in their writing described the output as bland, “cruise ship comedy material from the 1950s,” lacking the lived perspective from which a joke is told [9].

Kindness and empathy follow the same pattern in a sharper form. When evaluators were blinded to the source, chatbot replies to patient questions were rated more empathetic than physicians’ replies [10]. But across nine studies with more than six thousand participants, identical messages were rated more empathic and supportive when attributed to a human, and even the belief that an AI had assisted reduced the perceived support [11]. Consumers pay more for handmade goods because they believe the goods contain the maker’s love [12]. Empathy from a person costs the person something. That cost is part of what is received.

Two caveats. These are label effects, and label effects can fade; the art study found the bias weaker among people with positive attitudes toward AI [8]. And most of the economy buys outputs, not attempts, so the human touch will not shelter every occupation.

This matters for my earlier argument. Responsibility was one anchor that does not move with capability. Meaning is another. Where a human presence is constitutive of an activity, a better machine does not force a retreat. The boundary it moves is not the one that matters.

Drafting and AI-use note. AI was used to generate the figure and search for the references.

References

[1] K. Bharti, “What Should the Human Part of Science Become?” Contextual Stories, August 15, 2026. https://contextual-stories.blogspot.com/2026/08/what-should-human-part-of-science-become.html.

[2] M. Campbell, A. J. Hoane Jr., and F.-h. Hsu, “Deep Blue,” Artificial Intelligence 134, 57–83 (2002). https://doi.org/10.1016/S0004-3702(01)00129-1.

[3] Chess.com, “Chess.com Reaches 250 Million Members,” February 27, 2026. https://www.chess.com/news/view/chesscom-reaches-250-million-members.

[4] M. Bassi, “Humanoid Robots Just Raced Alongside Human Runners in a World-First Half-Marathon. Here’s How It Went,” Smithsonian Magazine, April 21, 2025. https://www.smithsonianmag.com/smart-news/humanoid-robots-just-raced-alongside-human-runners-in-a-world-first-half-marathon-heres-how-it-went-180986469/.

[5] B. Suits, The Grasshopper: Games, Life and Utopia (University of Toronto Press, 1978; 3rd ed., Broadview Press, 2014).

[6] W. J. Baumol and W. G. Bowen, Performing Arts: The Economic Dilemma (Twentieth Century Fund, 1966).

[7] D. B. Shank, C. Stefanik, C. Stuhlsatz, K. Kacirek, and A. M. Belfi, “AI composer bias: Listeners like music less when they think it was composed by an AI,” Journal of Experimental Psychology: Applied 29, 676–692 (2023). https://doi.org/10.1037/xap0000447.

[8] L. Bellaiche, R. Shahi, M. H. Turpin, A. Ragnhildstveit, S. Sprockett, N. Barr, A. Christensen, and P. Seli, “Humans versus AI: whether and why we prefer human-created compared to AI-created artwork,” Cognitive Research: Principles and Implications 8, 42 (2023). https://doi.org/10.1186/s41235-023-00499-6.

[9] P. W. Mirowski, J. Love, K. Mathewson, and S. Mohamed, “A Robot Walks into a Bar: Can Language Models Serve as Creativity Support Tools for Comedy? An Evaluation of LLMs’ Humour Alignment with Comedians,” in Proceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency (FAccT ’24), 2024. https://doi.org/10.1145/3630106.3658993.

[10] J. W. Ayers, A. Poliak, M. Dredze, et al., “Comparing Physician and Artificial Intelligence Chatbot Responses to Patient Questions Posted to a Public Social Media Forum,” JAMA Internal Medicine 183, 589–596 (2023). https://doi.org/10.1001/jamainternmed.2023.1838.

[11] M. Rubin, J. Z. Li, F. Zimmerman, D. C. Ong, A. Goldenberg, and A. Perry, “Comparing the value of perceived human versus AI-generated empathy,” Nature Human Behaviour 9, 2345–2359 (2025). https://doi.org/10.1038/s41562-025-02247-w.

[12] C. Fuchs, M. Schreier, and S. M. J. van Osselaer, “The Handmade Effect: What’s Love Got to Do with It?” Journal of Marketing 79(2), 98–110 (2015). https://doi.org/10.1509/jm.14.0018.