AI translation has become a core tool in the language service industry by virtue of its efficiency and automation advantages, but the limitations in the realization of its pragmatic function lead to significant deviation from human translation, which restricts the accuracy and adaptability of cross-cultural communication. In this regard, the study of the deviation between AI translation and human translation under the perspective of pragmatic function is proposed. Firstly, the essence of pragmatic function is defined as the ability of language to realize communicative intentions in specific contexts, and by comparing the logic of AI translation with that of human translation, it is clarified that human translators are able to dynamically adjust their translation strategies to achieve functional equivalence through cognitive reasoning and cross-cultural awareness. It is pointed out that the difference between AI translation and human translation in context adaptation is essentially the path difference between data-driven static mapping and cognitive-driven dynamic understanding, which is also reflected in the path choice between explicit symbol matching and implicit functional reconstruction. In terms of cultural values integration, AI translation focuses more on symbol transfer while human translation can significantly improve the cross-cultural adaptability of translation through contextual weight calibration and other operations. Based on this, we propose a collaborative framework of "AI initial screening context association+ human calibration function adaptation", construct a translation paradigm oriented by functional equivalence, and realize the deep complementarity between human and machine at the level of pragmatic function through the hierarchical context model, the intention classification system, and the cultural-functional transcoding protocol, with the aim of providing theoretical support for the upgrading of AI translation technology and innovation of the collaboration model of translation industry between human and machine. It aims to provide theoretical support for the upgrading of AI translation technology and the innovation of translation industry human-machine cooperative mode.

Li Shuke. A Study on the Deviation between AI Translation and Human Translation from the Perspective of Pragmatic Functions. Academic Journal of Humanities & Social Sciences (2025), Vol. 8, Issue 12: 88-94. https://doi.org/10.25236/AJHSS.2025.081214.

[1] Shein S S .The Study of the Language Nature of Dawei (Tavoyan) and Myanmar Language (Burmese) using Statistical Machine Translation[J].2023 IEEE Conference on Computer Applications (ICCA), 2023:11-15. 

[2] Dhomne S, Chandak M B, Raipurkar A R S .A LAYERED APPROACH OF MACHINE TRANSLATION USING TRANSLATION MEMORY ON EDGE COMPUTING[J].journal of theoretical and applied information technology, 2024, 102(7):3103-3110.

[3] Dang P .The extraction method used for English-Chinese machine translation corpus based on bilingual sentence pair coverage[J].Open Computer Science, 2024, 14(1):pp. 1166-1180. 

[4] Jailson B Q, Masaaki S,Díaz-López IreneGordiyenko YuliyaFraser Christopher S.Ramakrishnan V. The structure of a human translation initiation complex reveals two independent roles for the helicase eIF4A[J].Nature structural & molecular biology, 2024, 31(3):455-464. 

[5] Guo Y, Liu T, Zhang A W W .End-to-end translation of human neural activity to speech with a dual-dual generative adversarial network[J].Knowledge- based systems, 2023, 277(Oct.9):1.1-1.11. 

[6] Vorontsov K V, Skachkov N A .Reranking Hypotheses in Translation Models Using Human Markup[J]. International, 2024, 63(4):679-686. 

[7] Manapbayeva Z, Zaurbekova G, Ayazbekova K,et al. AI in Literary Translation: ChatGPT-4 vs. Professional Human Translation of Abai's Poem 'Spring' [J].Procedia Computer Science, 2024, 251:526-531. 

[8] Saurer M, Leibundgut M, Nadimpalli H,et al. Molecular basis of translation termination at noncanonical stop codons in human mitochondria[J]. Science, 2023, 380:531 - 536. 

[9] Saha M B P,ROY, A. S, Saha, P,et al.Stable RNA G-Quadruplex in the 5’-UTR of Human cIAP1 mRNA Promotes Translation in an IRES-Independent Manner[J].Biochemistry, 2024, 63(4):475-486. 

[10] Mcnee A, Kannan A, Jull P,et al. Expanding Human Breg for Cellular Therapy in Transplantation: Time for Translation[J].Transplantation, 2024, 109 (6):926-937. 

[11] Huang D, Xiang S .Speech recognition and intelligent translation under multimodal human-computer interaction system[J]. Journal of Intelligent Systems, 2024, 33(1):798-810. 

[12] Kruger A, Kovalchuk D, Shiriaev D R J .Decoding the Enigma: Translation Termination in Human Mitochondria[J].Human Molecular Genetics, 2024, 33( R1):R42-R46. 

[13] Shuvalov A V, Klishin A A, Biziaev N S,et al. Human eRF1 Translation Regulation[J].Molecular Biology, 2024, 58(4):708-717. 

[14] Nurdenara C, Al M W F .Image translation between human face and wayang orang using U-GAT-IT[J].IAES International Journal of Artificial Intelligence, 2024, 13(2):2451-2458.

[15] Lewis C J T, Xie L H, Bhandarkar S M,et al. Quantitative profiling of human translation initiation reveals elements that potently regulate endogenous and therapeutically modified mRNAs[J].Molecular Cell, 2025, 85(2):445-459.e5.