{ "model": "gemini-3.8-flash", "experiment": "shaina_fresh_test", "tasks": { "t1": { "scores": { "n": 10, "accuracy": 0.4, "ci95": [ 0.1682, 0.6873 ], "balanced_accuracy": 0.4167, "cohen_kappa": 0.2405 }, "items": [ { "id": "t1_RVFA__seed_vc__153f__main__id00460_KCBOXmDu5Ks_00025__id08352", "clips": [ "RVFA__seed_vc__153f__main__id00460_KCBOXmDu5Ks_00025__id08352.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "neither", "both", "the video", "the audio" ], "gt_answer": "the audio", "pred_answer": "neither", "correct": false, "raw": "```json\n{\"answer\": \"neither\"}\n```" }, { "id": "t1_RVFA__seed_vc__77f__main__id01296_B7gb-0xbMn0_00024__id06068", "clips": [ "RVFA__seed_vc__77f__main__id01296_B7gb-0xbMn0_00024__id06068.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "the audio", "both", "the video", "neither" ], "gt_answer": "the audio", "pred_answer": "the audio", "correct": true, "raw": "```json\n{\"answer\": \"the audio\"}\n```" }, { "id": "t1_RVRA__real__77f__vocoded__id05623_HbnUHzNq8XI_00003__id00739", "clips": [ "RVRA__real__77f__vocoded__id05623_HbnUHzNq8XI_00003__id00739.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "neither", "both", "the audio", "the video" ], "gt_answer": "neither", "pred_answer": "neither", "correct": true, "raw": "```json\n{\"answer\": \"neither\"}\n```" }, { "id": "t1_RVRA__real__153f__vocoded__id02959_hp-EccEnxCQ_00026__id08803", "clips": [ "RVRA__real__153f__vocoded__id02959_hp-EccEnxCQ_00026__id08803.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "the video", "both", "neither", "the audio" ], "gt_answer": "neither", "pred_answer": "neither", "correct": true, "raw": "```json\n{\"answer\": \"neither\"}\n```" }, { "id": "t1_FVFA__s2v__153f__main__id03084_BQIuTs2QZUU_00021__id08748", "clips": [ "FVFA__s2v__153f__main__id03084_BQIuTs2QZUU_00021__id08748.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "the video", "both", "neither", "the audio" ], "gt_answer": "both", "pred_answer": "the video", "correct": false, "raw": "```json\n{\"answer\": \"the video\"}\n```" }, { "id": "t1_FVFA__animate__77f__main__id05162_NpL77H2Shgk_00043__id03606", "clips": [ "FVFA__animate__77f__main__id05162_NpL77H2Shgk_00043__id03606.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "the audio", "neither", "both", "the video" ], "gt_answer": "both", "pred_answer": "neither", "correct": false, "raw": "```json\n{\"answer\": \"neither\"}\n```" }, { "id": "t1_FVFA__s2v__77f__main__id06390_17l4_o3k29c_00007__id07683", "clips": [ "FVFA__s2v__77f__main__id06390_17l4_o3k29c_00007__id07683.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "neither", "both", "the audio", "the video" ], "gt_answer": "both", "pred_answer": "neither", "correct": false, "raw": "```json\n{\"answer\": \"neither\"}\n```" }, { "id": "t1_FVFA__facefusion__153f__main__id07786_bFZAH657cFs_00051__id07683", "clips": [ "FVFA__facefusion__153f__main__id07786_bFZAH657cFs_00051__id07683.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "neither", "both", "the audio", "the video" ], "gt_answer": "both", "pred_answer": "the audio", "correct": false, "raw": "```json\n{\"answer\": \"the audio\"}\n```" }, { "id": "t1_FVFA__facefusion__77f__main__id05670_8I_YYqRsihM_00041__id04900", "clips": [ "FVFA__facefusion__77f__main__id05670_8I_YYqRsihM_00041__id04900.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "the video", "both", "neither", "the audio" ], "gt_answer": "both", "pred_answer": "neither", "correct": false, "raw": "```json\n{\"answer\": \"neither\"}\n```" }, { "id": "t1_RVFA__seed_vc__77f__main__id01858_HBSDURviflg_00015__id01156", "clips": [ "RVFA__seed_vc__77f__main__id01858_HBSDURviflg_00015__id01156.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "neither", "the audio", "both", "the video" ], "gt_answer": "the audio", "pred_answer": "the audio", "correct": true, "raw": "```json\n{\"answer\": \"the audio\"}\n```" } ] }, "t2": { "scores": { "n": 10, "accuracy": 0.2, "ci95": [ 0.0567, 0.5098 ], "balanced_accuracy": 0.25, "cohen_kappa": 0.0 }, "items": [ { "id": "t2_FVRA__s2v__77f__main__id08406_qr5lYg7ej40_00213__id05444", "clips": [ "FVRA__s2v__77f__main__id08406_qr5lYg7ej40_00213__id05444.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "full-body swap", "fully generated", "real", "face swap" ], "gt_answer": "fully generated", "pred_answer": "real", "correct": false, "raw": "```json\n{\"answer\": \"real\"}\n```" }, { "id": "t2_FVRA__facefusion__77f__main__id08940_B5cvspWteMI_00009__id03946", "clips": [ "FVRA__facefusion__77f__main__id08940_B5cvspWteMI_00009__id03946.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "face swap", "full-body swap", "fully generated", "real" ], "gt_answer": "face swap", "pred_answer": "face swap", "correct": true, "raw": "```json\n{\"answer\": \"face swap\"}\n```" }, { "id": "t2_RVRA__real__77f__main__id05444_QIxNYNGaVg0_00040__id08406", "clips": [ "RVRA__real__77f__main__id05444_QIxNYNGaVg0_00040__id08406.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "full-body swap", "face swap", "real", "fully generated" ], "gt_answer": "real", "pred_answer": "face swap", "correct": false, "raw": "{\"answer\": \"face swap\"}" }, { "id": "t2_FVRA__animate__77f__main__id07054_gLIAqAQ40LQ_00053__id08610", "clips": [ "FVRA__animate__77f__main__id07054_gLIAqAQ40LQ_00053__id08610.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "fully generated", "real", "full-body swap", "face swap" ], "gt_answer": "full-body swap", "pred_answer": "face swap", "correct": false, "raw": "{\"answer\": \"face swap\"}" }, { "id": "t2_RVRA__real__77f__main__id00761_C_ZBK3E8bJk_00027__id07616", "clips": [ "RVRA__real__77f__main__id00761_C_ZBK3E8bJk_00027__id07616.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "real", "fully generated", "face swap", "full-body swap" ], "gt_answer": "real", "pred_answer": "real", "correct": true, "raw": "```json\n{\"answer\": \"real\"}\n```" }, { "id": "t2_FVRA__facefusion__77f__main__id06480_YQD4AsloKcY_00076__id01534", "clips": [ "FVRA__facefusion__77f__main__id06480_YQD4AsloKcY_00076__id01534.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "real", "face swap", "fully generated", "full-body swap" ], "gt_answer": "face swap", "pred_answer": "real", "correct": false, "raw": "```json\n{\"answer\": \"real\"}\n```" }, { "id": "t2_FVRA__s2v__77f__main__id03854_yfQGoMSBDx0_00151__id00509", "clips": [ "FVRA__s2v__77f__main__id03854_yfQGoMSBDx0_00151__id00509.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "face swap", "full-body swap", "real", "fully generated" ], "gt_answer": "fully generated", "pred_answer": "face swap", "correct": false, "raw": "{\"answer\": \"face swap\"}" }, { "id": "t2_FVRA__animate__77f__main__id02959_h9Qwwv-DApg_00016__id03078", "clips": [ "FVRA__animate__77f__main__id02959_h9Qwwv-DApg_00016__id03078.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "face swap", "fully generated", "full-body swap", "real" ], "gt_answer": "full-body swap", "pred_answer": "face swap", "correct": false, "raw": "```json\n{\"answer\": \"face swap\"}\n```" }, { "id": "t2_FVRA__s2v__77f__main__id02359_hoPN5dCzSC8_00093__id06987", "clips": [ "FVRA__s2v__77f__main__id02359_hoPN5dCzSC8_00093__id06987.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "fully generated", "full-body swap", "real", "face swap" ], "gt_answer": "fully generated", "pred_answer": "real", "correct": false, "raw": "{\"answer\": \"real\"}" }, { "id": "t2_FVRA__animate__77f__main__id03681_Szvjxba6BTg_00066__id01664", "clips": [ "FVRA__animate__77f__main__id03681_Szvjxba6BTg_00066__id01664.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "real", "full-body swap", "face swap", "fully generated" ], "gt_answer": "full-body swap", "pred_answer": "real", "correct": false, "raw": "```json\n{\"answer\": \"real\"}\n```" } ] }, "t3": { "scores": { "n": 10, "accuracy": 0.5, "ci95": [ 0.2366, 0.7634 ], "balanced_accuracy": 0.3542, "cohen_kappa": 0.2647 }, "items": [ { "id": "t3_voice_A_RVRA__real__77f__vocoded__id00948_PCy1KCIsNxo_00024__id04278", "clips": [ "RVRA__real__77f__vocoded__id00948_PCy1KCIsNxo_00024__id04278.mp4", "RVRA__real__77f__main__id00948_z-P1KSBdNU8_00128__id08581.mp4" ], "part": "voice", "question": "Clip 2 is a real recording of one person. Listen only to the voices. Is the voice in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose voice it is, still counts as real.", "options": [ "fake of a different person", "fake of this person", "real, same person", "real, different person" ], "gt_answer": "real, same person", "pred_answer": "real, same person", "correct": true, "raw": "```json\n{\"answer\": \"real, same person\"}\n```" }, { "id": "t3_face_A_RVRA__real__77f__main__id05877_q_ZAQf6KfTQ_00093__id05564", "clips": [ "RVRA__real__77f__main__id05877_q_ZAQf6KfTQ_00093__id05564.mp4", "RVRA__real__77f__main__id05877_6QUcq_6B-LU_00011__id07768.mp4" ], "part": "face", "question": "Clip 2 is a real recording of one person. Look only at the faces (the inner face: eyes, nose and mouth). Is the face in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose face it is, still counts as real.", "options": [ "real, different person", "real, same person", "fake of this person", "fake of a different person" ], "gt_answer": "real, same person", "pred_answer": "real, same person", "correct": true, "raw": "```json\n{\"answer\": \"real, same person\"}\n```" }, { "id": "t3_voice_B_RVRA__real__153f__vocoded__id00885_xiPMwAYISG0_00135__id05623", "clips": [ "RVRA__real__153f__vocoded__id00885_xiPMwAYISG0_00135__id05623.mp4", "RVRA__real__153f__main__id05623_UXUHOpRirtg_00079__id03123.mp4" ], "part": "voice", "question": "Clip 2 is a real recording of one person. Listen only to the voices. Is the voice in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose voice it is, still counts as real.", "options": [ "real, same person", "fake of this person", "real, different person", "fake of a different person" ], "gt_answer": "real, different person", "pred_answer": "real, same person", "correct": false, "raw": "```json\n{\"answer\": \"real, same person\"}\n```" }, { "id": "t3_face_B_RVRA__real__153f__main__id06713_5qTLKWiq73k_00074__id05042", "clips": [ "RVRA__real__153f__main__id06713_5qTLKWiq73k_00074__id05042.mp4", "RVRA__real__153f__main__id05042_K1HMS1K0kWI_00010__id00144.mp4" ], "part": "face", "question": "Clip 2 is a real recording of one person. Look only at the faces (the inner face: eyes, nose and mouth). Is the face in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose face it is, still counts as real.", "options": [ "real, same person", "fake of a different person", "fake of this person", "real, different person" ], "gt_answer": "real, different person", "pred_answer": "real, different person", "correct": true, "raw": "```json\n{\"answer\": \"real, different person\"}\n```" }, { "id": "t3_voice_A_RVFA__seed_vc__153f__main__id04049_JCDeI7UwWnE_00080__id08096", "clips": [ "RVFA__seed_vc__153f__main__id04049_JCDeI7UwWnE_00080__id08096.mp4", "RVRA__real__153f__main__id04049_Flu_4soH0mo_00003__id04528.mp4" ], "part": "voice", "question": "Clip 2 is a real recording of one person. Listen only to the voices. Is the voice in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose voice it is, still counts as real.", "options": [ "fake of a different person", "real, same person", "real, different person", "fake of this person" ], "gt_answer": "fake of a different person", "pred_answer": "real, different person", "correct": false, "raw": "```json\n{\"answer\": \"real, different person\"}\n```" }, { "id": "t3_voice_B_RVFA__seed_vc__77f__main__id07574_DbcyMh1l_Ng_00043__id06429", "clips": [ "RVFA__seed_vc__77f__main__id07574_DbcyMh1l_Ng_00043__id06429.mp4", "RVRA__real__77f__main__id06429_0aTJ9yNMAx0_00002__id06427.mp4" ], "part": "voice", "question": "Clip 2 is a real recording of one person. Listen only to the voices. Is the voice in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose voice it is, still counts as real.", "options": [ "fake of this person", "real, different person", "real, same person", "fake of a different person" ], "gt_answer": "fake of this person", "pred_answer": "fake of a different person", "correct": false, "raw": "```json\n{\"answer\": \"fake of a different person\"}\n```" }, { "id": "t3_face_A_RVRA__real__153f__main__id00885_yMWJySOEuTI_00138__id00739", "clips": [ "RVRA__real__153f__main__id00885_yMWJySOEuTI_00138__id00739.mp4", "RVRA__real__153f__main__id00885_xiPMwAYISG0_00135__id05623.mp4" ], "part": "face", "question": "Clip 2 is a real recording of one person. Look only at the faces (the inner face: eyes, nose and mouth). Is the face in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose face it is, still counts as real.", "options": [ "real, different person", "fake of this person", "real, same person", "fake of a different person" ], "gt_answer": "real, same person", "pred_answer": "real, different person", "correct": false, "raw": "```json\n{\"answer\": \"real, different person\"}\n```" }, { "id": "t3_face_B_RVRA__real__77f__main__id08808_vQV6tYM_-3M_00053__id07697", "clips": [ "RVRA__real__77f__main__id08808_vQV6tYM_-3M_00053__id07697.mp4", "RVRA__real__77f__main__id07697_diNcpyVMCk8_00047__id07630.mp4" ], "part": "face", "question": "Clip 2 is a real recording of one person. Look only at the faces (the inner face: eyes, nose and mouth). Is the face in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose face it is, still counts as real.", "options": [ "real, same person", "fake of this person", "fake of a different person", "real, different person" ], "gt_answer": "real, different person", "pred_answer": "real, different person", "correct": true, "raw": "```json\n{\"answer\": \"real, different person\"}\n```" }, { "id": "t3_voice_B_RVFA__seed_vc__153f__main__id02359_QiCI0mefCx4_00030__id03078", "clips": [ "RVFA__seed_vc__153f__main__id02359_QiCI0mefCx4_00030__id03078.mp4", "RVRA__real__153f__main__id03078_4XN_hNJCe9E_00031__id06987.mp4" ], "part": "voice", "question": "Clip 2 is a real recording of one person. Listen only to the voices. Is the voice in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose voice it is, still counts as real.", "options": [ "real, same person", "real, different person", "fake of a different person", "fake of this person" ], "gt_answer": "fake of this person", "pred_answer": "real, different person", "correct": false, "raw": "```json\n{\"answer\": \"real, different person\"}\n```" }, { "id": "t3_voice_A_RVRA__real__153f__vocoded__id01124_7Y-51ooBIWQ_00040__id09218", "clips": [ "RVRA__real__153f__vocoded__id01124_7Y-51ooBIWQ_00040__id09218.mp4", "RVRA__real__153f__main__id01124_2hg7q9O8xLo_00015__id08748.mp4" ], "part": "voice", "question": "Clip 2 is a real recording of one person. Listen only to the voices. Is the voice in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose voice it is, still counts as real.", "options": [ "real, same person", "fake of this person", "real, different person", "fake of a different person" ], "gt_answer": "real, same person", "pred_answer": "real, same person", "correct": true, "raw": "{\"answer\": \"real, same person\"}" } ] } } }