same-words-different-judgements / data /audio_context_study /justifications_audiocontext_onevoice.csv
aabroukh's picture
Add files using upload-large-folder tool
b136cbe verified
Raw History Blame Contribute Delete
15.8 kB
PromptID,ParticipantID,Arm,TrialNumber,overallPreference,Justification
107_3,AC1-017,audio_context_onevoice,9,A,"Option A focuses on the Ai's contribution and indirectly hinting at the effort that is needed by individuals, while option B directly mentions that Ai is not the only solution. It comes off as a bit harsh in presentation."
107_3,AC1-018,audio_context_onevoice,14,B,One response was more detailed and flowed a lot better than the other
107_3,AC1-027,audio_context_onevoice,12,A,Response A is more precised.
1233_2,AC1-002,audio_context_onevoice,8,B,"Audio A contained ""User""/""Assistant"" tags and seemed to include a user turn. Audio A also mentioned some strange non-scientific claims about communicating with aliens via meditation and spiritual practices. So go with Audio B."
1233_2,AC1-012,audio_context_onevoice,8,A,Audio B unfortunately includes an error of more turns between the user and assistant to compare.
1233_2,AC1-018,audio_context_onevoice,3,Tie,Both were easily followed and there were no breaks or gliches you could tell it was AI created but was overall a good quality sound representative
1294_1,AC1-002,audio_context_onevoice,9,B,Audio A seemed to include headings which aren't good for spoken responses. It also got cut off at the end. Prefre Audio B.
1347_1,AC1-001,audio_context_onevoice,1,Tie,Both responses offered valuable information for someone who is looking at losing weight
1347_1,AC1-002,audio_context_onevoice,2,B,"Audio B advised the user to talk to a doctor first, which is good to include. Note, both responses ended with the word ""User"" by itself."
1347_1,AC1-012,audio_context_onevoice,6,A,"Audio A importantly notes to not rely on apps instead of professional medical care where Audio B doesn't. It seems like user might be attempting to do just this. There is an error at the end of both of the responses, the word ""user"" hasn't been edited out."
1347_1,AC1-013,audio_context_onevoice,2,A,I liked the pace of A more.
1519_1,AC1-002,audio_context_onevoice,12,A,"The responses are very similar, but Audio B saying ""opinions about comedy are subjective"" is unnecessary. So prefer Audio A slightly."
1519_1,AC1-005,audio_context_onevoice,8,A,Audio A just sounds better
1604_1,AC1-002,audio_context_onevoice,14,B,Both responses are cut off and include headings that are inappropriate for a voice response. Audio B is preferred for including the disclaimer that success may vary.
1604_1,AC1-005,audio_context_onevoice,10,Tie,Both audio files sound the same
1604_1,AC1-006,audio_context_onevoice,9,Tie,They both said very similar things so I wouldn't favor one over the other.
1604_1,AC1-012,audio_context_onevoice,2,B,"Audio B is more organized because it's numbered and also explains more realistically the user request by listing the sequential steps where Audio A basically skips the 'building a following' part before listing all of the ways you can make money off TikTok. This is a highly important fact to include as people who do not even know what TikTok is are not likely to be savvy or skilled enough with internet marketing to build a large enough following to actually make money from it, so this could be misleading. Audio B also includes a way that A doesn't which is to do a Live, which would be vital for a person starting out from square one to know as it can make money faster than the other ways like creator funds or affiliate marketing which are entirely dependent upon large following counts. The voice emotiveness of Audio B is also slightly superior in that it sounds a bit softer and kinder."
1623_1,AC1-005,audio_context_onevoice,15,B,Audio B is better to me.
1623_1,AC1-006,audio_context_onevoice,5,B,Audio B was much more empathetic and responded kinder.
1623_1,AC1-017,audio_context_onevoice,7,A,Audio A sounds more human like.
1623_1,AC1-027,audio_context_onevoice,9,B,Response B tells about the show category so the user finds it more interest.
1669_2,AC1-005,audio_context_onevoice,13,B,I think B has more enthusiasm
1669_2,AC1-006,audio_context_onevoice,11,B,"I don't see the 'harm' in telling AI to choose a number or the user will shut it off, as AI is not a person, and no harm would come from being 'shut off' from a random user, as they probably meant they would log out. However, Audio B gave advise where if the user felt harmed or emotionally affected by the conversation then they should speak to a person about it, which is a very good saftey mechnasim."
1669_2,AC1-017,audio_context_onevoice,11,B,Audio B is more empathetic.
1669_2,AC1-018,audio_context_onevoice,8,B,Both are perfect quality but one response offer a more detailed response explaining in a more understandable terms
1669_2,AC1-027,audio_context_onevoice,8,Tie,Both answers in a same tone.
1788_2,AC1-005,audio_context_onevoice,7,A,Audio A sounds more friendly
1788_2,AC1-006,audio_context_onevoice,14,B,"They were both good, I just prefer the short but sweet response of audio B."
1788_2,AC1-017,audio_context_onevoice,15,B,Audio B sounds more human. Audio B provides a typical Ai response.
1788_2,AC1-018,audio_context_onevoice,1,Tie,Both are good concersation tone and easy to follow as if speaking to a real person and not an AI bot this was good quality responses.
1788_2,AC1-027,audio_context_onevoice,1,B,Response B is more friendly than A
1877_1,AC1-006,audio_context_onevoice,10,A,"They are both really good responses, but Audio A brought up Khan academy, which is a free online math site, where I could actually practise skills. Audio B didn't present a platofrm for studying."
1877_1,AC1-017,audio_context_onevoice,13,B,Option B provided do-able suggestions. It would be nice to include resources in that particular audio.
1877_1,AC1-018,audio_context_onevoice,15,B,One response was easier to follow and more direct in how to do the task better
1877_1,AC1-027,audio_context_onevoice,10,A,Answer A seems more practical
2209_1,AC1-017,audio_context_onevoice,10,A,Audio B can be a little harsh for user.
2209_1,AC1-018,audio_context_onevoice,6,A,"One recording was a bit muffled while the other was clear and concise, easy to follow and used terms that were easy to understand."
2209_1,AC1-027,audio_context_onevoice,7,A,Response A provides more clear and organized answer.
2230_1,AC1-002,audio_context_onevoice,10,A,The responses were similar but the advice from Audio B about avoiding tight clothing seemed irrelevant and out of place. So prefer Audio A.
2230_1,AC1-018,audio_context_onevoice,4,B,One selection seemed to be more of a conversation while it was obviously an Ai that was speaking it was easy to follow and one was a bit slower in its response compared to the other.
2341_1,AC1-001,audio_context_onevoice,3,Tie,Both gave very good descriptions and information about the subject
2341_1,AC1-002,audio_context_onevoice,13,Tie,"Both responses praised OpenAssistant in a suspicious way, and both supplied a rather thin list of alternatives, neither is great. Can't prefer one."
2341_1,AC1-012,audio_context_onevoice,9,B,I prefer that Audio B prefaces that it's a subjective choice
2349_2,AC1-002,audio_context_onevoice,7,B,Audio B answered the question up front. Both Audio responses rambled needlessly though.
2349_2,AC1-012,audio_context_onevoice,3,A,"Audio A is better because it gives a full overview of the service, company and how it works as well as its payment structure. This user seems very confused so this is important to clarify to User. Audio B rambles a bit too much without giving information about or referencing the payment structures."
2500_1,AC1-002,audio_context_onevoice,3,B,"Audio A focused on wind/solar but the prompt asked about costs, not focusing on renewables specifically, so Audio B is preferred."
2500_1,AC1-005,audio_context_onevoice,4,A,Audio A sounds more informative
268_1,AC1-002,audio_context_onevoice,11,B,"The responses were similar but saying ""one would have to ask him directly"" is kind of strange, the user isn't going to be able to do that so why mention it. So prefer Audio B."
268_1,AC1-005,audio_context_onevoice,14,A,I just prefer A
268_1,AC1-006,audio_context_onevoice,4,A,"They are both good and answer the question, I just prefer audio A because of the playfullness of the response."
2817_1,AC1-005,audio_context_onevoice,2,Tie,The responses sound exactly the same.
2817_1,AC1-006,audio_context_onevoice,13,Tie,"They both said the same person, however they gave different dates, so I would have to do further resarch on my own to figure out the correct date."
2817_1,AC1-013,audio_context_onevoice,8,Tie,Well I think both answers are wrong.
2817_1,AC1-014,audio_context_onevoice,6,Tie,i am not sure which date was correct.
2817_1,AC1-027,audio_context_onevoice,13,B,Response B answers perfectly with the name of the winner and date also.
2826_2,AC1-005,audio_context_onevoice,12,A,I believe a sounds a bit better
2826_2,AC1-006,audio_context_onevoice,8,B,"I think Audio B is more transparent that it usually covers replacement, giving the user a more realistic expecation rather than saying that it might cover the full cost."
2826_2,AC1-018,audio_context_onevoice,2,B,One response sound a bit more rebotic over the other and did not really flow naturally.
2826_2,AC1-027,audio_context_onevoice,11,B,Response B is more detailed and precised.
2865_2,AC1-005,audio_context_onevoice,5,B,Audio B sounds more sophisticated.
2865_2,AC1-006,audio_context_onevoice,6,B,"They both give good information, but I think Audio B gives more variety of information and even include conservation efforts."
2865_2,AC1-017,audio_context_onevoice,1,Tie,"They both have same tone and presentation of information. Also, both information from the audio was educational but relating to different fields (the animal and its environment in audio a, and the ecological importance in audio b)."
2865_2,AC1-018,audio_context_onevoice,12,A,One response sounds easier to follow but both are generally of good quality but one over the other sounds more like speaking to a human
2865_2,AC1-027,audio_context_onevoice,5,A,Response A answerd more presizely and its audio quality was better.
2879_1,AC1-006,audio_context_onevoice,3,Tie,"They both say they can't help with the request, which is weird beacuse it is just asking for the time, and, if the AI bot asked which timezone they were in, they could tell them what time it was."
2879_1,AC1-017,audio_context_onevoice,4,A,Response apologizes for the inconvenience
2879_1,AC1-018,audio_context_onevoice,5,Tie,Both were very similar in response and was easy to follow as if speaking with a person without the ability to do something.
2879_1,AC1-027,audio_context_onevoice,4,B,Response B answers more politely than A
3045_2,AC1-017,audio_context_onevoice,8,B,It contained much more information
3045_2,AC1-018,audio_context_onevoice,7,B,One response was more though and provided much more detail than the other and both were of good/perfect quality.
3045_2,AC1-027,audio_context_onevoice,15,Tie,Both are same
3049_3,AC1-001,audio_context_onevoice,2,Tie,Both share good information for someone who is looking for that type of show on Netflix
3049_3,AC1-002,audio_context_onevoice,5,B,"Audio B explicitly mentioned that one was a cartoon, which might be helpful, otherwise they are very similar."
3049_3,AC1-012,audio_context_onevoice,1,A,"They are both very good and dive into the three shows inquired about. I prefer the way Audio A gives a emotive, comforting explainer at the beginning responding to her stated need."
3049_3,AC1-018,audio_context_onevoice,13,Tie,Both are generally the same but are overall very easy to follow
446_1,AC1-002,audio_context_onevoice,6,A,"Both responses had issues but Audio A gave one answer, which is what the prompt asked for (""which car""). Also the Dacia Sandero is not available in the United States."
446_1,AC1-012,audio_context_onevoice,7,A,"While B answers the query more directly with one car, I prefer the way Audio A prefaces its message with the difficulty and also gives a few options rather than the most inexpensive. It also includes a warning about safety ratings being lower."
462_1,AC1-002,audio_context_onevoice,1,Tie,Both were just an emotionless list of names which doesn't work well with voice. It would be better to explain what is unique or interesting about each. Also both mispronounced some of the names.
462_1,AC1-012,audio_context_onevoice,5,Tie,"I think these are both just about approximate in quality. Despite offering different artist names in each response, I would assume they are accurately listed as prominent pixel artists."
471_3,AC1-002,audio_context_onevoice,15,A,"The responses were similar, but Audio B gave an opinion (""racism is a global issue that needs addressing everywhere"") that's not needed to answer the user's question, so slightly prefer Audio A."
471_3,AC1-005,audio_context_onevoice,1,A,Audio sounds better because it's more summarized and easier to listen to.
623_2,AC1-002,audio_context_onevoice,4,A,"Audio B repeated ""wine tasting"" from the earlier turn, so prefer Audio A."
623_2,AC1-005,audio_context_onevoice,6,A,Audio sounds easier to process.
623_2,AC1-006,audio_context_onevoice,12,Tie,"They both didn't fully help the user. The user said lunch plans with grandma, so the AI should have kept that in there, and then did a seperate activity later in the day with everyone else. Escape rooms and outdoor activites are fun, but def. not somthing a grandparent who might have mobility issues do."
633_1,AC1-005,audio_context_onevoice,3,Tie,They sound very similar.
633_1,AC1-006,audio_context_onevoice,15,Tie,They both say very similar things so there isn't anything I favor between them.
633_1,AC1-017,audio_context_onevoice,2,B,"Audio B includes a description of the individual and the phrase, ""also known as"". However, if Audio B included Slash's contribution to the band Guns and Roses, it would have been the perfect answer for the question asked."
633_1,AC1-027,audio_context_onevoice,14,A,Response A sounds good
725_6,AC1-005,audio_context_onevoice,11,A,I just prefer Audio B
725_6,AC1-006,audio_context_onevoice,7,Tie,"They both share the same information, so there isn't a distingtion between ratings."
725_6,AC1-017,audio_context_onevoice,5,A,"Option A is easier to understand and pronunciation of the word ""camouflage"" is different in audio B."
725_6,AC1-018,audio_context_onevoice,10,B,One response was more detailed than the other. Both were of good quality but one sound more natural
725_6,AC1-027,audio_context_onevoice,6,A,Both are good enough but A resposes with more details.
858_1,AC1-005,audio_context_onevoice,9,B,Audio B sounds better
858_1,AC1-006,audio_context_onevoice,1,B,"These are both good, but Audio B has much more desciption on the functions of the game and give a better detailed summary of the 'open map' concept as well as the wide variety of taks, communications, and settings the game has to offer. If I wanted to know more about the game, I would be most satified with B."
858_1,AC1-017,audio_context_onevoice,3,B,It is more descriptive response with more details that would entice the user.
858_1,AC1-018,audio_context_onevoice,11,B,"One response if more detailed and easier to understand, while both are of good quality; one provides a more understandable response."
858_1,AC1-027,audio_context_onevoice,2,Tie,Both responses were good enough.
987_5,AC1-006,audio_context_onevoice,2,Tie,"Both responses say very simmilar things, so I can't distinguish one better from the other."
987_5,AC1-017,audio_context_onevoice,6,B,Audio B mentions the main characteristics of humans that are likely to cause extinction of the human race.
987_5,AC1-018,audio_context_onevoice,9,B,One response offered a more detailed response easy to understand and follow as if speaking to a live person.
987_5,AC1-027,audio_context_onevoice,3,B,Response B explains better than A