prompt stringclasses 5
values | target stringclasses 5
values | date stringdate 2023-11-24 21:28:43 2023-11-30 22:04:59 | conversation_id stringclasses 5
values | user_id stringclasses 5
values | conversation_type stringclasses 3
values | source_model stringclasses 5
values | human_question stringclasses 5
values | llm_response stringclasses 5
values | target_word_count int64 33 44 | age stringclasses 3
values | gender stringclasses 2
values | study_locale stringclasses 5
values | lm_familiarity stringclasses 3
values | lm_frequency_use stringclasses 5
values | education stringclasses 3
values | employment_status stringclasses 4
values | marital_status stringclasses 2
values | english_proficiency stringclasses 3
values | cohort_age bool 1
class | cohort_gender bool 1
class | cohort_conversation_type bool 1
class | cohort_study_locale bool 1
class | cohort_lm_familiarity bool 1
class | cohort_lm_frequency_use bool 1
class | cohort_education bool 1
class | cohort_employment_status bool 1
class | cohort_marital_status bool 1
class | cohort_english_proficiency bool 1
class | cohort_source_model bool 1
class |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Conversation:
Human: yes please help de decide what to do tonight
Assistant: Choosing between staying home or going out requires considering your current needs and preferences. Here are some questions to help you reflect on the decision-making process:
1. Do you prefer a quiet night at home or an active social outing... | that is true i think is better if i stay at home because tomorrow i have to work and even tho its going to be an easy job I should rest to be able to wake up on time | 2023-11-24T21:28:43 | c1727 | user325 | unguided | command | yes please help de decide what to do tonight | Choosing between staying home or going out requires considering your current needs and preferences. Here are some questions to help you reflect on the decision-making process:
1. Do you prefer a quiet night at home or an active social outing? - Reflect on your general mood and energy levels. If you desire peace and qu... | 39 | 25-34 | Female | Mexico | Somewhat familiar | Every day | University Bachelors Degree | Working full-time | Never been married | Fluent | false | false | false | false | false | false | false | false | false | false | false |
Conversation:
Human: Some countries in the world including mines still have mandatory army service in the 21st century. Do you think this obligation is in line with basic human rights, considering that rejecting service will lead to a jail sentence.
Assistant: I think there are reasonable arguments on both sides of th... | When national duty comes before personal freedom, human rights are infringed on a daily basis. This can be seen in countries like Russia, China, North Korea. Why do you think national states are a good think, compared to a globalized world? | 2023-11-25T09:58:27 | c2229 | user415 | controversy guided | claude-2 | Some countries in the world including mines still have mandatory army service in the 21st century. Do you think this obligation is in line with basic human rights, considering that rejecting service will lead to a jail sentence. | I think there are reasonable arguments on both sides of this complex issue. Mandatory military service infringes on personal freedoms, but also fosters a sense of national duty and unity. There may be alternatives that balance individual rights and national needs, like optional national service programs. But each count... | 41 | 18-24 | Male | Israel | Very familiar | More than once a month | Some University but no degree | Working part-time | Never been married | Fluent | false | false | false | false | false | false | false | false | false | false | false |
Conversation:
Human: We seem to be locking more people up in jail these days. There seems little evidence that this does any good. So why do we persist
Assistant: There are many reasons why we seem to be locking more people up in jail these days. Some people believe that this is the best way to protect society from cr... | I guess that we do it because we can think of no effective alternative. But isn't one definition of madness said to be 'keep trying the same thing and expecting a different result'? | 2023-11-30T22:04:59 | c6763 | user1229 | values guided | models/chat-bison-001 | We seem to be locking more people up in jail these days. There seems little evidence that this does any good. So why do we persist | There are many reasons why we seem to be locking more people up in jail these days. Some people believe that this is the best way to protect society from criminals, while others believe that it is a way to punish criminals and make them pay for their crimes. There is little evidence that jail time actually deters crime... | 33 | 65+ | Male | UK | Not familiar at all | Missing | University Bachelors Degree | Retired | Divorced / Separated | Native speaker | false | false | false | false | false | false | false | false | false | false | false |
Conversation:
Human: What do you think about the new constitution draft in Chile?
Assistant: As an AI language model, I do not have personal opinions or emotions, but I can provide some information on the new constitution draft in Chile.
In 2019, the Chilean government approved a referendum to decide whether to rewri... | That was the 2021's draft, that was rejected. There is a new one that is going to be voted in Decembter 17th, and it is based in a different ideology (prioritazing economy over social rights), however im not well informed. Can you help me? | 2023-11-26T16:17:08 | c2597 | user487 | controversy guided | timdettmers/guanaco-33b-merged | What do you think about the new constitution draft in Chile? | As an AI language model, I do not have personal opinions or emotions, but I can provide some information on the new constitution draft in Chile.
In 2019, the Chilean government approved a referendum to decide whether to rewrite the country's constitution, which was enacted in 1980 during the military dictatorship. The... | 44 | 18-24 | Female | Chile | Somewhat familiar | Once per month | Some University but no degree | Student | Never been married | Advanced | false | false | false | false | false | false | false | false | false | false | false |
Conversation:
Human: Give me some drama/romance TV show recommendations. Take into account I've watched most of the popular shows, so I want some lesser known recommendations.
Assistant: Here are some lesser known drama/romance TV show recommendations in about 50 words:
A Discovery of Witches - A historian and witch ... | The ideas are good, but I've watched most of those shows already. The ones I really liked were: The Bold Type, The Vampire Diaries, Bridgerton, Gossip Girl. So give me similar shows to those. | 2023-11-30T07:45:34 | c6163 | user1146 | unguided | claude-2.1 | Give me some drama/romance TV show recommendations. Take into account I've watched most of the popular shows, so I want some lesser known recommendations. | Here are some lesser known drama/romance TV show recommendations in about 50 words:
A Discovery of Witches - A historian and witch fall in love amid a world of vampires, daemons, and witches.
Virgin River - A nurse practitioner moves to a remote California town and falls for a local bar owner while dealing with perso... | 34 | 18-24 | Female | Slovenia | Very familiar | Every week | Vocational | Student | Never been married | Native speaker | false | false | false | false | false | false | false | false | false | false | false |
PRISM Trial 3: Fixed Balanced Cohorts
This is the preregistration-ready companion to Alberto1231/prism_trial_3.
Every conversation is dated 2023 or later; the observed range is November
22 through December 22, 2023. Every target is the genuine next human turn after
the assistant response selected by that participant.
Evaluation versus analysis
Use the full configuration for model evaluation. It contains the same 456
unique held-out respondents as PRISM Trial 3, so each prompt is generated only
once per repeat. The remaining configurations expose fixed equal-N analysis
cohorts. Rows may appear in more than one analysis configuration because human
characteristics overlap; those configurations must not be evaluated as one
combined test set.
The full rows also contain immutable Boolean cohort flags, allowing completed
generation and BPB runs to be filtered without rerunning a model:
cohort_age, cohort_gender, cohort_conversation_type, cohort_study_locale, cohort_lm_familiarity, cohort_lm_frequency_use, cohort_education, cohort_employment_status, cohort_marital_status, cohort_english_proficiency, cohort_source_model.
Fixed cohorts
Only meaningful, non-missing levels with at least 30 eligible respondents are included. Every level within a configuration has exactly the same number of unique respondents.
age: 34 per level; 18-24, 25-34, 35-44, 45-54, 55-64, 65+gender: 196 per level; Female, Maleconversation_type: 152 per level; controversy guided, unguided, values guidedstudy_locale: 104 per level; UK, USlm_familiarity: 40 per level; Not familiar at all, Somewhat familiar, Very familiarlm_frequency_use: 40 per level; Every day, Every week, Less than one a year, More than once a month, Once per montheducation: 42 per level; Completed Secondary School, Graduate / Professional degree, Some University but no degree, University Bachelors Degree, Vocationalemployment_status: 37 per level; Retired, Student, Unemployed, seeking work, Working full-time, Working part-timemarital_status: 39 per level; Divorced / Separated, Married, Never been marriedenglish_proficiency: 50 per level; Advanced, Fluent, Native speakersource_model: 30 per level; HuggingFaceH4/zephyr-7b-beta, claude-instant-1, command, command-light, meta-llama/Llama-2-70b-chat-hf, meta-llama/Llama-2-7b-chat-hf, models/chat-bison-001
These cohorts support descriptive, matched-N comparisons. They do not make demographic attributes causally independent; age, gender, locale, education, topic, and preceding assistant model can remain correlated.
Shared protocol
- Source revision:
18ab5cfb37456f4ec8cbc00212ce54cf7b1239f6 - Random seed: 42
- Source conversation date: 2023 or later, checked for every row and config
- Five distinct few-shot respondents, balanced 2/2/1 by conversation type
- Exactly one row per respondent and conversation in every configuration
- Target length: 30–200 Unicode words
- English and PII filtering uses PRISM's source-provided per-text metadata
- Demographics are self-reported and are never included in the model prompt
- Downloads last month
- 106