metadata

license: mit
task_categories:
  - question-answering
language:
  - en
size_categories:
  - 10K<n<100K
pretty_name: alignment-research-dataset
dataset_info:
  - config_name: all
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        dtype: string
      - name: source_filetype
        dtype: string
      - name: converted_with
        dtype: string
      - name: book_title
        dtype: string
      - name: paged_url
        dtype: string
      - name: summary
        sequence: string
      - name: karma
        dtype: int32
      - name: votes
        dtype: int32
      - name: words
        dtype: int32
      - name: comment_count
        dtype: int32
      - name: tags
        sequence: string
      - name: modified_at
        dtype: string
      - name: source_type
        dtype: string
      - name: venue
        dtype: string
      - name: newsletter_category
        dtype: string
      - name: highlight
        dtype: int32
      - name: newsletter_number
        dtype: string
      - name: summarizer
        dtype: string
      - name: opinion
        dtype: string
      - name: prerequisites
        dtype: string
      - name: read_more
        dtype: string
      - name: alias
        dtype: string
      - name: data_last_modified
        dtype: string
      - name: abstract
        dtype: string
      - name: author_comment
        dtype: string
      - name: journal_ref
        dtype: string
      - name: doi
        dtype: string
      - name: primary_category
        dtype: string
      - name: categories
        sequence: string
      - name: bibliography_bib
        sequence:
          - name: title
            dtype: string
      - name: docx_name
        dtype: string
      - name: chapter_names
        sequence: string
      - name: file_name
        dtype: string
      - name: description
        dtype: string
      - name: filename
        dtype: string
    splits:
      - name: train
        num_bytes: 318682720
        num_examples: 9700
    download_size: 365472655
    dataset_size: 318682720
  - config_name: agentmodels
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_filetype
        dtype: string
      - name: converted_with
        dtype: string
      - name: book_title
        dtype: string
    splits:
      - name: train
        num_bytes: 386274
        num_examples: 21
    download_size: 1203135
    dataset_size: 386274
  - config_name: aiimpacts
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: paged_url
        dtype: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 3219417
        num_examples: 253
    download_size: 3290483
    dataset_size: 3219417
  - config_name: aipulse.org
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: paged_url
        dtype: string
    splits:
      - name: train
        num_bytes: 652332
        num_examples: 23
    download_size: 655946
    dataset_size: 652332
  - config_name: aisafety.camp
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: paged_url
        dtype: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 96728
        num_examples: 9
    download_size: 99056
    dataset_size: 96728
  - config_name: aisafety.info
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_filetype
        dtype: string
    splits:
      - name: train
        num_bytes: 798694
        num_examples: 294
    download_size: 830259
    dataset_size: 798694
  - config_name: alignmentforum
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: karma
        dtype: int32
      - name: votes
        dtype: int32
      - name: words
        dtype: int32
      - name: comment_count
        dtype: int32
      - name: tags
        sequence: string
      - name: modified_at
        dtype: string
      - name: source_type
        dtype: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 73852390
        num_examples: 2643
    download_size: 75161883
    dataset_size: 73852390
  - config_name: alignment_newsletter
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: converted_with
        dtype: string
      - name: source_type
        dtype: string
      - name: venue
        dtype: string
      - name: newsletter_category
        dtype: string
      - name: highlight
        dtype: int32
      - name: newsletter_number
        dtype: string
      - name: summarizer
        dtype: string
      - name: opinion
        dtype: string
      - name: prerequisites
        dtype: string
      - name: read_more
        dtype: string
    splits:
      - name: train
        num_bytes: 3202660
        num_examples: 1504
    download_size: 3553667
    dataset_size: 3202660
  - config_name: arbital
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_filetype
        dtype: string
      - name: alias
        dtype: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 3295285
        num_examples: 728
    download_size: 3624231
    dataset_size: 3295285
  - config_name: arxiv
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_type
        dtype: string
      - name: converted_with
        dtype: string
      - name: data_last_modified
        dtype: string
      - name: abstract
        dtype: string
      - name: author_comment
        dtype: string
      - name: journal_ref
        dtype: string
      - name: doi
        dtype: string
      - name: primary_category
        dtype: string
      - name: categories
        sequence: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 140132334
        num_examples: 1144
    download_size: 155841593
    dataset_size: 140132334
  - config_name: audio_transcripts
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_filetype
        dtype: string
      - name: converted_with
        dtype: string
    splits:
      - name: train
        num_bytes: 1312932
        num_examples: 37
    download_size: 1334992
    dataset_size: 1312932
  - config_name: carado.moe
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_type
        dtype: string
    splits:
      - name: train
        num_bytes: 1046524
        num_examples: 248
    download_size: 1077923
    dataset_size: 1046524
  - config_name: cold_takes
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_type
        dtype: string
    splits:
      - name: train
        num_bytes: 2122519
        num_examples: 102
    download_size: 3304055
    dataset_size: 2122519
  - config_name: deepmind_blog
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_type
        dtype: string
    splits:
      - name: train
        num_bytes: 131197
        num_examples: 10
    download_size: 132884
    dataset_size: 131197
  - config_name: distill
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_type
        dtype: string
      - name: converted_with
        dtype: string
      - name: abstract
        dtype: string
      - name: journal_ref
        dtype: string
      - name: doi
        dtype: string
      - name: bibliography_bib
        sequence:
          - name: title
            dtype: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 2251795
        num_examples: 51
    download_size: 2435649
    dataset_size: 2251795
  - config_name: eaforum
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: karma
        dtype: int32
      - name: votes
        dtype: int32
      - name: words
        dtype: int32
      - name: comment_count
        dtype: int32
      - name: tags
        sequence: string
      - name: modified_at
        dtype: string
      - name: source_type
        dtype: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 8427856
        num_examples: 401
    download_size: 8558545
    dataset_size: 8427856
  - config_name: gdocs
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_filetype
        dtype: string
      - name: converted_with
        dtype: string
      - name: docx_name
        dtype: string
    splits:
      - name: train
        num_bytes: 450068
        num_examples: 12
    download_size: 454095
    dataset_size: 450068
  - config_name: gdrive_ebooks
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_filetype
        dtype: string
      - name: converted_with
        dtype: string
      - name: chapter_names
        sequence: string
      - name: file_name
        dtype: string
    splits:
      - name: train
        num_bytes: 19583452
        num_examples: 20
    download_size: 19746001
    dataset_size: 19583452
  - config_name: generative.ink
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_type
        dtype: string
    splits:
      - name: train
        num_bytes: 516173
        num_examples: 19
    download_size: 525315
    dataset_size: 516173
  - config_name: gwern_blog
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: description
        dtype: string
    splits:
      - name: train
        num_bytes: 735083
        num_examples: 7
    download_size: 742259
    dataset_size: 735083
  - config_name: importai
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_type
        dtype: string
    splits:
      - name: train
        num_bytes: 243239
        num_examples: 15
    download_size: 246559
    dataset_size: 243239
  - config_name: jsteinhardt_blog
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: paged_url
        dtype: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 657649
        num_examples: 39
    download_size: 669942
    dataset_size: 657649
  - config_name: lesswrong
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: karma
        dtype: int32
      - name: votes
        dtype: int32
      - name: words
        dtype: int32
      - name: comment_count
        dtype: int32
      - name: tags
        sequence: string
      - name: modified_at
        dtype: string
      - name: source_type
        dtype: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 50402273
        num_examples: 1741
    download_size: 51420245
    dataset_size: 50402273
  - config_name: markdown.ebooks
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_type
        dtype: string
      - name: filename
        dtype: string
    splits:
      - name: train
        num_bytes: 1878436
        num_examples: 4
    download_size: 1897775
    dataset_size: 1878436
  - config_name: miri
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: paged_url
        dtype: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 6712562
        num_examples: 492
    download_size: 6856015
    dataset_size: 6712562
  - config_name: nonarxiv_papers
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_filetype
        dtype: string
      - name: abstract
        dtype: string
      - name: filename
        dtype: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 10369405
        num_examples: 200
    download_size: 10408494
    dataset_size: 10369405
  - config_name: qualiacomputing
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: paged_url
        dtype: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 5456683
        num_examples: 301
    download_size: 5536074
    dataset_size: 5456683
  - config_name: reports
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_filetype
        dtype: string
      - name: abstract
        dtype: string
      - name: filename
        dtype: string
    splits:
      - name: train
        num_bytes: 4604618
        num_examples: 55
    download_size: 4621199
    dataset_size: 4604618
  - config_name: vkrakovna_blog
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: paged_url
        dtype: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 360866
        num_examples: 46
    download_size: 369604
    dataset_size: 360866
  - config_name: waitbutwhy
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        dtype: string
      - name: source_type
        dtype: string
    splits:
      - name: train
        num_bytes: 157217
        num_examples: 2
    download_size: 158533
    dataset_size: 157217
  - config_name: yudkowsky_blog
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: paged_url
        dtype: string
      - name: summary
        sequence: string
    splits:
      - name: train
        num_bytes: 594207
        num_examples: 23
    download_size: 604224
    dataset_size: 594207
  - config_name: ml_safety_newsletter
    features:
      - name: id
        dtype: string
      - name: source
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: large_string
      - name: url
        dtype: string
      - name: date_published
        dtype: string
      - name: authors
        sequence: string
      - name: source_type
        dtype: string
    splits:
      - name: train
        num_bytes: 110266
        num_examples: 9
    download_size: 112020
    dataset_size: 110266

AI Alignment Research Dataset

The AI Alignment Research Dataset is a collection of documents related to AI Alignment and Safety from various books, research papers, and alignment related blog posts. This is a work in progress. Components are still undergoing a cleaning process to be updated more regularly.

Sources

The following list of sources may change and items may be renamed:

agentmodels
aiimpacts
aisafety.camp
aisafety.info
alignmentforum
alignment_newsletter
arbital
arxiv - alignment research papers from arxiv
audio_transcripts - transcripts from interviews with various researchers and other audio recordings
carado.moe
cold_takes
deepmind_blog
distill
eaforum - selected posts
gdocs
gdrive_ebooks - books include Superintelligence, Human Compatible, Life 3.0, The Precipice, and others
generative.ink
gwern_blog
importai
jsteinhardt_blog
lesswrong - selected posts
markdown.ebooks
miri - MIRI
ml_safety_newsletter
nonarxiv_papers - other alignment research papers
qualiacomputing
reports
vkrakovna_blog
waitbutwhy
yudkowsky_blog

Keys

Not all of the entries contain the same keys, but they all have the following:

id - unique identifier
source - based on the data source listed in the previous section
title - title of document
text - full text of document content
url - some values may be 'n/a', still being updated
date_published - some 'n/a'
authors - list of author names, may be empty
summary - list of human written summaries from various newsletters, may be empty

The values of the keys are still being cleaned up for consistency. Additional keys are available depending on the source document.

Usage

Execute the following code to download and parse the files:

from datasets import load_dataset
data = load_dataset('StampyAI/alignment-research-dataset')

To only get the data for a specific source, pass it in as the second argument, e.g.:

from datasets import load_dataset
data = load_dataset('StampyAI/alignment-research-dataset', 'lesswrong')

Limitations and Bias

LessWrong posts have overweighted content on doom and existential risk, so please beware in training or finetuning generative language models on the dataset.

Contributing

The scraper to generate this dataset is open-sourced on GitHub and currently maintained by volunteers at StampyAI / AI Safety Info. Learn more or join us on Discord.

Rebuilding info

This README contains info about the number of rows and their features which should be rebuilt each time datasets get changed. To do so, run:

datasets-cli test ./alignment-research-dataset --save_info --all_configs

Citing the Dataset

For more information, here is the paper and LessWrong post. Please use the following citation when using the dataset:

Kirchner, J. H., Smith, L., Thibodeau, J., McDonnell, K., and Reynolds, L. "Understanding AI alignment research: A Systematic Analysis." arXiv preprint arXiv:2022.4338861 (2022).