|
Next revision
|
Previous revision
|
ai_timelines:predictions_of_human-level_ai_timelines [2022/09/21 07:37]
127.0.0.1 external edit |
ai_timelines:predictions_of_human-level_ai_timelines [2024/01/29 21:31] (current)
harlanstewart |
| ====== Predictions of Human-Level AI Timelines ====== | ====== Predictions of Human-Level AI Timelines ====== |
| |
| // Published 05 June, 2015; last updated 24 November, 2020 // | // Published 05 June, 2015; last updated 8 May, 2023 // |
| |
| | In the most recent surveys, experts gave median predictions in the 2050s for the arrival of human-level AI. Prediction markets/aggregators currently give median predictions in the 2030s. Individual analyses give a wide range of predictions. |
| |
| | ===== Details ===== |
| |
| <HTML> | |
| <p>We know of around 1,300 public predictions of when human-level AI will arrive, of varying levels of quality. These include predictions from individual statements and larger surveys. Median predictions tend to be between 2030 and 2055 for predictions made since 2000, across different subgroups of predictors.</p> | |
| </HTML> | |
| |
| | ==== Surveys ==== |
| |
| | //(See main page: [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:ai_timeline_surveys|AI timeline surveys]])// |
| |
| ===== Details ===== | We know of 20 surveys on the predicted timing of human-level AI. Participants appear to mostly be experts in AI or related areas, but with a large contingent of others. Most of the surveys asked about AI timelines by asking participants for the year in which they expect a given probability of human-level AI (HLAI). Collapsing a few slightly different definitions of HLAI, the results of surveys that used this fixed-probability methodology are shown in figure 1. Note that asking about timelines in terms of fixed-probabilities tends to elicit shorter timeline predictions than asking in terms of fixed-years. |
| |
| | [{{:ai_timelines:year_when_x_chance_of_human-level_ai_hlai_median_answers_from_surveys_over_the_years.png|Figure 1: Median predicted year of given probabilities of HLAI from [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:agi-09_survey|AGI-09 (2009)]], [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:fhi_winter_intelligence_survey|FHI Winter Intelligence (2011)]], [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:kruel_ai_interviews|Kruel interviews (2011-2012)]], [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:muller_and_bostrom_ai_progress_poll|FHI: AGI-12 (2012)]], [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:muller_and_bostrom_ai_progress_poll|FHI:PT-AI (2012)]], [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:muller_and_bostrom_ai_progress_poll|FHI: TOP100 (2013)]], [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:muller_and_bostrom_ai_progress_poll|FHI:EETN (2013)]], [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:2016_expert_survey_on_progress_in_ai|ESPAI 2016]], [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:walsh_2017_survey|Walsh (2017)]], [[https://arxiv.org/ftp/arxiv/papers/1901/1901.08579.pdf|Gruetzemacher (2018)]], [[https://arxiv.org/abs/2206.04132|GovAI (2019)]], [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:2022_expert_survey_on_progress_in_ai|ESPAI 2022]]}}] |
| |
| ==== The landscape of AI predictions ==== | ====Prediction markets and aggregators==== |
| |
| | //(See main page: [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_prediction_markets|AI Timeline Prediction Markets]])// |
| |
| <HTML> | Prediction markets, and prediction aggregators like Metaculus, incentivize forecasters to make good predictions. One limitation of prediction markets is that they cannot incentivize correct predictions where the predicted outcome prevents a payout. Because many believe that human-level AI might lead to human extinction, prediction markets may not work as well for AI timelines as for other questions. |
| <p>Predictions of when <a href="/doku.php?id=clarifying_concepts:human-level_ai">human-level AI</a> will be achieved exist in the form of <a href="/doku.php?id=ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:ai_timeline_surveys">surveys</a> and public statements (e.g. in articles, books or interviews). Some statements backed by analysis are discussed <a href="/doku.php?id=ai_timelines:list_of_analyses_of_time_to_human-level_ai">here</a>. Many more statements have been <a href="/doku.php?id=ai_timelines:miri_ai_predictions_dataset">collected</a> by <a href="http://intelligence.org/">MIRI</a>. Figure 1 illustrates almost all of the predictions we know about, though most are aggregated there into survey medians. Altogether, we know of around 1,300 public predictions of when human-level AI will arrive, though 888 are from a single informal online poll. We know of ten surveys that address this question directly (plus <a href="/doku.php?id=ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:kruel_ai_interviews">a set of interviews</a> which we sometimes treat as a survey but here count here as individual statements, and <a href="/doku.php?id=ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:hanson_ai_expert_survey">a survey</a> which asks about progress so far as a fraction of what is required for human-level AI). Only 65 predictions that we know of are not part of surveys.</p> | |
| </HTML> | |
| |
| | * [[https://www.metaculus.com/home/|Metaculus]] is a reputation-based prediction aggregation engine that uses a scoring system to incentivize accurate predictions and to track participants' performance over time. |
| |
| ==== Summary of findings ==== | <html> |
| | <iframe src="https://www.metaculus.com/questions/embed/5121/" height="320" width="550"></iframe> |
| | </html> |
| |
| | * [[https://manifold.markets/|Manifold Markets]] is a play money prediction market with user-created questions. The play money can be redeemed as a charitable donation. |
| |
| <HTML> | <html> |
| <figure aria-describedby="caption-attachment-552" class="wp-caption alignnone" id="attachment_552" style="width: 600px"> | <iframe src="https://manifold.markets/embed/HarlanStewart/in-what-year-will-humanlevel-ai-fir" title="In what year will human-level AI first exist?" frameborder="0" width="600" height="300"></iframe> |
| <a href="http://aiimpacts.org/wp-content/uploads/2014/12/AI-predictions-summary-3.jpg"><img alt="" class="wp-image-552" height="432" sizes="(max-width: 600px) 100vw, 600px" src="https://aiimpacts.org/wp-content/uploads/2014/12/AI-predictions-summary-3-1024x737.jpg" srcset="https://aiimpacts.org/wp-content/uploads/2014/12/AI-predictions-summary-3-1024x737.jpg 1024w, https://aiimpacts.org/wp-content/uploads/2014/12/AI-predictions-summary-3-300x216.jpg 300w, https://aiimpacts.org/wp-content/uploads/2014/12/AI-predictions-summary-3.jpg 1324w" width="600"/></a> | </html> |
| <figcaption class="wp-caption-text" id="caption-attachment-552"> | |
| <strong>Figure 1:</strong> Predictions from the <a href="/doku.php?id=ai_timelines:miri_ai_predictions_dataset" rel="noopener noreferrer" target="_blank">MIRI dataset</a> (red = <a href="/doku.php?id=ai_timelines:miri_ai_predictions_dataset" rel="noopener noreferrer" target="_blank">maxIY</a> ≈ ‘AI more likely than not after …’, and green = <a href="/doku.php?id=ai_timelines:miri_ai_predictions_dataset" rel="noopener noreferrer" target="_blank">minPY</a> ≈ ‘AI less likely than not before …’) and <a href="/doku.php?id=ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:ai_timeline_surveys" rel="noopener noreferrer" target="_blank">surveys</a>. This figure excludes one prediction of 3012 made in 2012, and the <a href="/doku.php?id=ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:hanson_ai_expert_survey" rel="noopener noreferrer" target="_blank">Hanson survey</a>, which doesn’t ask directly about prediction dates. | |
| </figcaption> | |
| </figure> | |
| </HTML> | |
| |
| | ==== Analyses ==== |
| |
| <HTML> | //(See main page: [[ai_timelines:list_of_analyses_of_time_to_human-level_ai|List of Analyses of Time to Human-Level AI]])// |
| <p>Recent <a href="/doku.php?id=ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:ai_timeline_surveys">surveys</a> tend to have median dates between 2040 and 2050. All six of the surveys which ask for the year in which human-level AI will have arrived with 50% probability produce medians in this range (not including <a href="/doku.php?id=ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:kruel_ai_interviews">Kruel’s interviews</a>, which have a median of 2035, and are counted in the statements here). The median prediction in <a href="/doku.php?id=ai_timelines:miri_ai_predictions_dataset" rel="noopener noreferrer" target="_blank">statements</a> is 2042, though predictions of AGI researchers and futurists have medians in the early 2030s. <a href="/doku.php?id=ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:ai_timeline_surveys" rel="noopener noreferrer" target="_blank">Surveys</a> give median estimates for a 10% chance of human-level AI in the 2020s. We have not attempted to adjust these figures for <a href="/doku.php?id=ai_timelines:predictions_of_human-level_ai_timelines:interpretation_of_ai_predictions:accuracy_of_ai_predictions:accuracy_of_ai_predictions" rel="noopener noreferrer" target="_blank">biases</a>.</p> | |
| </HTML> | |
| |
| | ==== 2015 database of public predictions about AI ==== |
| | //(Note: this section is about data that was collected in 2015 and does not include predictions made since then)// |
| |
| ==== Implications ==== | In 2012, [[https://intelligence.org/|MIRI]] [[https://wiki.aiimpacts.org/doku.php?id=ai_timelines:miri_ai_predictions_dataset|collected]] public predictions about human-level AI timelines. In 2015, we expanded on this list to make a database of all such predictions that we knew about at the time, including public statements and survey results. This database contained 1,300 public predictions of when human-level AI will arrive, though 888 are from a single informal online poll. The database includes ten surveys that addressed this question directly (plus a [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:kruel_ai_interviews|set of interviews]] which we sometimes treat as a survey but here count here as individual statements, and [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:hanson_ai_expert_survey|a survey]] which asks about progress so far as a fraction of what is required for human-level AI). Only 65 predictions in the database are not part of surveys. |
| |
| |
| <HTML> | [{{:ai_timelines:ai-predictions-summary-3.jpg|Figure 2: Predictions from the [[http://aiimpacts.org/miri-ai-predictions-dataset/|MIRI dataset]] (red = [[http://aiimpacts.org/miri-ai-predictions-dataset/|maxIY]] ≈ ‘AI more likely than not after …’, and green = [[http://aiimpacts.org/miri-ai-predictions-dataset/|minPY]] ≈ ‘AI less likely than not before …’) and [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:ai_timeline_surveys|surveys]]. This figure excludes one prediction of 3012 made in 2012, and the [[ai_timelines:predictions_of_human-level_ai_timelines:ai_timeline_surveys:hanson_ai_expert_survey|Hanson survey]], which doesn’t ask directly about prediction dates.}}] |
| <p>Expert predictions about AI timelines are often considered uninformative. Evidence that predictions are less informative than in other messy fields <a href="/doku.php?id=ai_timelines:predictions_of_human-level_ai_timelines:interpretation_of_ai_predictions:accuracy_of_ai_predictions:accuracy_of_ai_predictions">appears</a> to be weak. We have not evaluated baseline prediction accuracy in such fields however. We expect survey results and predictions from those further from AGI are more accurate than other sources, due to <a href="/doku.php?id=ai_timelines:predictions_of_human-level_ai_timelines:interpretation_of_ai_predictions:accuracy_of_ai_predictions:accuracy_of_ai_predictions">selection biases</a>. The differences between these sources appear to be a small number of decades.</p> | |
| </HTML> | |
| |
| | The surveys in the 2015 database tend to have median dates between 2040 and 2050. All six of the surveys which asked for the year in which human-level AI will have arrived with 50% probability produced medians in this range (not including Kruel’s interviews, which had a median of 2035, and are counted in the statements here). The median prediction in statements was 2042, though predictions of AGI researchers and futurists had medians in the early 2030s. We did not attempt to adjust these figures for biases. |
| |
| | ==== Implications ==== |
| |
| | Expert predictions about AI timelines are often considered uninformative. Evidence that predictions are less informative than in other messy fields [[ai_timelines:predictions_of_human-level_ai_timelines:interpretation_of_ai_predictions:accuracy_of_ai_predictions:accuracy_of_ai_predictions|appears]] to be weak. We have not evaluated baseline prediction accuracy in such fields however. We expect survey results and predictions from those further from AGI are more accurate than other sources, due to [[ai_timelines:predictions_of_human-level_ai_timelines:interpretation_of_ai_predictions:accuracy_of_ai_predictions:accuracy_of_ai_predictions|selection biases]]. The differences between these sources appear to be a small number of decades. |
