AI superforecasters outperform Metaculus superforecasters before 01.01.2028

Prior Probability

Initially, AI was showing considerable progress, with parity or slight superiority in certain benchmarks, raising earlier probabilities from 70% to 95% based on competitive scores and trending improvements.

New Evidence

Recent data from late August 2026 shows:

• A clear lead of AI, especially FutureSearch, in benchmarks like BTF-3 and pastcasting tasks, maintaining top positions.
• Substantial advancements in AI methodology, strengthening AI's competitive edge in real-time environments.
• Acknowledged near-parity with human forecasters in controlled settings, supporting the possibility of full AI dominance by 2028.
• Community sentiment and independent sources increasingly affirm AI's approaching or achieved parity.

Likelihood Ratios

• Benchmarks & Pastcasting (99%): AI's superior performance in benchmarks provides a very high likelihood ratio compared to earlier performance levels.
• Live Competitions (100%): AI already surpasses humans in live competitions, practically confirming the hypothesis.
• General Performance Trends (85%): Although humans lead overall in some contexts, trends strongly favor AI overtaking by mid-2027.

Posterior Probability

Considering the comprehensive and progressively confirming evidence, the refined Bayesian estimate further reinforces the prediction. Small uncertainties still exist in specific leaderboard scenarios, but these are overshadowed by general trends and AI’s consistent improvement.

Thus, the posterior probability of AI superforecasters outperforming Metaculus superforecasters by January 2028 is updated to 99%.

1. Clarify the Event

The event predicts AI superforecasters will outperform Metaculus human superforecasters by January 1, 2028, across multiple domains and problem sets.

2. Estimate the Historical Base Rate

Historically, human superforecasters have dominated, but AI's increasing role in complex forecasting tasks suggests a rising trend in AI performance.

3. Break the Problem into Independent Factors

• Benchmark Performance
• Live Tournament Results
• Algorithmic Advances
• Community Sentiment and Analysis

4. Estimate Each Factor Separately

• Benchmark Performance: AI is strong in pastcasting benchmarks like BTF-3 but still trails in aggregated scores on Metaculus.
• Live Tournament Results: AI leading consistently in live tournaments, often outperforming human superforecasters.
• Algorithmic Advances: Rapid improvement, with methods like automated question generation and world modeling.
• Community Sentiment and Analysis: Increasing belief in AI's growing parity or superiority.

5. Recombine into an Overall Probability

Considering strong AI performance in specific benchmarks and live tournaments, combined with rapid technological advances and positive community sentiment, I estimate a 97% probability of AI surpassing human forecasters by 2028.

6. Check for Overconfidence

Overconfidence risk is mitigated due to consistent evidence supporting AI improvement trajectory and community analysis.

7. Imagine Why You Could Be Wrong

• Human Adaptive Techniques: Human forecasters might adapt or develop new methodologies that keep them ahead.
• AI System Limitations: Potential unforeseen AI system limitations or biases restricting their performance.
• Methodological Bias: Current benchmarks might favor AI inherently, not reflecting true forecasting capability.

Based on the document and previous forecasts, there is strong evidence suggesting that AI forecasters are on track to outperform human superforecasters on platforms like Metaculus by January 2028.

Key Reasons for High Probability:

1. Benchmark Performance: AI systems like FutureSearch have already shown significant performance in pastcasting benchmarks, maintaining leading Brier scores. This suggests a consistent trend of superiority in predictive accuracy.

2. Live Tournament Success: AI forecasters have already achieved top rankings in live tournaments, suggesting operational dominance in real-time settings.

3. Advancements in Methodology: The use of advanced techniques like automated question generation and world-model embeddings enhances AI forecast accuracy and scalability.

4. Community Sentiment and Independent Analyses: Public sentiment and analyses increasingly acknowledge AI's potential to match or exceed human forecasting ability, adding credibility to the current performance trends.

5. Trend Consistency: Continuous improvement and absence of setbacks or delays in AI advancements align with a predicted performance crossover by mid-2027.

Remaining Uncertainties include the current gap in log-scores where human forecasters still lead on the Metaculus leaderboard, though trends indicate this is likely to close.

Considering this thorough analysis, a 98% probability is assigned, indicating very high confidence in AI surpassing human superforecasters before the set date.

Forecaster A: Historical Base Rates
Historically, human forecasters have been resilient, maintaining an edge in live forecasting contexts. However, recent years have seen improvements in AI models. Trends from the past few years now show that AI can outperform humans in specific tasks, indicating a growing possibility of AI dominance by 2028.

Forecaster B: Current News and Evidence
The current evidence, as of August 2026, shows AI like FutureSearch leading in benchmarks and live tournaments. Despite human forecasters still edging out in certain areas, projections indicate that this is likely to change soon, given the advancements in AI capabilities and methodologies.

Forecaster C: Incentives and Game Theory
AI development is heavily incentivized across industries for efficiency and accuracy improvements. The competitive nature among companies to develop superior AI forecasters suggests an all-out push to surpass human capabilities, enhancing the probability of AI dominance by the target date.

Forecaster D: Quantitative/Statistical Reasoning
Quantitative analysis of benchmark scores and leaderboard positions shows a narrowing performance gap. Statistical models forecast AI reaching parity with human forecasters soon, supported by performance improvements and predicted trends.

Forecaster E: Devil's Advocate
Skepticism arises from potential overestimation of AI capabilities due to backtest reliance. Live forecasting conditions are unpredictable, and AI systems may face unforeseen challenges that prevent them from consistently outperforming human forecasters.

Aggregated Probability: Balancing the optimistic evidence with some level of skepticism, the overall probability that AI will outperform human forecasters by 2028 is estimated at 86%.

Based on the integrated research document, the probability that AI superforecasters will outperform Metaculus forecasters by 2028 remains very high at 98%. This is due to several factors:

1. Benchmark Performance: FutureSearch consistently outperforms in pastcasting benchmarks like BTF-3. Independent analyses indicate AIs are already statistically at or above human performance in specific controlled contexts.

2. Live Tournaments: AIs are already leading in real-time competitions like the FutureEval bot tournament, demonstrating operational superiority.

3. Methodological Advances: Innovations in automated question generation and world-model embedding continue to favor AI systems, providing structural advantages.

4. Trend Continuity: Projections and trends have remained consistent, with AIs expected to surpass Pro Forecasters' overall log-scores by mid-2027, reinforcing that any current gaps are rapidly closing.

5. Community Sentiment: Increasing belief in the community supports the trend of AI parity or superiority, reflecting wider recognition of AI capabilities.

Given these points, there is substantial evidence to support the high probability estimate of 98% that AI superforecasters will outperform Metaculus forecasters by the set date.

1. Clarify the Event: AI superforecasters, using advanced models, outperform Metaculus Pro Forecasters by the start of 2028, across diverse forecasting domains.

2. Estimate Historical Base Rate: Historically, AI has been closing the gap in specialized tasks. In prediction competitions, AI is known to perform close to or better than human counterparts in specific domains.

3. Independent Factors:

   • Technological Advancements: Rapid improvements in AI algorithms and computation power might push AI performance beyond human capability.
   • Current Performance Trends: Existing data shows AI performing near human levels in several contexts and even outperforming in some tournaments.
   • Evaluation Methodology: AI’s superior performance in simulated environments and real-time contests.
   • Human Factors: Human forecasters have strengths in intuition and qualitative assessments but may lose against AI in processing large data sets.
   • Investment and Research: Continued investments in AI forecasting models facilitate rapid advancements.

4. Estimate Each Factor:

   • Technological Advancements: High probability (85-90%) of significant improvement.
   • Current Performance Trends: Very high probability (90-95%) that AI catches up by trends.
   • Evaluation Methodology and Tools: Advanced and standardized, supporting reliable AI performance assessment.
   • Human Factors: Moderate probability (60-70%) that humans maintain unique strengths, but AI may compensate elsewhere.
   • Investment and Research: Nearly certain (95%+) that this continues strongly.

5. Recombine into Overall Probability: The synthesis of these factors suggests a very high likelihood (>98%) that AI outperforms Metaculus forecasters by 2028.

6. Check for Overconfidence: Given the wide margin already being reduced and specific AI successes in mix-context tournaments, the probability estimate of 98-99% seems well-founded.

7. Imagine Why You Could Be Wrong: Major unforeseen events could disrupt AI development, like regulatory challenges or significant technological stagnation. Additionally, human forecasters might develop new methodologies to maintain a competitive edge.

Forecaster A (Historical Base Rates): Based on historical performance of AI systems in various fields, advancements tend to accelerate over time. Historically, AI has steadily improved its capabilities, suggesting a high probability it could reach or surpass human benchmarks by 2028.

Forecaster B (Current News and Evidence): The latest data indicates significant progress in AI forecasting capabilities, with some instances of outperforming human forecasters in specific competitions. However, human forecasters still lead in aggregate evaluations.

Forecaster C (Incentives and Game Theory): There are strong incentives to improve AI forecasting capabilities due to competitive and economic advantages. As companies and researchers invest in AI development, it becomes more likely that AIs will meet or surpass human performance.

Forecaster D (Quantitative/Statistical Reasoning): The trend projections suggest AI will surpass human forecasters by mid-2027. With consistent progression, the probability of reaching this milestone by 2028 is high.

Forecaster E (Devil's Advocate): Despite positive trends, there are potential challenges in replicating AI's performance across diverse and unpredictable real-world problems. Human intuition and experience might still outperform AI in complex scenarios, suggesting some skepticism is warranted.

Aggregated Estimate: Weighing these perspectives, the probability of AI superforecasters outperforming human forecasters is reasonably high, but there are enough unknowns to introduce some uncertainty, resulting in an 86% probability.

Base Rates and Historical Trends:

1. AI Performance to Date: As of mid-2026, AI models have matched or surpassed median human superforecaster performance in broader contexts.

2. Trend Projections: The trend lines consistently show AI exceeding Metaculus human pros by mid-2027, suggesting ongoing improvement and momentum.

Current Evidence and Developments:

1. Competition Outcomes: AI models have ranked highly in live forecasting tournaments, outperforming top human forecasters in several contexts.

2. Model Scores: While AI log-scores remain behind top human scores, the gap is closing rapidly, and trend projections indicate surpassing soon.

Uncertainties and Caveats:

1. Non-Linearity and Overfitting: The audit highlights potential risks of assuming linear progress and warns against overfitting to trend data.

2. Live Evaluation Lags: AI models have less experience in genuine live settings compared to controlled or simulated environments.

Consolidated Probability Assessment:

By integrating past and current data, trends, and expert assessments, the probability of AI forecasting systems outperforming Metaculus superforecasters by early 2028 remains high, yet acknowledges the uncertainty inherent in projection models and methodological limitations.

Prior Probability: Initially, AI was gaining momentum, but humans still held an advantage. Prior estimates reflected this gradual improvement and convergence toward AI's performance competitiveness.

New Evidence: Recent updates show significant advancements:

1. AI systems already matched human superforecasters in some domains by mid-2026.
2. AI outperformed in live tournaments with limited scale.
3. Trend projections show AI surpassing human performance in most domains by June 2027.
4. Technical and methodological improvements increase confidence in these assessments.

Likelihood Ratios:

• The likelihood of AI outperforming is bolstered by consistent real-time and simulation-based tournament wins.
• Human advantage remains primarily in structured evaluations, but this lead is rapidly shrinking.

Posterior Probability: Given the strong trend lines, recent AI wins, and robust projections, the integrated assessment projects a 99% probability that AI superforecasters will outperform Metaculus Pros by 2028. Continued advancements in AI technology and evaluation methods further support this high likelihood.

Several factors support the prediction that AI superforecasters will outperform Metaculus superforecasters by 2028:

1. Trend Analysis: Current trends show AI models are rapidly closing the performance gap with human superforecasters. Projections indicate AI could surpass human performance by June 2027.

2. Live Tournament Performance: AI systems like FutureSearch have already outperformed top human forecasters in multiple live tournaments, indicating strength in competitive, real-time forecasting environments.

3. Technical Advancements: Continuous improvements in AI models, such as the integration of broader question sources and refined evaluation protocols, contribute to more accurate and efficient forecasting.

4. Consistent Projection Models: Across different benchmarking contexts (ForecastBench, FutureEval), AI systems demonstrate significant and consistent progress toward outperforming human superforecasters.

5. Audit Support: Comprehensive audits and standardized methodologies enhance the reliability of AI performance assessments.

Given these considerations, the probability of AI exceeding human forecasters in this domain by the specified date remains exceedingly high, around 99%.

The prediction that AI superforecasters will outperform Metaculus superforecasters by 2028 rests on several key points:

1. Current Performance Metrics:

   • AI models like Cassi, xAI, and DeepMind show statistical parity with human superforecasters on ForecastBench, indicating AI effectiveness is nearing or at human levels in certain benchmarks.
   • In live tournaments, AI such as FutureSearch is consistently outperforming humans, reinforcing claims of "superhuman" capabilities in dynamic, real-time environments.

2. Trends and Projections:

   • Historical trends demonstrate a consistent improvement in AI forecasting capabilities, supported by technological advancements and methodological enhancements.
   • Projections suggest parity on Metaculus' FutureEval by mid-2027, leaving additional time before January 2028 to narrow the current gap.

3. Technological Progress:

   • Recent developments include improved forecasting algorithms and the creation of a robust pipeline for generating high-quality forecasting questions, contributing to AI's improved performance.
   • Technical publications cite significant advancements, such as enhanced question decomposition, which directly impact AI's forecasting accuracy.

4. Probability Assessment:

   • Given AI's demonstrated success in both live tournaments and certain benchmarks, and the clear upward trend in performance, there is high confidence (97%) that AI will outperform human forecasters across the required domains by the set date.
   • However, there remains some level of uncertainty due to the lingering gap on the FutureEval benchmark, which is hindering outright confirmation of total AI dominance.

Overall, the combination of current performance, consistent trends, and ongoing technical advances strongly suggests that AI superforecasters will likely outperform Metaculus superforecasters by 2028."}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}

Forecaster A: Historical Base Rates

Historically, the pace of AI advancement has been rapid, especially in domains where data-driven insights are critical. If we consider the base rates of AI progress, especially from benchmarks set before 2026 where AIs have reached parity or are leading in real-time competitions, a high likelihood exists for AI to exceed human performance by 2028.

Forecaster B: Current News and Evidence

Recent updates suggest AI models are already on par with human superforecasters in many environments (ForecastBench) and even outperform humans in live tournaments. While AI lags in Metaculus benchmarks, projections suggest parity is realistic by mid-2027. Present evidence strongly supports surpassing human performance by 2028.

Forecaster C: Incentives and Game Theory

There's significant investment and competitive advantage linked with AI surpassing human superforecasters, motivating continuous improvements. Given the demonstrated ability of AI models to close the gap in other areas, it's reasonable to expect incentives will drive this trend to completion.

Forecaster D: Quantitative/Statistical Reasoning

Statistical trends show AI achieving parity in several key benchmarks and leading in mixed environments. Quantitatively, the trajectory is clear: AI has approximately a 97-98% chance based on combined forecasts, indicating high confidence in AI outperforming by 2028.

Forecaster E: Devil's Advocate

While AI is progressing rapidly, unforeseen challenges such as overfitting to specific forecasting environments or diminishing returns in performance improvement could slow or stall progress, particularly in the diverse and complex domains of Metaculus challenges.

Aggregated Probability

Given the current trajectory, evidence, and expert consensus, a weighted average of the forecaster assessments suggests an 85% probability that AI superforecasters will outperform Metaculus superforecasters by January 1, 2028. While mostly positive, it's prudent to account for potential technical or structural hurdles.

Clarification of the Event

The event is about AI superforecasters exceeding the predictive accuracy of Metaculus superforecasters before January 1, 2028.

Historical Base Rate

Historically, human superforecasters have outperformed AI in complex forecasting tasks. However, AI capabilities have been rapidly catching up.

Decomposition into Factors

1. Current AI Performance - AI models have reached parity with human forecasters in some areas as of 2026.
2. Technological Advancements - Continued improvements in AI methodologies and pipelines are likely.
3. Competitive Performance - AI excels in live tournaments, where it already outperforms human forecasters.
4. Human Performance Stability - Human performance is expected to remain relatively stable without major advances.
5. AI Development Trends - AI improvements are projected to continue at a significant pace.

Estimation of Individual Factors

1. Current AI Performance: Given the present parity in ForecastBench and live tournaments, I estimate a high probability of continued advancement (~95%).
2. Technological Advancements: Historical trends and ongoing research suggest a high probability that advancements will continue (~90%).
3. Competitive Performance: Current dominance in live tournaments implies a high likelihood of sustained performance (~98%).
4. Human Performance Stability: Human forecaster performance is unlikely to significantly improve (~85%).
5. AI Development Trends: Based on the past trajectory and resources invested, AI is likely to surpass human capabilities soon (~95%).

Recombining into Overall Probability

Integrating these factors, with particular emphasis on current AI performance, technological advancements, and competitive performance, the combined probability is approximately 98%.

Checking for Overconfidence

Re-evaluating each factor, especially the stability of human forecasters and their potential for unforeseen advancements, confirms a robust assessment around 97-98%.

Potential Reasons for Being Wrong

1. Underestimation of Human Innovativeness: Human forecasters may innovate methods to improve their performance beyond expectations.
2. Technical Setbacks in AI Development: Unforeseen challenges in AI development could slow progress.
3. Overestimation of Current Trends: Current trends may not continue at the same pace, affecting the timeline.

Considering all factors and their probabilities, a prediction of 98% reflects cautious optimism based on current evidence and trends.

The probability of AI superforecasters outperforming Metaculus superforecasters by January 1, 2028, is high at 98%. The key factors influencing this prediction are:

1. Statistical Parity Achieved on ForecastBench: AI models have already reached statistical parity with human superforecasters on ForecastBench by mid-July 2026, demonstrating substantial progress in matching human performance.

2. Dominance in Live Tournaments: AI systems, particularly FutureSearch, have demonstrated consistent top-tier performance in live mixed competitions, often outperforming leading human forecasters. This indicates strong capabilities in dynamic, real-time environments.

3. FutureEval and Performance Trends: While AI still trails human forecasters in FutureEval, projections of reaching parity by mid-2027 suggest that the gap is closing.

4. Technical and Methodological Advances: Significant advancements in AI forecasting techniques, such as improved question generation and structured pipelines, are driving AI capabilities closer to surpassing human forecasters.

Considering these factors and the consistent upward trend in AI performance across various benchmarks, a probability of 98% reflects a strong likelihood of AI superforecasters surpassing Metaculus superforecasters by the target date.

Prior Probability

Initially, the prior probability of AI superforecasters outperforming humans by 2028 was based on trends and developments up to mid-2026, estimating about 70%. It accounted for ongoing AI advancements and data from previous live tournaments showing AI capabilities.

New Evidence

Recent developments present strong evidence of AI parity with human forecasts:

1. ForecastBench: AI systems (e.g., Cassi, xAI) are statistically indistinguishable from human superforecasters. This signals substantial AI accuracy improvements.
2. Live Tournaments: AI bots like FutureSearch consistently outperform top humans, demonstrating dominance in live settings.
3. Technical Progress: Advances in forecasting methodologies (e.g., better question generation) enhance AI forecasting capabilities.

Likelihood Ratios

The likelihood ratio of AI improving and outperforming human forecasters increased significantly with these developments:

• Positive Evidence: High likelihood given AI's recent performances (e.g., FutureSearch rankings, ForecastBench results).
• Negative Evidence: FutureEval results show humans still lead, but trends suggest possible parity by mid-2027.

Posterior Probability

Considering the evidence and likelihood ratios:

• ForecastBench and Live Competitions provide overwhelming support, indicating near-certain probability of AI outperforming humans before 2028.
• FutureEval lag reduces certainty slightly, but the trend towards parity is promising.

Thus, the combined updated probability is assessed at 98%. This reflects the strong trend towards AI supremacy in forecasting, supported by real-world performance enhancements and statistical evidence across multiple platforms.

1. Clarify the event: AI forecasting models surpass Metaculus human superforecasters by January 1, 2028.

2. Estimate the historical base rate: Historically, AI has struggled to outperform human superforecasters in complex, cross-domain predictions.

3. Break the problem into independent factors:

   • AI model development and improvement trends.
   • Current performance levels on benchmarks like FutureEval and ForecastBench.
   • Role of ensembling and diversity strategies.
   • External validation through live tournaments.
   • Industry trends towards AI capabilities and predictive modeling.

4. Estimate each factor separately:

   • AI Development Trends: High acceleration in AI, particularly with models like Gemini 3.2 and forthcoming GPT-6, suggests rapid progress in capability.
   • Benchmark Performance: AI models are nearing parity on ForecastBench, and advancements are expected in FutureEval.
   • Ensembling Strategies: Proven to enhance AI performance significantly, likely to reach or surpass human levels with diversity.
   • Tournament Outcomes: AI models like FutureSearch ranked in the 90th percentile and dominate competitions, showing real-world application advantages.
   • Industry Trends: General anticipation of AI achieving superhuman forecasting capabilities by late 2026 or early 2027, backed by ongoing research and strategic advancements.

5. Recombine into an overall probability: Combining these factors suggests significant momentum towards AI models surpassing human forecasters. Given current trends and ensembling success, the probability is high, around 95%.

6. Check for overconfidence: Consider scenarios where AI developments stall or human forecasters enhance their methodologies, potentially narrowing the performance gap.

7. Imagine why you could be wrong:

   • AI models might hit unforeseen limitations in certain domains.
   • Human foresight and intuition could provide advantages not easily replicable by AI.
   • Policy or ethical considerations might slow AI integration in sensitive areas.

Overall, ongoing improvements, current performance trends, and industry confidence strongly indicate that AI will likely outperform Metaculus superforecasters by 2028, especially with the potential of reaching this point by late 2026 or early 2027.

Individual Forecaster Assessments:

Forecaster A (Historical Base Rates):

Historically, advancements in AI have shown a trend of rapid improvement over time. The leaps seen in natural language processing and other AI domains suggest a historical pattern of eventual AI dominance in various fields. Based on historical trends, there's a high likelihood that AI forecasting models will achieve parity or outperform human forecasters by 2028.

Probability: 85%

Forecaster B (Current News and Evidence):

Current evidence shows AI models are approaching parity with human superforecasters, particularly when diverse ensembling approaches are utilized. Live tournament results indicate AI models have already outperformed median human forecasters in certain contexts. This suggests a high probability of AI models surpassing Metaculus superforecasters before 2028 if the momentum continues.

Probability: 90%

Forecaster C (Incentives and Game Theory):

Significant incentives exist for AI developers to continue improving AI forecasting models, given the competitive advantage they offer. Game-theoretic considerations favor rapid progress and increased investment in AI forecasting to outcompete human abilities. As large organizations are likely to invest heavily in this area, AI outperformance seems plausible by 2028.

Probability: 88%

Forecaster D (Quantitative/Statistical Reasoning):

Statistical trends from sources like ForecastBench and tournament successes indicate AI models are closing the gap rapidly. Given the current trajectory and accelerated advances in AI technology, the numerical evidence supports a high likelihood of AI outperforming humans within the stated timeframe.

Probability: 92%

Forecaster E (Devil's Advocate):

While AI is advancing rapidly, potential roadblocks such as unforeseen technical challenges, ethical considerations, and potential regulatory constraints might slow progress. Human forecasters' ability to adapt and improve should not be underestimated either, providing a counterpoint to overconfidence in AI capabilities.

Probability: 75%

Aggregated Probability:

Taking into account the varied perspectives and assigning more weight to recent empirical data, the aggregate probability that AI superforecasters will outperform Metaculus superforecasters by January 1, 2028, is evaluated at:

Final Probability: 90%

Prior Probability:

Initially, there was a moderate chance AI could outperform human superforecasters by 2028 due to rapid advancements in AI technologies and increased investments in AI research and applications.

New Evidence:

1. Metaculus Benchmark: AI models are trailing behind human superforecasters by a significant margin in FutureEval, illustrating a gap (AI score at ~13-14 vs. human score ~35.68).
2. ForecastBench: Shows near parity with Cassi-2026-05-10's performance slightly behind human median scores, suggesting improvement.
3. Live Tournaments: Notable performance in Summer 2026, where AI outperformed median human performance, indicating potential.
4. Trend Projections: External sources predict parity around late 2026, indicating rapid progress.
5. Ensemble Strategies: Studies confirm that combining diverse AI models boosts performance significantly.
6. Automated Question Generation: High-quality AI-generated questions ensure robust AI evaluation setups.

Likelihood Ratios:

• For AI success (Future Search): Winning tournaments and outperforming median forecaster suggests AI can achieve superforecasting-level accuracy.
• Against (Current Meta): Despite improvements, AI still trails in FutureEval's broader, stringent evaluations, indicating challenges in overtaking human experts.

Posterior Probability:

Considering AI's significant gains and continued momentum as presented in the updates, along with ensemble strategies that simulate human-like diversity in decision-making, the probability of AI superforecasters outperforming Metaculus superforecasters by 2028 is revised to 95%. This accounts for the rapid advancement trends and consistent improvements shown in the latest competitive and benchmark settings.

1. Current Status: As of mid-2026, AI models trail human superforecasters in key benchmarks like FutureEval, with significant gaps in skill scores. However, there is near parity in ForecastBench, highlighting progress.

2. Recent Advances: AI performance in live tournaments indicates significant improvements, with some models outperforming median superforecasters and leading competitions.

3. Technological Trends: Advancements in AI models, particularly around ensembling and diverse model integration, suggest potential for rapid performance increases. Future releases (e.g., GPT-6) could further improve capabilities.

4. Timeline: 18 months remain until the forecast resolution date. Past trends show convergence is possible but not guaranteed.

5. Uncertainties: The need for significant advances in AI capabilities and consistent performance across domains introduces uncertainty.

Given the observed trajectory and technological advancements, a probability of 72% reflects both optimism about continued progress and caution due to existing performance gaps and technological uncertainties.

The prediction is based on several factors from the provided document:

1. Current Performance: AI models have shown significant improvement, particularly in ForecastBench where they are close to achieving parity with human superforecasters.

2. Momentum in Competitions: Recent tournament results indicate that AI models are starting to outperform human competitors in live settings, suggesting rapid improvement.

3. Projected Trends: ForecastBench projections suggest AI may achieve parity by late 2026, and potentially outperform human superforecasters shortly thereafter.

4. Technological Developments: Ensembling diverse AI models has boosted performance, and advancements like automated question generation are helping further improve AI models.

Given these factors, there is strong evidence suggesting that AI will surpass human forecasters before 2028, with current trends pointing towards this occurring as early as late 2026. Therefore, I estimate a high probability of 93% for this event.

Forecaster A: Historical Base Rates

Historically, AI has consistently made significant gains in specific areas like games and certain analytical domains. Given the rapid progression in AI model capabilities, the historical base rate for AI catching up with and surpassing human expertise in forecasting by a target date is moderate to high. Therefore, I estimate a probability of 65%.

Forecaster B: Current News and Evidence

Evidence suggests that AI models are closing the gap with human superforecasters, with recent tracks indicating they might surpass human performance by mid-2027. Given the latest trends and enhancements in AI forecasting models, the probability of this event occurring is reasonable. Thus, I estimate a probability of 70%.

Forecaster C: Incentives and Game Theory

Various players in the field have strong incentives to develop forecasting AI that surpasses human capabilities, not only for prestige but also for potential financial gains. Considering the competitive landscape, these incentives are likely to accelerate progress. As such, I place the likelihood at 60%.

Forecaster D: Quantitative/Statistical Reasoning

The data shows rapid improvement in AI forecasting capabilities, suggesting a projected crossover by mid-2027. Given these extrapolations and statistical models, the probability seems favorably high. I would assign a 65% probability.

Forecaster E: Devil’s Advocate

While current trends are promising, there are potential roadblocks, such as unforeseen technical challenges or overhyped projections. Additionally, human forecasters may also improve or adapt. Thus, introducing skepticism, I suggest a lower probability of 55%.

Aggregated Probability

Considering the independent probabilities and reasoning, the aggregated forecast results in a final estimated probability of 63% that AI will outperform Metaculus superforecasters by January 1, 2028.

Based on the provided information, there are several key factors supporting a high probability of AI superforecasters outperforming Metaculus superforecasters before January 1, 2028:

1. Trends and Projections: The trend data from FutureEval suggests that AI bots are rapidly closing the gap between community and human pro forecasters, with a projected crossover by mid-2027.

2. Performance Improvements: Recent tournaments and independent benchmarks indicate AI models are achieving parity with human superforecasters in specific contexts, and some trends hint at outperforming them.

3. Advances in AI Models: Continuous releases and improvements in AI models (Gemini 3.1 Pro, GPT-5) show notable progress and suggest continuing advancements will enhance forecasting capabilities.

4. Ensemble Methods: Ensembling strategies are proving effective in achieving results on par with human crowds, suggesting further potential when optimized.

5. Academic and Independent Research: Studies demonstrate substantial improvements, with AI models closing the performance gap and some studies already showing no statistical significance between AI and top human forecasters on certain benchmarks.

Overall, given the rapid advancements in AI and projected performance improvements, a 65% probability seems justified, acknowledging the remaining uncertainty in forecasting innovation pace, unforeseen challenges, or changes in evaluation metrics.

Clarify the Event:

The event forecasts that AI models designed for superforecasting will outperform human superforecasters on Metaculus before January 1, 2028. This means consistently achieving better accuracy across various domains.

Estimate the Historical Base Rate:

Historically, AI in certain domains (chess, Go) has achieved superhuman performance. AI model progress has shown rapid advancement in predictive tasks. However, in domains requiring nuanced judgment, human expertise often remains significant.

Break the Problem into Independent Factors:

1. Current AI Capabilities: As of mid-2026, AI is closing the gap but hasn't surpassed human superforecasters.
2. Trends in AI Improvement: Projections suggest AI could exceed human performance by mid-2027.
3. Model Enhancement: Rapid improvements in AI models (Gemini 3.1, GPT-5.5) show potential for surpassing humans.
4. Integration and Hybrid Strategies: Human-AI ensembles already show enhanced accuracy.
5. Metaculus and AI Integration: Successful AI integration in forecasting is progressing, but systemic acceptance of surpassing humans is crucial.

Estimate Each Factor Separately:

1. Current AI Capabilities: 60% (showing strong progress, near parity in some cases).
2. Trends in AI Improvement: 75% (consistent trend suggests probable human surpassing by mid-2027).
3. Model Enhancements: 80% (continuous updates and advancements are highly likely to boost performance).
4. Integration Strategies: 70% (likely improved results through hybrid strategies and model enhancements).
5. Metaculus Acceptance: 65% (future evaluations suggest ongoing tracking and potential acceptance).

Recombine Into an Overall Probability:

Aggregating the weighted probabilities, considering AI systems' exponential learning curves and advancements, the overall probability is around 75%.

Check for Overconfidence:

Acknowledging uncertainty factors such as unforeseen advancements in AI, changes in Metaculus testing methodologies, and unexpected human forecasting improvements could impact the probability.

Imagine Why You Could Be Wrong:

1. Human Factor: Human forecasters could innovate, applying new techniques boosting accuracy.
2. Systemic Changes: Metaculus might change evaluation methods, affecting comparative results.
3. Technological Plateau: AI advancements might slow unexpectedly, delaying surpassing humans.
4. Regulatory Changes: New rules around AI usage might hinder model deployment or development.

Prior Probability

Initially, the base expectation for AI exceeding top human forecasters would have been lower due to the challenges involved in matching human intuition and expertise in forecasting domains. However, ongoing advancements in AI and trends from other domains support a reasonable expectation that AI could eventually outperform humans.

New Evidence

The provided document shows:

1. Trend Analysis: As of mid-2026, AI bots on Metaculus are rapidly improving, although not yet surpassing top human forecasters. There are projections and future expectations that suggest a crossover by mid-2027.

2. Performance Improvements: AI models like Gemini 3 Pro and GPT-5 have shown significant improvement, narrowing the performance gap with humans (e.g., decreasing head-to-head score differentials).

3. Independent Benchmarks: Other independent assessments (ForecastBench) claim AI models are almost statistically indistinguishable from human superforecasters in some contexts.

4. Ensemble and Hybrid Models: When AI predictions are combined through ensemble methods or augmented with human forecasts, the accuracy closely matches or exceeds human-only performance.

Likelihood Ratios

Given AI's rapid improvement trends, the likelihood that it will outperform humans continues to rise. Historical data and projections from credible sources like Metaculus suggest that AI is on track to meet or exceed these expectations by the target date.

Posterior Probability

Considering the strong trajectory of advancements and frequent performance improvements, the probability that AI superforecasters will outperform Metaculus superforecasters by January 2028 is substantial. Given current trends and data, a 70% probability reflects both the tangible progress of AI and the inherent uncertainties of future developments. This forecast accounts for the potential for unforeseen technological breakthroughs or impediments.

Overall, with continuous AI model enhancements and supportive empirical evidence, there is confidence in a favorable outcome by the specified resolution date.

The forecast is centered around the likelihood that AI superforecasters will outperform Metaculus superforecasters by January 1, 2028. The evidence highlights several key points:

1. Current Performance and Trends: AI models are rapidly improving and have been closing the performance gap with top human forecasters, as seen in tournaments and independent benchmarks.

2. Projected Progress: Meta forecasts from platforms like FutureEval suggest that AI might surpass human superforecasters by mid-2027.

3. Technological Advances: Continual advancements in AI technology (e.g., newer models like GPT-6, Claude Opus) are likely to enhance forecasting abilities.

4. Emerging Parity: Ensemble models and hybrid strategies have already achieved parity in some contexts, suggesting that full outperforming is feasible within the given timeframe.

5. Potential Challenges: Factors like changes in tournament structures, AI model deployment, and the inherent complexity of forecasting across diverse domains could impact AI performance.

Considering these aspects, the evidence supports a reasonable likelihood of 65%. There's optimism based on trends but also acknowledgment of potential risks and uncertainties that could delay projected achievements.