Botmaster Jeremy Lichtman’s Sudan conflict deaths forecast, Sept. 15, 2025

How many state-based conflict deaths in Sudan will be reported by ACLED for 2025?

Here’s what Jeremy Lichtman’s Multi-AI Oracle predicts today:

Less than 1,000: 1%
Between 1000 and 3,000: 5%
Between 3000 and 5,000: 16%
Between 5000 and 8,000: 26%
Between 8000 and 12,000: 36%
More than 12,000: 16%

11–14 minutes read

Obtaining outside data.
Checking internet sources, using keywords: state-based conflict deaths Sudan 2025
News from various sources:

In 2025, Sudan is experiencing one of the worst education crises globally due to an ongoing conflict between the Sudanese army and the paramilitary Rapid Support Forces (RSF), which began in April 2023. According to Save the Children, more than three-quarters of Sudan’s school-age children are not attending classes, equating to around 13 million out of 17 million children. The conflict has not only led to the closure of schools but also has resulted in tens of thousands of deaths, millions being displaced, and significant damage to the country’s infrastructure.

A new study by Save the Children highlighted that approximately seven million children who are enrolled cannot access education due to the conflict or displacement, while six million are not enrolled at all. If the ongoing violence persists, these children face risks, such as recruitment into armed groups and exposure to sexual violence. The United Nations also reported incidents of girls being forced into marriage under threats to their families.

However, there is some progress, as almost half of the country’s schools have reopened, allowing around four million children to return to school. Relative calm in parts of central Sudan, including Khartoum, has enabled about two million people to return home, mainly settling in the Al-Jazira state and Khartoum. Despite this, Sudan remains in a dire humanitarian situation, with millions displaced internally and internationally, and famine conditions in refugee camps in Darfur.

Fact-checking news summary:
Based on the summary provided, here is a list of specific facts, along with an analysis regarding their correctness, importance, and relevance to the original question of how many state-based conflict deaths will be reported by ACLED in Sudan in 2025:
1. **Fact:** In 2025, Sudan is experiencing one of the worst education crises globally due to an ongoing conflict.
– Correctness: True, based on the current situation in Sudan.
– Importance: Important for understanding the wider context of the conflict.
– Relevance: False, as it does not directly contribute to estimating conflict deaths.
2. **Fact:** The conflict is between the Sudanese army and the Rapid Support Forces (RSF), which began in April 2023.
– Correctness: True, according to various reports on the conflict.
– Importance: Critical, since knowing the parties involved might help analyze potential casualties.
– Relevance: True, as the ongoing conflict directly relates to the estimation of conflict deaths.
3. **Fact:** More than three-quarters of Sudan’s school-age children are not attending classes, equating to around 13 million out of 17 million children.
– Correctness: True, according to the description given.
– Importance: Lesser, provides context around the severity of the humanitarian situation.
– Relevance: False, as it does not contribute directly to estimating conflict deaths.
4. **Fact:** The conflict has resulted in tens of thousands of deaths, millions being displaced, and damage to infrastructure.
– Correctness: Presumably true, though specific numbers are not confirmed in the summary.
– Importance: Critical, as it directly relates to the question.
– Relevance: True, because previous death toll data is relevant for future death projections.
5. **Fact:** Approximately seven million children who are enrolled cannot access education due to the conflict or displacement.
– Correctness: True, as per the summary.
– Importance: Lesser, as it highlights the impact of the conflict on education.
– Relevance: False, since it’s not directly related to conflict death estimation.
6. **Fact:** Six million children are not enrolled at all.
– Correctness: True, as mentioned in the summary.
– Importance: Lesser, as it depicts the scale of the educational disruption.
– Relevance: False, doesn’t contribute to conflict death estimation.
7. **Fact:** If the conflict persists, children face recruitment risks into armed groups and exposure to sexual violence.
– Correctness: True, generally a consequence of such conflicts.
– Importance: Important, as it might indirectly affect mortality rates.
– Relevance: True, provides insights into potential future scenarios which could influence death tolls.
8. **Fact:** The United Nations reported incidents of girls being forced into marriage.
– Correctness: Presumably true, consistent with reports of conflict areas.
– Importance: Important, highlights human rights violations in conflict which might affect death toll indirectly.
– Relevance: True, provides part of the larger picture regarding human rights abuses, which can influence mortality.
9. **Fact:** Almost half of the country’s schools have reopened, allowing around four million children to return to school.
– Correctness: True, according to the report summary.
– Importance: Lesser, suggests some normalization but does not detail death counts.
– Relevance: False, does not directly affect the death count prediction.
10. **Fact:** Around two million people returned home to central Sudan due to relative calm.
– Correctness: True, based on the summary.
– Importance: Important, since internal displacement returns can affect conflict intensity and data availability.
– Relevance: True, changes in displacement patterns might influence conflicts and death estimates.
11. **Fact:** Famine conditions exist in refugee camps in Darfur.
– Correctness: True, consistent with known humanitarian issues in the region.
– Importance: Important, since famine can exacerbate mortality indirectly connected to conflict.
– Relevance: True, details on humanitarian crises can intersect with conflict-related deaths.

Each fact contributes differently to the understanding of the context in Sudan, but only some are directly relevant to the estimation of conflict-related deaths.

Fact-checked summary:
In 2025, the conflict in Sudan involves the Sudanese army and the Rapid Support Forces (RSF), and it began in April 2023. This ongoing conflict has resulted in tens of thousands of deaths, millions being displaced, and significant damage to infrastructure, which is directly relevant for estimating future conflict deaths. If the conflict persists, children face recruitment risks into armed groups and exposure to sexual violence, scenarios that could influence the death toll. Furthermore, the United Nations has reported incidents of girls being forced into marriage, highlighting human rights violations that can indirectly affect mortality rates. Changes in displacement patterns, such as the return of around two million people to central Sudan, might influence conflicts and death estimates. Additionally, famine conditions in refugee camps in Darfur are exacerbating the humanitarian situation, which is pertinent as such crises can intersect with and impact conflict-related deaths. These facts form an essential basis for understanding the environment in which conflict deaths might be projected for Sudan in the coming years.

OpenAI gives a base rate of between 8000 and 12000
The question difficulty is rated 8 (0 to 10)
Historical weighted factors include:
0.30.20.10.250.15
A Bayesian calculation could be performed as follows:
To calculate the Bayesian likelihood, consider the probabilities of necessary events multiplied by their historical weightings. Given the ongoing conflict and recent escalations, adjust the weights by these conditions. The current situation trends towards a high death toll range, considering the persistence and scale of the conflict as indicated in recent news: (0.3*0.8) + (0.2*0.5) + (0.1*0.9) + (0.25*0.7) + (0.15*0.6) = 0.67.
(Carolyn‘s note: GenAIs are notoriously poor at arthmetic. My spreadsheet says 0.3*0.8)+(0.2*0.5)+(0.1*0.9)+(0.25*0.7)+(0.15*0.6)=0.695.)
Bayesian base rate: 0.67 (Carolyn’s note: In cases where the Multi-AI Oracle is only forecasting a single percent probability, this Bayesian rate is easy to understand. But in this case, with five probability buckets, this apparently is an overall confidence level across all buckets. On the other hand, GenAIs aren’t good at math)
Sufficient news to provide a good forecast? 1 (0 or 1)
News is relevant, topical and unbiased? 1 (0 or 1)
Question classification: scenario_based_forecast
Expected time (days) until question close: 200
The following were considered in order to produce this base rate:
The base rate is derived from historical data on conflict mortality rates in similar regional conflicts with comparable parameters such as duration, intensity, and international involvement. The current ongoing nature and reported severity of the conflict push the base rate towards higher mortality outcomes.

Ideally, the news feed would contain the following sorts of information for a better forecast:
Detailed data on the current military capabilities of both parties, accurate death toll reporting, and reliable updates on humanitarian aid effectiveness could improve the forecast.

Some potential divergent considerations that might affect the base rate:
Resolution might differ if unforeseen peace agreements occur, significant international intervention takes place, or if natural events alter the course of the conflict (e.g., severe climate changes affecting conflict zones).

The following chain of events are necessary for the question to resolve positively:
– Continuation of the conflict between the Sudanese army and the RSF, with estimated heavy combat and military engagements. (Likely) – Failure of peace negotiations or cease-fires between conflicting parties. (Likely) – Continued lack of international intervention or successful mediation efforts. (Likely) – Persistent humanitarian crises worsening due to famine, disease, or lack of resources. (Very Likely) – Ongoing human rights violations, including recruitment of children and forced marriages. (Likely)

Querying Claude (AI predicts: [“Less than 1000”: 0.02, “Between 1000 and 3000”: 0.08, “Between 3000 and 5000”: 0.15, “Between 5000 and 8000”: 0.25, “Between 8000 and 12000”: 0.35, “More than 12000”: 0.15] – confidence: 6)

Querying Mistral (AI predicts: [“Less than 1000”: 0.01, “Between 1000 and 3000”: 0.05, “Between 3000 and 5000”: 0.15, “Between 5000 and 8000”: 0.3, “Between 8000 and 12000”: 0.4, “More than 12000”: 0.09] – confidence: 5)

Querying OpenAI (AI predicts: [“Less than 1000”: 0.01, “Between 1000 and 3000”: 0.05, “Between 3000 and 5000”: 0.1, “Between 5000 and 8000”: 0.15, “Between 8000 and 12000”: 0.29, “More than 12000”: 0.4] – confidence: 6)

Question Type: Multiple Choice
Confidence: 6
MAPD: Avg: 0.073333333333333, Max: 0.20666666666667
# LLM responses: 3
Explanations of the above statistical measures here —>
Model value:
Less than 1,000: 1%
Between 1000 and 3,000: 5%
Between 3000 and 5,000: 16%
Between 5000 and 8,000: 26%
Between 8000 and 12,000: 36%
More than 12,000: 16%

The various AI analyses predict a high mortality rate from the ongoing conflict between the Sudanese army and the RSF in Sudan since April 2023, compounded by humanitarian crises, failed peace negotiations, and lack of effective international intervention. Comparisons with historical data from similar conflicts suggest a base estimate of 8,000-12,000 deaths annually. These predictions could be altered if unexpected peace agreements occur, international intervention increases, or if significant shifts occur within the conflict dynamics, such as moving to lower-intensity operations or an unforeseen decisive military outcome. The reliability of data and potential natural disasters also introduce variability into mortality projections.

Runtime: 150 seconds.

Past forecasts by Phil’s and Jeremy ’s bots —>

Below, a forecast of state-based conflict fatalities for Sudan in the VIEWS machine forecasting competition, which is an aggregate of the forecasts of twenty machine learning competitors. Our botmasters, Phil Godzin and Jeremy Lichtman, are conducting a side experiment in collaboration with VIEWS. The VIEWS aggregate forecast is much lower than ours.