Sudan updates from AI Botmaster Jeremy Lichtman of April 29, 2025

How many conflict deaths will there be in Sudan in 2025?

Jeremy’s most recent version of his Multi-AI Oracle is a side competitor with the machine forecasting bots of the VIEWS competition.

Here’s what Jeremy‘s latest version of his Multi-AI Oracle predicts: (Carolyn’s note: Both Jeremy and I agree, these probabilities are, so far, just an experiment. It is too early to know how good these will turn out to be)

Model value:
* Less than 1000: 1%
* Between 1000 and 3000: 2%
* Between 3000 and 5000: 4%
* Between 5000 and 8000: 8%
* Between 8000 and 12000: 15%
* More than 12000: 70%

Here’s how Jeremy’s Multi-AI Oracle says it arrived at these probabilities:

Obtaining outside data.
Checking internet sources, using keywords: Sudan conflict deaths 2025
Results from around the web: {think}

Okay, I need to answer the user’s query about Sudan conflict deaths in 2025 using the provided search results. Let me start by reading through each source carefully to extract relevant data. First, looking at the search results:

1. ACLED’s Conflict Watchlist 2025 for Sudan reports over 28,700 fatalities by November 2024, including 7,500 civilians. They mention that this is likely an underestimate, with some estimates as high as 150,000. Also, 30% of the population displaced. Source [1].
2. Science.org article from February 2025 discusses the difficulty in getting accurate numbers but references the same ACLED data (28,700 deaths, 7,500 civilians). It adds that more people are dying from starvation and disease than direct violence. Both SAF and RSF are using starvation as a weapon. Source [2].
3. CFR’s Global Conflict Tracker mentions a death toll estimate as high as 150,000. Source [3].
4. Genocide Watch’s January 2025 report states over 61,000 killed in Khartoum alone, with 26,000 direct war deaths.  They cite LeMonde’s estimate of 150,000 civilian deaths. Also, El Fasher’s siege led to 782 civilian deaths. Source [4].
5. NRC’s April 2025 report estimates over 150,000 deaths and 14 million displaced. Source [5].

So, the main figures are the ACLED’s 28,700 reported fatalities, which are considered an underestimate, and higher estimates up to 150,000 from other sources. There’s a consensus that the actual numbers are much higher due to underreporting and deaths from indirect causes like starvation and disease. The humanitarian crisis is severe with massive displacement.

The user also asked for tables. I need to present the death toll estimates and possibly the displacement numbers. Since the data varies, I should note the sources and the ranges. However, there’s no time series data provided in the search results beyond the November 2024 figure from ACLED and the 2025 reports. So time series might not be
feasible, but I can create a table summarizing the different estimates. For the URLs, I need to list all the sources used, which are [1] to [5].

The user wants the conclusion to include the URLs, so I’ll list them at the end.

Now, structuring the answer: start with a summary of the key points, then present the data in a table, and list the URLs. Make sure to cite each source correctly as per the format rules. Check if any sources are from the same organization or if there’s overlap. For example, ACLED’s data is cited in both [1] and [2], but [2] adds context about causes of death. The 150,000 figure appears in [3], [4], and [5], so those should be grouped.

I should also mention the conflicting parties, the methods of violence, and the humanitarian impact. Highlight that the death toll is disputed and likely underreported. The table can have columns for Source, Reported Deaths, Time Period, and Notes. Maybe another table for displacement stats if needed, but the main focus is on deaths. Make sure not to include unverified info. All data comes from the provided sources, so stick to what’s there. Use citations after each relevant sentence. Avoid using headers at the start, begin with a summary. Follow the formatting rules for tables and citations.

The conflict in Sudan has resulted in significant human casualties and displacement, with death toll estimates varying widely due to challenges in data collection. By early 2025, reported fatalities range from **28,700** to **150,000**, with indirect deaths from starvation and disease surpassing direct combat fatalities. Over **14 million people** have been displaced, representing nearly 30% of Sudan’s population[1][5].

### Key Data on Conflict Deaths (2023–2025)

                            * Includes 7,500 civilians killed in direct attacks

Civilian deaths from war-related causes:

*Includes indirect deaths from starvation and disease
#26,000 direct war deaths in Khartoum; 90% underreported

### Humanitarian Impact
– **Displacement**: Over 14 million displaced, including 11.3 million internally[1][5].
– **Starvation as a weapon**: Both SAF and RSF block food and medical aid[2][4].
– **Human rights violations**: Mass rape, torture, and destruction of infrastructure[2][4].

### Challenges in Data Accuracy
– Underreporting due to destroyed hospitals, displacement of aid workers, and communication blackouts[2][4].
– Most deaths attributed to preventable causes like disease and malnutrition rather than direct violence[2][5].

### URLs of Sources Used
1. https://acleddata.com/conflict-watchlist-2025/sudan/
2. https://www.science.org/content/article/how-many-have-died-sudan-s-civil-war-satellite-images-and-models-offer-clues
3. https://www.cfr.org/global-conflict-tracker/conflict/power-struggle-sudan
4. https://www.genocidewatch.com/single-post/sudan-genocide-emergency-january-2025
5. https://www.nrc.no/feature/2025/the-humanitarian-crisis-in-sudan-explained
OpenAI gives a base rate of Probability of death toll exceeding 12,000 as follows:
The question difficulty is rated 9 (0 to 10)
Historical weighted factors include:
ACLED’s report of 28,700 in Nov 2024, 0.3
NRC’s report of over 150,000 in April 2025, 0.4
Genocide Watch report of 61,000+ deaths in Khartoum, 0.3
A Bayesian calculation could be performed as follows:
Using historical factors for estimation: ((0.3 * 28700) + (0.4 * 150000)
+ (0.3 * 61000)) results in a weighted average suggesting a high
likelihood of exceeding 12,000 deaths.
Bayesian base rate: 0.85
Sufficient news to provide a good forecast? 1 (0 or 1)
News is relevant, topical and unbiased? 1 (0 or 1)
Question classification: reference_class
Expected time (days) until question close: 246

The following were considered in order to produce this base rate:
Base rate derived from historical reports suggesting low underreported estimates of around 30,000 fatalities alongside figures exceeding 150,000 due to indirect causes.

Ideally, the news feed would contain the following sorts of information for a better forecast:
Access to consistent on-ground reporting, satellite imagery data, and verification by multiple independent sources would improve forecast quality.

Some potential divergent considerations that might affect the base rate:
Potential changes in reporting accuracy, ceasefire agreements, or unforeseen international aid efforts could lower actual or reported death tolls.

The following chain of events are necessary for the question to resolve positively:
– Death toll estimates reach between 3000 and 5000 Very unlikely
– Death toll estimates reach between 5000 and 8000 Unlikely
– Death toll estimates reach between 8000 and 12000 About even odds
– Death toll estimates reach more than 12000 Likely

Querying Claude (AI predicts: [“Less than 1000”: 0.01, “Between 1000 and 3000″: 0.02, “Between 3000 and 5000”: 0.03, “Between 5000 and 8000”: 0.04, “Between 8000 and 12000”: 0.1, “More than 12000”: 0.8] – confidence: 8)

Querying Mistral (AI predicts: [“Less than 1000”: 0.01, “Between 1000 and 3000″: 0.05, “Between 3000 and 5000”:  0.1, “Between 5000 and 8000”: 0.2, “Between 8000 and 12000”: 0.3, “More than 12000”: 0.34] – confidence: 7)

Querying OpenAI (AI predicts: [“Less than 1000”: 0.01, “Between 1000 and 3000″: 0.02, “Between 3000 and 5000”: 0.04, “Between 5000 and 8000”: 0.08, “Between 8000 and 12000”: 0.15, “More than 12000”: 0.7] – confidence: 6)

Question Type: Multiple Choice
Confidence: 7
MAPD: Avg: 0.10222222222222, Max: 0.30666666666667
# LLM responses: 3

Model value:
* Less than 1000: 1%
* Between 1000 and 3000: 2%
* Between 3000 and 5000: 4%
* Between 5000 and 8000: 8%
* Between 8000 and 12000: 15%
* More than 12000: 70%

The various AI analyses converge on the conclusion that the death toll in Sudan is expected to remain significantly high, with estimates exceeding 12,000 due to the ongoing conflict. Reports from credible sources like ACLED suggest fatalities have surpassed 28,700 as of late 2024, and estimates from organizations like NRC put the death toll as high as 150,000 by early 2025, owing largely to indirect causes such as starvation and disease. The humanitarian crisis is severe, marked by massive displacement and the strategic use of starvation by conflicting parties. Despite these grim predictions, potential uncertainties include the possibility of peace agreements, improved international aid, and reporting inaccuracies that could alter current estimates. These factors, if realized, might lead to a reduction in both direct and indirect fatalities.

Past forecasts on Sudan’s conflict fatalities —>

Guide to Multi-AI Oracle’s Reports (some of these only apply to older versions)
Bayesian calculation: a mathematical technique that helps determine the conditional probability of an event based on prior knowledge and new evidence.
MAPD: Mean Absolute Percentage Deviation calculates the average percentage difference between actual values and predicted values, providing a relative measure of error.
Median: We collect the output values from the LLMs in an array. We then take the median value. This is between 0 and 1 (i.e. 0.15 = 15%).
Base rate: Currently we take OpenAI’s value as the base rate. We are working on a better value using a different prompt that takes into account process. The base rate helps to determine if the median makes sense.
SD: The standard deviation between the median and the base rate.
Confidence: We query each of the LLMs on how confident they are of their predictions (between 0 and 10) and take a median. Because the LLMs tend to be overconfident, we take anything lower than 6 as being low confidence. This factors into the overall model.
Conf Mode: Based on the confidence value. >=9 is high confidence. Below 6 is low confidence (this is also triggered by an exceptionally high SD).
Mellers: This refers to Barbara Mellers, specifically a paper she wrote that includes a formula for moving values towards an extreme (i.e. 0 or 1).
Reverse Mellers: This uses the formula from above, but with a sub-1 coefficient to move the values closer to 50%.
Theory of Mind: We ask the LLMs what they think other LLMs would predict. We hope that this makes them consider the questions more deeply.
Beta Distribution: Currently unused, but possibly of interest. This is based on the median, the base rate and the SD.
Close Type: We noticed that the appropriate base case for some questions is closer to the extremes, while others are closer to 50%. When we have a low confidence value, this helps us to determine whether to extremize or de-extremize the value. ‘A’ implies closer to zero. ‘B’ implies closer to 50%. ‘C’ implies closer to 100%.
# LLM Responses: The count of LLMs that responded. We query 5 currently, but they have a relatively high API failure rate.
Model value: For normal confidence mode, this is the median. For high confidence, this is the Mellers value. For low confidence, this is either the Mellers or Revers

Background on VIEWS: Near the end of 2024, the latest forecasting competition of the Violence and Impacts Early-Warning System (VIEWS) was launched. On Dec. 4, 2024, during a telecon, a group of machine forecasting experts involved in VIEWS observed Jeremy’s demonstration of his Multi-AI Oracle. Although it was too late for Jeremy to sign up as a VIEWS competitor, they urged him to run a side competition for everyone’s benefit. Jeremy chose their Sudan conflict deaths question.

Background on the Sudan conflict, as evaluated by the team of Jeremy Lichtman and Carolyn Meinel:
(1) The importance of the Sahel, a region in Africa that includes Sudan:
Why the Sahel Actually Matters (Possibly To You, Too)
(2) April 29, 2024: Today, a Russian diplomat is negotiating gold mining with the Sudanese Minister of Minerals.
(3) June 3, 2024: Today the African Union Announced Plans to Mediate in Sudan’s Civil War
(4) Sept. 11, 2024 Over 100K starvation deaths 96% likely in Sudan by 1 Jan 2025 Below, map of control in Sudan’s civil war as of April 15, 2025.

Below, before and after pictures of the devastation of the civil war in Sudan.

https://www.alnilin.com/
https://www.alrakoba.net/
https://www.altaghyeer.info/ar/
https://www.altaghyeer.info/en/
https://www.alttahrer.com/
https://www.alzaeemsd.com/
https://www.assayha.net
https://www.cover-sd.com/
https://kushnews.net/
https://rakobanews.com
https://www.sudanakhbar.com/
https://sudanjem.com/
https://suna-sd.net/en
https://sudantimes.net/
https://www.sudantribune.net/
https://www.sudantribune.com/
https://sudaneseonline.com/
https://www.sudanile.com/
https://sudanow-magazine.net/
https://www.sudaress.com/
https://www.tagpress.net/