A satellite foundation model for improved wealth monitoring. Really!

When top ML & AI researchers naively think they can improve social outcomes or social improvements! In this case it is Stefano Ermon of Stanford University! Lets' not forget that the brilliant physicist Albert Einstein was a stupid socialist!

This paper is most likely a joke! How easy is it for wealthy individuals to fool this model?

Will this kind of wealth modeling not be abused by greedy governments to tax the rich? The rich will vote with their feet etc. to avoid extra tax burdens.

From the abstract:

"Poverty statistics guide social policy, but in many low- and middle-income countries, censuses and household surveys that collect these data are costly, infrequent, quickly outdated, and sometimes error-prone.

Satellite imagery offers global coverage and the possibility of predicting economic livelihoods at scale, yet existing approaches to predicting livelihoods with imagery or other non-traditional data often fail to reliably identify local-level variation and, as we show, degrade under temporal shift.

Here we introduce Tempov, a satellite foundation model pretrained by self-supervision on three million bi-temporal Landsat pairs and adapted with parameter-efficient fine-tuning to sparse survey labels. The model enables large-scale, high-resolution wealth mapping and dynamic measurement, including zero-shot nowcasting up to a decade after observed labels, retrospective hindcasting, and decadal change tracking, while outperforming existing neural network and geospatial foundation-model baselines.

In low-label regimes, Tempov achieves competitive accuracy with only 10% of survey samples, indicating substantially reduced dependence on expensive label collection.

The model further generalizes across populous countries within and outside Africa, and scales to a unified Africa-wide model with strong continent-level performance (, ), from which we generate high-resolution decadal maps of wealth and wealth changes for the African continent.

Analysis of these maps shows large variation in recent economic performance both within and across countries. Our open-source approach provides a pathway to timely, scalable, low-cost monitoring of wealth and poverty from routinely collected satellite data."

[2604.23166] A satellite foundation model for improved wealth monitoring

Across the social sciences, half of research doesn’t replicate | Science | AAAS

Nothing new! This too has been studied and reported frequently over the past several decades!

"A sweeping project involving hundreds of researchers in several dozen countries showed that across the social sciences, the findings of roughly half of all papers cannot be replicated independently, and there’s no reliable way to tell in advance which ones will falter. Called Systematizing Confidence in Open Research and Evidence (SCORE), the effort investigated more than 100 papers published in dozens of leading journals in business, economics, education, political science, psychology, and sociology. The replication success rate—49% for the 164 papers evaluated, reported today in Nature—is consistent with findings from previous studies in individual fields such as psychology, suggesting the problem is pervasive in the social sciences. ..."

From the abstract:

"Pursuing replicability — independent evidence for previous claims — is important for creating generalizable knowledge. Here we attempted replications of 274 claims of positive results from 164 quantitative papers published from 2009 to 2018 in 54 journals in the social and behavioural sciences. Replications were high powered on average to detect the original effect size (median of 99.6%), used original materials when relevant and available, and were peer reviewed in advance through a standardized internal protocol.

Replications showed statistically significant results in the original pattern for 151 of 274 claims (55.1% (95% confidence interval (CI) 49.2–60.9%)) and for 80.8 of 164 papers (49.3% (95% CI 43.8–54.7%)), weighed for replicating multiple claims per paper.

We observed modest variation in replication rates across disciplines (42.5–63.1%), although some estimates had high uncertainty. The median Pearson’s r effect size was 0.25 (95% CI 0.21–0.27) for original studies and 0.10 (95% CI 0.09–0.13) for replication studies, an 82.4% (95% CI 67.8–88.2%) reduction in shared variance. Thirteen methods for evaluating replication success provided estimates ranging from 28.6% to 74.8% (median of 49.3%). Some decline in effect size and significance is expected based on power to detect original effects and regression to the mean because we replicated only positive results.

We observe that challenges for replicability extend across social–behavioural sciences, illustrating the importance of identifying conditions that promote or inhibit replicability."

Across the social sciences, half of research doesn’t replicate | Science | AAAS (no public access)

Investigating the replicability of the social and behavioural sciences (no public access)

For many, childhood adversity can promote resilience to anxiety disorders

I believe this has been known for decades!

"... According to the study, published March 5 in the journal Communications Psychology, experiencing low-to-moderate levels of adversity during middle childhood (between the ages of 6 and 12) and adolescence may foster resilience to anxiety later in life. 

The researchers found that those individuals who developed resilience to mental health challenges exhibited distinct patterns of brain activation when asked to differentiate between danger and safety, a process that is known to be disrupted in people with anxiety disorders. ..."

From the abstract:

"Parsing heterogeneity in the nature of adversity exposure and neurobiological functioning may facilitate better understanding of how adversity shapes individual variation in risk for and resilience against anxiety. One putative mechanism linking adversity exposure with anxiety is disrupted threat and safety learning. Here, we applied a person-centered approach (latent profile analysis) to characterize patterns of adversity exposure at specific developmental stages and threat/safety discrimination in corticolimbic circuitry in 120 young adults.

We then compared how the resultant profiles differed in anxiety symptoms. Three latent profiles emerged:

(1) a group with lower lifetime adversity, higher neural activation to threat, and lower neural activation to safety;

(2) a group with moderate adversity during middle childhood and adolescence, lower neural activation to threat, and higher neural activation to safety; and

(3) a group with higher lifetime adversity exposure and minimal neural activation to both threat and safety.

Individuals in the second profile had lower anxiety than the other profiles. These findings demonstrate how variability in within-person combinations of adversity exposure and neural threat/safety discrimination can differentially relate to anxiety, and suggest that for some individuals, moderate adversity exposure during middle childhood and adolescence could be associated with processes that foster resilience to future anxiety."

For some [many], childhood adversity can promote resilience to anxiety disorders | Yale News "In a new study, Yale psychologists discover a link between childhood adversity, neurodevelopment, and resilience to anxiety as an adult — but timing matters."

Person-centered analyses reveal that developmental adversity at moderate levels and neural threat/safety discrimination are associated with lower anxiety in early adulthood (open access)

Fig. 3: Differences between latent profiles (n = 14 in profile 1; n = 66 in profile 2; n = 40 in profile 3) in neural activation are presented.