AI Threat Assessment · 26 May 2026

Good Business Lab

Development Economics Research
STILL BREATHING
3.8/ 10

“Good Business Lab spent a decade building the most rigorous randomized controlled trials in development economics — the gold standard for proving whether microcredit actually works or if that new farming technique really helps smallholders. Their PhD economists run experiments so methodologically pristine that peer reviewers weep with joy. The problem is that Claude can now analyze their publicly available datasets, replicate their statistical models, and generate policy recommendations without the 18-month fieldwork timeline or the $200K grant budget that funds the cricket sponsorship.”

Business Model
6.5
Automation Risk
7.0
Moat Strength
2.5
Adaptability
4.0
Need Survival
3.0
AI Threat Level
BUSINESS MODEL REPLACEABILITY

Their value proposition was 'we design experiments that prove causation, not just correlation' — which commanded premium consulting fees until GPT-4 started running power calculations and suggesting randomization strategies from a prompt. The experimental design survives; the monopoly on knowing how to do it doesn't.

6.5
WORKFORCE AUTOMATION RISK

Data analysis, literature reviews, and statistical modeling are gone — Claude writes the regression code, runs the robustness checks, and formats the tables for publication with the thoroughness of a Berkeley PhD who never needs coffee breaks or sabbaticals.

7.0
MOAT STRENGTH

The real moat is institutional trust: governments and NGOs pay Good Business Lab because ministers and program officers trust their signatures on impact evaluations in ways they'll never trust an AI output. This is a genuine trust-era brand moat in high-stakes policy decisions — but only until AI proves it can run cleaner experiments with larger sample sizes and catch the methodological errors human researchers miss.

2.5
AI ADAPTABILITY SIGNALS

Their research publications mention machine learning applications in development work, but their job postings still read like they're hiring for 2015 — 'PhD in Economics preferred, R programming skills required' — as if the future of evidence generation looks exactly like the past, but with better laptops.

4.0
WILL THE NEED SURVIVE AI?

Rigorous impact evaluation absolutely survives — governments still need to know if programs work before spending billions. But AI doesn't just speed up the analysis; it can design better experiments, catch confounding variables human researchers miss, and run simulations at scales impossible for field teams.

3.0
Verdict

“AI won't eliminate development economics research — it will just make the humans realize they've been charging premium rates for statistical analysis that a well-prompted LLM does more thoroughly, faster, and without the grant application paperwork. The institutional trust buys them a decade; the methodology monopoly was already gone before they noticed.”

Scores are based on public information and AI analysis. This is an affectionate roast, not a financial assessment. The best companies use this as a mirror, not a verdict.

Roast another →