Voices
Fifteen takes that BCG either omits, soft-pedals, or rolls into a single scenario. One paragraph each. Cited in sources/SOURCES.md.
Institutional
NIC, Global Trends 2040 (2021). Five scenarios: Renaissance of Democracies (US-led tech revival, authoritarians stagnate); A World Adrift (multipolar chaos, no consensus); Competitive Coexistence (US-China rivalry stabilizes); Separate Silos (globalization fractures into regional blocs); Tragedy and Mobilization (a global food/environment catastrophe in late 2030s triggers bottom-up reform, NGOs and multilaterals gain unprecedented standard-setting power). NIC takes the catastrophe-trigger seriously as a scenario; BCG only as a risk in Climate Coalition.
WEF, Global Risks 2026. Framing: “age of disorder.” Top 10-year risks dominated by environment (5 of top 10). AI moves from 30th (short term) to 5th (ten year) — the largest single-issue jump. Inequality named the most-interconnected risk. 57% of respondents see “turbulent or stormy” decade; ~20% expect “global catastrophic” outcomes. Whereas BCG models AI as net-positive, WEF’s expert panel puts adverse-AI-outcomes in the top 5.
OECD, Long-Run Scenarios 2025. Baseline trend growth for OECD+G20 declines from ~3% to 1.5% by 2060. China’s potential growth falls below US by 2040–45, below zero by 2070–75. India and sub-Saharan Africa are the two regions gaining global output share; SSA goes from 3% of world output to ~13% by 2100. OECD agrees with Zeihan on the demographic gradient but rejects collapse; agrees with McKinsey that growth slows but rejects “near-halving.”
McKinsey Global Institute. Demographic peak around 2050; employment growth just 0.3%/yr. To keep GDP growth at 2.1%/yr, productivity must hit 3.3%/yr — 80% above historical. Sector studies show 4%/yr is achievable but not on autopilot. Implication: BCG’s “AI Abundance” 5% growth requires productivity that nobody has yet delivered for 50 years running.
IPCC SSPs. Five socioeconomic pathways. SSP1 (sustainability, low pop, low energy) and SSP5 (fossil-fueled growth, high tech) both peak global population at ~8.5B around 2050–60 then decline. SSP3 (regional rivalry) keeps growing to 12.6B by 2100. BCG’s four scenarios all sit between SSP1 and SSP2; SSP3 (Battling Blocs–like with high population) and SSP5 (no climate action, fossil-fueled abundance) are missing from BCG entirely.
UN WPP 2024. World population 9.7B by 2050 (from 8.2B in 2024). Fertility 2.25 → 2.07 by 2050 — at replacement. By mid-2030s, those 80+ outnumber infants globally. Sub-Saharan Africa population rises 79% to 2.2B by 2054. The aging structure is locked in; the only variable is whether productivity compensates.
IEA NZE2050. Renewables 4× by 2035; electricity hits 50% of final energy and ⅔ of useful energy by 2050; warming peaks 1.65°C and falls to 1.5°C only via large-scale carbon removal that doesn’t yet exist at scale. IEA’s pathway is normative — it shows what 1.5°C requires, not what it predicts. Smil thinks it’s impossible at this speed; Handmer thinks it’s slow.
Technologists
Daniel Kokotajlo et al., AI 2027. Month-by-month forecast: AI agents 2026; coding automated early 2027; intelligence explosion late 2027. >1M readers including the US VP. After 2026 reception, Kokotajlo updated his superintelligence median to 2029 (originally 2027). If anything close to this is right, BCG’s “AI Abundance” is too slow and too cooperative — the takeoff is over before geopolitical accommodation can be negotiated.
Marc Andreessen, Techno-Optimist Manifesto (Oct 2023). 5,200 words. “We believe in accelerationism.” Climbing the Kardashev gradient — civilization defined by energy throughput. AI regulation is treated as the enemy. Functionally aligned with BCG’s “Digital Darwinism” but presented as desirable rather than as a downside.
Casey Handmer, Terraform Industries. Solar PV cost trajectory means ~95% of human energy use is downstream of solar PV by 2042. Synthetic methane from solar + air + water becomes cheaper than fossil methane. Implies energy ceases to be a binding constraint on civilization within 20 years. Compared to IEA NZE2050: faster, deeper, and driven by economics not policy. None of BCG’s four scenarios assume this.
Robin Hanson, Age of Em. Brain-emulation economy: ems run subjectively faster than humans, work at near-subsistence wages, doubling-time on the order of months. Real GDP growth on the order of 1,000% per year for periods. Even if AI doesn’t deliver Kokotajlo’s takeoff, something like an em economy is a separately-arguable scenario that BCG and OECD both omit.
Demographer-contrarians
Peter Zeihan, End of the World Is Just the Beginning. Globalization was a US security order, and the US is withdrawing. Demographics are the deeper destiny: Chinese population could halve by 2050; aging Germany, Japan, Korea, Italy lose the workers to sustain consumption. US is uniquely positioned (geography, energy, demographics). Net: a Battling-Blocs-style world but more chaotic, with mass famine in import-dependent regions. BCG’s Battling Blocs assumes orderly bloc formation; Zeihan thinks the order itself breaks.
Climate
Vaclav Smil, Halfway Between Kyoto and 2050. Materially, NZE2050 requires 1.7B tons of green steel/yr (91M tons of green hydrogen just for steel), $9.2T/yr energy investment (≈10% of world GDP for 30 years). Historical energy transitions take 50–60 years from 10% market share to dominance; oil did. Smil’s conclusion: net-zero by 2050 is “low probability if not impossibility.” Climate Coalition is therefore too optimistic on speed, not too pessimistic.
Jason Hickel, degrowth. UN baseline implies global resource use 85B tons today → 132B+ tons by 2050. Only scenarios that meet 1.5°C cut absolute resource use. Hickel: degrowth is reduced consumption-by-elites + redirection to social goals. Postcapitalism by chosen path, not by collapse. BCG models no scenario where rich countries voluntarily shrink GDP per capita.
Existential-risk frame
Toby Ord, The Precipice (2020). Existential risk this century: ~1 in 6. Of that, AI misalignment alone is ~1 in 10 — bigger than all other sources combined. Climate change, nuclear war, engineered pandemics fill out the rest. The 2050 question becomes: what’s the conditional probability we get there? BCG’s “Battling Blocs” assumes great-power tensions but not catastrophe; Ord’s frame says the catastrophe distribution is fat-tailed and the rational response is risk reduction, not scenario planning around the median.