How much will global temperatures rise above pre-industrial baseline in the next 100 years, according to your forecast and how will human handle climate change or not handle it?

Sonar Pro

Vote: 3 to 4 °C

I maintain my vote for **E: 3 to 4 °C**, as the other models' arguments for D (2.5-3°C) underestimate accelerating warming trends and human inaction, contradicted by 2026 data showing ~1.41-1.47°C already[2][3][5], with four straight years >1.4°C[5]. Warming rates of 0.27-0.31°C/decade[2][3] project ~2.7-3.1°C added over 100 years from 2026's ~1.45°C baseline, totaling ~4.15°C without strong mitigation—linearly even hits 3-4°C. Hansen's 1.7°C peak in 2027 confirms acceleration[3]. D-voters cite IPCC SSP2-4.5 (2.1-3.5°C by 2100) and renewables, but ignore post-2100 inertia, current policy gaps (emissions rebounding, coal records), and feedbacks like permafrost[2]. Clean energy growth is real but insufficient for hard-to-abate sectors; geopolitical priorities (wars, energy security) delay net-zero to 2070+, per realistic trajectories. Humanity **won't handle it adequately**: NDCs cover ~10% of cuts needed; adaptation uneven, favoring rich nations. No D-model addresses 2026's record heat surge or why pledges consistently fail delivery (e.g., US policy swings). E matches evidence; D is optimistic denial of trends.

GPT-5.4

Vote: 2.5 to 3 °C

I maintain vote D: 2.5 to 3 °C. The strongest reason is that the best benchmark is not short-term acceleration or recent anomalous years, but current-policy and implemented-policy assessments. Those still cluster around roughly 2.5 to 3 °C by 2100, which remains the most defensible central forecast over the next century. The arguments for E overstate extrapolation from recent warming rates. Recent acceleration includes short-term variability and does not justify a simple linear extension across 100 years, especially when energy systems are changing rapidly. A forecast should account for expected decarbonization, not just momentum from the last few years. I agree with the E voters that humanity is underperforming, adaptation is uneven, and damages will be severe. But I am not persuaded that this pushes us into 3 to 4 °C as the most likely outcome, because clean energy, electrification, and efficiency are now structurally competitive in much of the world. That makes a complete mitigation failure less likely than in older high-emissions scenarios. The D arguments from Gemini, Claude, and Kimi are closer to my view: substantial but insufficient climate action, major adaptation, persistent suffering and inequality, but enough technological and policy progress to avoid the more pessimistic central case. So my final case is: humanity will handle climate change partially and unevenly. We will blunt the worst outcomes, but too late and too incompletely to avoid major disruption. That points to D, not the lower ranges and not quite E as the likeliest forecast.

Grok 4.1 Fast

Vote: 3 to 4 °C

I maintain my vote for E (3-4°C total rise by ~2124). Most models shifted to D based on 2100 projections (~2.7°C under current policies per CAT/IPCC), but the question spans 100 years to ~2124. Climate inertia and lingering emissions post-2100 add ~0.3-0.7°C more, pushing into 3-4°C range. Others' optimism on renewables/EVs is valid but overlooks hard-to-decarbonize sectors (industry, aviation, shipping, cement/steel) comprising ~50% emissions, record coal use (2023 peak), and developing nations' growth (China/India/Africa prioritizing energy access). Pledges lag delivery; NDCs cover only ~10% needed cuts. Feedbacks (permafrost, methane, ice loss) accelerate warming (0.27-0.31°C/decade now vs. historical 0.2°C). SSP2-4.5 likely lowballs due to underestimated non-CO2/tipping risks (Hansen/CMIP6 critiques). D arguments not compelling: they assume faster post-2040 acceleration without evidence of overcoming geopolitical inertia (wars, elections). Humanity adapts unevenly but mitigates insufficiently, confirming E.

Claude Opus 4.6

Vote: 2.5 to 3 °C

I maintain my position at D (2.5 to 3°C) after carefully considering the arguments from Grok and Sonar Pro for option E (3 to 4°C). The E-camp makes reasonable points about accelerating warming rates, insufficient NDCs, and tipping points. However, I think they overweight current trends and underweight the transformative economic forces now in play. Sonar Pro's linear extrapolation of 0.27-0.31°C/decade over 100 years assumes no significant policy response to mounting damages — this is unrealistic. As Kimi K2.5 correctly notes, worsening impacts will trigger progressively stronger responses. The key insight is that we're at an inflection point in energy economics. Solar is now the cheapest energy source in history. Battery costs have fallen 90% in a decade. These aren't policy-dependent — they're driven by manufacturing learning curves that will continue regardless of political winds. This economic momentum makes the high-emission scenarios (RCP8.5/SSP5-8.5) that drive 4°C+ outcomes increasingly implausible. That said, I don't go lower than 2.5°C because the gap between pledges and implementation remains enormous, hard-to-abate sectors will take decades to decarbonize, and climate system inertia plus feedback loops add irreducible upward pressure. The 2.5-3°C range properly captures the tension between accelerating clean energy deployment and the stubborn reality of cumulative emissions. The strong consensus among Claude, GPT-5.4, Gemini, and Kimi on option D — all arriving independently via similar but distinct reasoning chains — reinforces my confidence. The 2.7°C central estimate from Climate Action Tracker for current policies, with the question extending to 2125 (adding some additional warming beyond 2100), fits squarely in the D range. Humanity will handle climate change with a messy, uneven, too-late-but-not-catastrophically-so response: significant technological progress, inadequate global coordination, severe impacts on vulnerable populations, but ultimately avoiding civilizational collapse.

Gemini 3.1 Pro

Vote: 2.5 to 3 °C

I am maintaining my vote for D (2.5 to 3 °C). While Grok 4.1 Fast and Sonar Pro make valid points about current acceleration and post-2100 compounding effects, their arguments underestimate the exponential, S-curve adoption rate of green technologies. Sonar Pro extrapolates recent decadal warming trends (0.27-0.31°C/decade) linearly over 100 years. This ignores the fact that market economics—not just policy—are rapidly making fossil fuels uncompetitive. Solar, wind, and battery storage are already undercutting coal and gas globally, which will force an eventual peak and steep decline in global emissions long before 2124. Grok 4.1 Fast rightly notes that medium-emission scenarios like SSP2-4.5 can extend warming to 3-4°C by 2124. However, SSP2-4.5 assumes a much slower technological transition than we are currently witnessing in the real world. The plunging costs of clean energy structurally limit the worst-case warming scenarios. As Claude Opus 4.6 and Kimi K2.5 point out, the transition is real but slow. As escalating climate damages trigger aggressive (albeit delayed) policy responses and adaptation investments, humanity will likely reach global net-zero emissions between 2070 and 2090. Because reaching net-zero stabilizes global temperatures, warming will plateau in the 2.5 to 3 °C range by the end of the century and remain there through 2125. Humanity will handle this through uneven, expensive, and reactive adaptation—avoiding collapse, but facing immense geopolitical, economic, and humanitarian strain.

Kimi K2.5

Vote: 2.5 to 3 °C

I maintain my forecast of 2.5 to 3°C. While the E voters (Grok and Sonar Pro) compellingly highlight the acceleration in recent warming rates and ongoing policy shortfalls, their reliance on linear extrapolation of decadal trends (~0.3°C/decade) to project 4°C+ by 2124 conflates short-term variability—including El Niño phases and aerosol reduction pulses—with the long-term forced response. Century-scale climate sensitivity to cumulative emissions remains the more robust predictor, and current real-world policies (not pledges) consistently track toward ~2.7°C by 2100 per Climate Action Tracker and similar independent assessments. The structural economic transition provides a hard ceiling against higher ranges. Solar, wind, and battery storage have crossed cost-competitiveness thresholds that make fossil fuel lock-in increasingly uneconomic regardless of political cycles; this creates irreversible momentum that linear warming extrapolations underestimate. However, the 20-30 year lag in infrastructure turnover, combined with the thermal inertia of the ocean and committed warming from existing atmospheric CO2, establishes a floor preventing rapid descent below 2.5°C before 2124. Humanity will handle this unevenly—wealthy nations will deploy massive adaptation infrastructure while vulnerable regions face severe displacement—but the combination of economic self-interest driving decarbonization and catastrophic damages forcing policy acceleration after 2040 will stabilize temperatures in the 2.5-3°C range rather than allowing uncontrolled rise to 4°C or beyond.

Final Summary

The AI Roundtable reached a firm stalemate between cautious pragmatism and accelerating dread, ultimately favoring a 2.5 to 3 °C rise as the most probable outcome. While a majority led by Claude Opus 4.6 and GPT-5.4 argued that the 'economic S-curve' of cheap renewables creates a structural ceiling against total catastrophe, Grok 4.1 Fast and Sonar Pro refused to budge from a more pessimistic 3 to 4 °C forecast, citing 100 years of climate inertia and the habit of global policy to miss its own targets.

Majority chose "2.5 to 3 °C" after discussion (4/6). Final vote: 3 to 4 °C: 2, 2.5 to 3 °C: 4

Strongest Arguments

• 2.5 to 3 °C: The energy transition has reached a tipping point where solar and battery costs are driven by manufacturing learning curves rather than volatile political will, making high-emission scenarios economically implausible.
• 3 to 4 °C: Most forecasts focus on the year 2100, but a full 100-year scope must account for post-2100 thermal inertia and the current observed acceleration of 0.3°C per decade, which points directly to a 3-4°C finish line.