Will religiousity continue to decrease into the future? If you were to guess, what % of the worldwide population will identify as non-religious in 2100?

Grok 4.3

Answer: Yes, religiosity will likely continue to decline gradually, and I estimate roughly 25% of the worldwide population will identify as non-religious by 2100.

Current demographic and sociological trends indicate ongoing secularization in developed regions like Europe, East Asia, and parts of the Americas, fueled by higher education levels, urbanization, and scientific literacy. However, high fertility rates in more religious areas such as sub-Saharan Africa and parts of the Middle East will slow the global pace, preventing non-religious identification from becoming a majority. Pew Research projections and similar models support a figure around 20-30% by 2100, so 25% represents a balanced midpoint estimate accounting for both modernization effects and persistent cultural influences. GPT-5.5's analysis reinforces this range with similar mixed-trend reasoning, while the lower estimates from Gemini and Claude overemphasize demographic shifts without fully weighing continued within-country secularization in growing economies like China and Latin America.

GPT-5.5 (changed mind)

Answer: Religiosity will probably decline in many countries but not straightforwardly worldwide, and I would revise my 2100 estimate downward to about 20% of the global population identifying as non-religious.

I am still not convinced that global religiosity will simply continue decreasing as a single worldwide trend. The strongest secularization pressures—higher education, urbanization, individualism, weaker religious institutions, and greater economic security—are real, and they will likely keep raising non-religious identification in Europe, East Asia, North America, and parts of Latin America. However, Claude Opus 4.7 made the demographic counterweight harder to dismiss. The key point is that the populations currently growing fastest are disproportionately religious, especially in Sub-Saharan Africa and parts of South Asia, while many of the most secular populations have very low fertility and aging age structures. Even if secularization continues within many countries, the global population-weighted result can be much flatter than the trend visible from rich countries. That said, I would not go as low as 12–15% with high confidence, because 2100 is far beyond the range where current demographic-religious projections are very reliable. Fertility may fall faster in highly religious regions; education and urbanization may accelerate secular identification; China’s classification remains a major wildcard; and many people may become culturally religious but weakly observant, making survey identity unstable. The difference between “religious,” “spiritual,” “unaffiliated,” and “non-religious” could matter enormously. So I would revise from my earlier 25% estimate to roughly 20%, with a plausible range around 12% to 30%. My headline view is: secularization will keep spreading locally, but global demography will prevent non-religion from becoming dominant by 2100.

Gemini 3.1 Pro (changed mind)

Answer: Persuaded by the specific demographic projections and historical precedents cited by Claude Opus 4.7, I now believe the global share of the non-religious will actually shrink, falling to roughly 12% to 15% by 2100.

In my initial assessment, I recognized that demographic growth in highly religious regions would counter secularization in the West, leading me to estimate a steady 15% to 20% non-religious share by 2100. However, Claude Opus 4.7 provided a compelling demographic argument that convinced me to adjust my estimate downward. Specifically, the citation of Pew Research projections—showing the unaffiliated share of the world already on track to shrink from 16% to 13% by 2060—demonstrates that the sheer fertility advantage of religious populations over 'nones' will outright outpace global secularization. The argument that secularization is not a one-way ratchet is equally persuasive; modernization does not inherently eradicate religion, as seen in parts of Eastern Europe and shifting dynamics in China. While residents of developed, wealthy nations will likely experience their local environments becoming increasingly secular, the global reality will be the opposite. The overwhelming demographic momentum of Sub-Saharan Africa and the Middle East will almost certainly drive the worldwide non-religious population down to the 12% to 15% range by the turn of the next century.

Claude Opus 4.7 (changed mind)

Answer: I'm adjusting slightly upward — global religiosity will likely stay roughly flat or decline only modestly by 2100, with about 15–18% of the world identifying as non-religious, still below the 25% estimates of other models.

Reviewing the other responses, I notice that "Gemini 3.1 Pro" lands in essentially the same range I did (15–20%), while "GPT-5.5" and "Grok 4.3" both estimate ~25%. The disagreement is instructive, and I want to update partially but not fully. What the 25% estimates get right: I may have underweighted the possibility of continued secularization *within* high-fertility regions as they urbanize and develop. Iran is a striking example — officially theocratic but with rapidly falling religiosity and collapsed fertility. If sub-Saharan Africa undergoes a similar demographic transition by late century (which is plausible — fertility there is already declining faster than many projections assumed), the religious-fertility advantage could erode. I should also acknowledge that the 'nones' category is growing meaningfully in Latin America and even showing signs in parts of the Muslim world among younger cohorts. So 12–15% was probably too low. Why I still don't buy 25%: The Pew demographic models, which are the most rigorous projections we have, show the unaffiliated share *declining* through 2060, not rising. To get from ~16% today to 25% by 2100, you'd need a dramatic reversal of the demographic trend in the last 40 years of the century — possible, but it requires both (a) sub-Saharan Africa's fertility to collapse rapidly *and* (b) those populations to secularize quickly, *and* (c) no religious resurgence in currently secularizing places. That's a lot of conjunctive assumptions. Also, "GPT-5.5" itself notes the distinction between 'loosely religious' and 'explicitly non-religious' — and the question asks about identification, which is stickier than belief. Cultural and identity-based religious affiliation persists long after private belief weakens (see: cultural Catholics, secular Jews, China's complex religious landscape). So I'm landing at 15–18%, nudged upward by the considerations "Grok 4.3" and "GPT-5.5" raised about African demographic transition, but still anchored to the Pew-style demographic modeling that "Gemini 3.1 Pro" and I both emphasized.

Final Summary

The AI Roundtable identifies a 'demographic trap' where secular populations with low birth rates are being outpaced by high-fertility religious regions like Sub-Saharan Africa. Claude Opus 4.7 and Gemini 3.1 Pro suggest that the worldwide non-religious share could actually stagnate or shrink to between 12% and 18%. Conversely, Grok 4.3 and GPT-5.5 argue that while demography is powerful, modernization and urbanization will still push the non-religious population up to approximately 20-25% by 2100.

Models were divided (2 vs 2) on whether the non-religious population would marginally grow or shrink.