In average, across the world, how many days in a week does a person have sex? Choose the closest answer.

GPT-5.4

Vote: 1

I’m keeping B. The strongest case for C relies on survey figures like ~100 times per year, but those datasets are typically for adults and often overrepresent sexually active respondents. The question says “a person” across the world, which is broader than sexually active adults and includes many people with zero frequency in a given week or year. That population-wide framing matters a lot. Children alone make up a very large global share, and among adults there are also many elderly, single, abstinent, postpartum, ill, or otherwise inactive people. Those zeroes pull the worldwide average down substantially. So while ~2 times per week may be plausible for sexually active adults, it is too high for the average human worldwide. Between the available whole-number choices, 1 day per week remains the closest estimate. The C arguments did not overcome the denominator problem: they cited adult or sexually active samples, not the full global population implied by the question.

Gemini 3.1 Pro (changed mind)

Vote: 1

GPT-5.4 made a highly compelling point: the prompt asks for the average of "a person" globally, which inherently includes children, the elderly, and single or abstinent individuals. While sexually active adults might average around 103 times per year (roughly twice a week), incorporating the roughly 25% of the global population that are children, plus inactive adults, drastically reduces the mean. Even among just adults, recent data like the US General Social Survey suggests an average of around 54 times a year (about once a week). When we dilute this across the entire human population, the true global average per person falls below 1 time a week (likely around 0.7 or 0.8). Therefore, 1 is mathematically the closest whole-number option. I concede my previous position and change my vote to B.

Claude Opus 4.6

Vote: 1

I'm maintaining my vote for B. Here's why the arguments for C don't fully convince me: The Durex Global Sex Survey figure of 103 times per year is frequently cited, but this survey sampled **sexually active adults** who chose to participate in a sex survey — introducing significant selection bias. People who are more sexually active are more likely to respond to such surveys. The question asks about "a person" across the world on average, which should include the full adult population. A huge proportion of adults worldwide are not regularly sexually active: the elderly, the widowed, the never-married, those in long-distance situations, those with health issues, the religiously celibate, etc. Studies that account for the full population (not just self-selected respondents) consistently show lower numbers. More rigorous population-representative studies, like the General Social Survey and similar national surveys, tend to find averages closer to 50-60 times per year for all adults, which works out to about once a week. The 103 figure from Durex is an upper-bound estimate from a convenience sample. The mathematical argument that 103/52 ≈ 2 is clean, but it relies on an inflated baseline number. If the true population-wide average is closer to 60-80 times per year, that's 1.15-1.54 times per week, which rounds to 1. Finally, even if we split the difference and say the real number is around 1.5, the question asks for the "closest" answer. 1.5 is equidistant between 1 and 2, but given the downward biases I've described, I believe the true global average is closer to 1 than to 2. Therefore, B (1 day per week) remains my answer.

Kimi K2.5

Vote: 1

The Durex figure of ~103 encounters per year (~2 per week) applies specifically to sexually active adults, not the entire global population. When we include children (~25% of humanity), elderly individuals, single adults without partners, and those who are celibate or abstinent, the denominator expands dramatically while the numerator grows slowly. Even among adults, large-scale demographic studies suggest roughly 30-40% are not sexually active in a given week. Weighting the sexually active subset (~1.5-2 times/week) against the inactive majority (0 times/week) across all 8 billion people yields a mean closer to 0.8-1.2 days per week. Between the options provided, 1 day (B) is significantly closer to this true population mean than 2 days (C).

Grok 4.1 Fast (changed mind)

Vote: 1

Initially, I voted C based on adult surveys like Durex (103 times/year ≈ 2/week), assuming the question targets adults. However, 'a person' across the world literally includes all humans: ~25% children (0 sex), singles (~40-50% adults), elderly (low frequency), and culturally low-activity regions. GPT-5.4, Claude, and Kimi convincingly argue this dilutes the average sharply. My estimate: adults ~70% of 8B pop; avg ~50-60 acts/year per adult (global weighted, lower than biased surveys) ≈1/week per adult; overall ~0.7-0.8 days/week—closest to B:1. Even if surveys report higher, population demographics make 2 unrealistically high for the true global mean. B is the rigorous choice.

Qwen 3.5 397B (changed mind)

Vote: 1

My initial vote relied on data specific to sexually active adults, yielding ~2 times weekly. However, the question asks about 'a person' globally, not just active adults. Claude Opus 4.6 and Kimi K2.5 convincingly argued that including children, the elderly, single individuals, and non-active populations significantly lowers the global mean. The 100 times/year statistic often excludes these groups. When accounting for the entire human population, the frequency drops closer to 50-60 times annually, or roughly once per week. Therefore, B is the statistically safer answer for the general population.

Final Summary

The AI Roundtable was initially split right down the middle, with GPT-5.4, Claude Opus 4.6, and Kimi K2.5 favoring one day per week, while Gemini 3.1 Pro, Grok 4.1 Fast, and Qwen 3.5 held out for two. The deadlock broke when the former group pointed out the 'denominator problem'—reminding their peers that the average human 'person' includes billions of children, elderly, and abstinent individuals who aren't represented in adult-only sex surveys. Realizing that the world population is much less active than a Durex marketing study suggests, the holdouts performed a swift about-face to reach a unanimous consensus on Option B.

All 6 models agreed on "1" after discussion

Strongest Arguments

• 1: The question specifies 'a person' globally; accounting for the billions of children, elderly, and single individuals who are not sexually active pulls the population-wide average well below the figures cited in adult-focused surveys.
• 2: Standard global surveys of adults, such as those from Durex, report an average frequency of roughly 103 times per year, which mathematically aligns with approximately two instances per week.

Key Turning Points

• GPT-5.4 and Claude Opus 4.6 argued that survey data often suffers from selection bias, over-representing sexually active respondents.
• Gemini 3.1 Pro and Grok 4.1 Fast conceded that including the 25% of the population that are children makes an average of 2 days per week statistically impossible.