- Subconscious Insights
- Posts
- Synthetic Data vs. Human Data: How AI is Outpacing Human Insights in Market Research
Synthetic Data vs. Human Data: How AI is Outpacing Human Insights in Market Research

In the article, “What Generative AI Reveals About the Human Mind,” TIME created a salient comparison of AI’s data prediction abilities and the human brain’s action-based learning. The fact is, we still don't know what Language Models are capable of, or even why certain behaviors emerge. However, we have discovered that for many use cases, simulated consumer behaviors are indistinguishable from human behaviors - with some meaningful differences.
Some models estimate prices much better than others. Some models simulate social decisions much better. Subconscious AI runs behavioral surveys on synthetic respondents to get useful, lightning fast results that are testably near-identical to how human beings would respond.
But, how do we do it? Is it reliable? And of course, can synthetic data really replace humans in market research studies?
Why now is the time for Synthetic Data:

People have been trying to do this for years, so why does it work now? The introduction of LLMs represents a transformative inflection point in synthetic data generation. In contrast to historical synthetic data models, Large Language Models can consider a much broader range of data points and circumstances, both explicitly and implicitly, that allow them to capture nuances and complexities in data.
This approach allows LLMs to more accurately predict specific outcomes, like how a 25-year-old woman in Dallas might vote on electric vehicle-related legislation, by considering factors that traditional models might overlook, such as proximity to charging stations or daily commute details.
Finding the most viable respondents can be a difficult process in some markets. With Subconscious.ai, we generate a synthetic population that mirrors your target audience - ready to deliver insights at scale with accuracy and precision around the “why” of human behavior. Real people are hard to find, but with LLM-generated respondents, your research never has to slow down. We increase the velocity of running sophisticated quantitative experiments from months to minutes.
Can synthetic data really replace humans?
Some estimates assume panel provided data is already 35% synthetic (with the price of human). Synthetic data can solve each of these problems, not as a replacement to humans, but as a supplement that's good enough to save your business time and money.
When humans are used to predict human behavior, about 80% of human behavior is explainable by the best-designed experiments in the world. When Subconscious AI is used to predict human behavior, about 75% of human behavior is explainable by our causal models. This means we can explain ~95% of a human study for the purposes of market research, at ~1,000x lower cost.
Synthetic vs. human baseline correlation over time
We have compared thousands of published studies to simulate results using the most state-of-the-art models. By combining these insights, we are augmenting human market researchers by offering more nuanced and comprehensive insights at fractions of the cost.
The Say-do gap refers to the discrepancy between what humans say they will do vs. what they actually end up doing.
Our approach provides:
1. High Precision: Simulating human behavior with near-human accuracy.
2. Efficiency: Faster insights without extensive timelines.
3. Cost-Effectiveness: Achieving high-quality results at a fraction of the cost.
This capability opens new avenues for understanding consumer behavior, designing effective policies, and conducting impactful research. Whether predicting market trends or crafting data-driven policies, our technology offers a reliable solution.
Interested in learning more about synthetic data, and how your team can use it for market research? Let’s talk!
If you’d like to try out the Subconscious.ai platform for your market research needs, you can run 2 free experiments by clicking “Get Started” here.