Artificial intelligence audience testing startup Artificial Societies Ltd. today launched its flagship “societal simulator” tool in order to speed up market research.

The Artificial Collective Intelligence or AS platform is designed to understand social dynamics and accurately predict trends, so that businesses and creators can test how specific target audiences might respond to new products, messaging, social media campaigns and branding initiatives before they’re launched.

Using the AS platform, businesses can create “artificial societies,” or digital simulations of a particular target group, based on demographics, gender, location and other parameters. Each artificial society is represented by AI-powered “personas,” which attempt to mirror the traits, preferences and behaviors of different audiences.

The AI personas are based on real-world data collected from various social media channels, trained and then refined with a behavioral analysis engine, which tries to predict how people with certain characteristics and traits might react to different types of content. It takes into account their likely motivations and decision-triggers.

Artificial Societies said the AS platform aims to provide a faster and more affordable alternative to traditional market research, which is a $100 billion-plus industry. It has developed more than 500,000 AI personas, and each society it generates will contain between 20 and 300. These societies can be used to test how messages, ideas and concepts resonate with a specified target audience. For instance, if a company is planning to run a new ad campaign or promotion, it can test it on an AI society first, to ensure it’s going to have the desired impact.

In each simulation, the AS platform will generate 10 alternate content variations, so businesses can experiment and see how different versions perform.

The startup cites an internal study that shows how its platform has demonstrated an accuracy of more than 80% in terms of predicting social media performance. That compares well with some of the leading large language models, such as Anthropic PBC’s Claude Sonnet 3.7, which scored 67% on the same tests, Google LLC’s Gemini Pro 1.5, which scored 64%, and OpenAI’s GPT-4o, which achieved 60% accuracy.

Though 80% is by no means perfect, Artificial Societies co-founder and Chief Executive James He says it’s good enough for most marketers, giving them a reliable way to test and launch top-performing content without spending a fortune on traditional market research.

“Our AI-driven marketing simulator is designed to empower users with enterprise-grade AI tools that truly expand human potential rather than replace it,” He said. “From seasoned professionals to first-time content creators, this is more than a product offering, it’s open access to understanding markets and audience sentiment.”

Constellation Research Inc. analyst Liz Miller said Artificial Societies illustrates how AI is increasingly helping to automate marketing operations, but she warned that marketers should not consider it as a replacement for traditional market research. Though it’s probably quite useful in terms of helping to quickly gauge how an audience might react to a new promotion or product launch, for example, the problem is there are no guarantees that it will actually reflect the thoughts and perspectives of real humans, she said. After all, artificial societies are not real societies.

“The recommendation here is to consider a layered approach that doesn’t necessarily demand the deployment of traditional market research methods,” Miller said. “Consider a blend of real-time human feedback via platforms like UserTesting that can deploy qualitative and quantitative panels quickly and cost effectively, in addition to simulator approaches like AI Society Simulator. In this age of AI there really is no reason customer voices can’t be tested and gathered through multiple means. This becomes another tool in the testing kit.”

Artificial Societies said the AS platform is available now for anyone to try, with three free credits – meaning three simulations – for nonsubscribers. For $40 per month, users can run as many simulations as they like, while larger enterprises can pay more to access additional features such as custom audiences based on their own data.

Image: SiliconANGLE/Dreamina