Simile Raises $100 Million for AI That Aims to Predict Human Behavior

That’s according to a report Thursday (Feb. 12) from Bloomberg News, which says the financing will help Simile develop AI tools designed to help companies predict human behavior. That includes guessing which products consumers might buy, or questions that analysts might pose during company earnings calls.

The report says Simile had emerged from stealth this week following months of developing an AI model trained on interviews with hundreds of people about their lives. The company’s system was also given data on historic transactions and text from scientific journals dealing with behavioral experiments.

The company says it hopes to use this technology to foresee decisions someone might make in any given situation. It does that with simulations populated by AI agents that represent the preferences of actual people.

Bloomberg notes that the company’s approach might give businesses another option for learning consumer preferences aside from turning to focus groups. Simile says CVS has been testing its service to help guide decisions about which products to stock and display.

As PYMNTS wrote earlier this week, the question of whether AI can grasp the values that guide human choices has become more pressing as artificial intelligence systems become a greater part of daily life and decision-making.

“Traditional AI is engineered to optimize efficiency, accuracy and speed,” that report said. “Humans, by contrast, make decisions shaped by culture, social norms and lived experience, factors that are rarely captured in algorithmic design.”

The report also points to new research from the University of Washington suggesting that machines could learn values by seeing how people behave in different cultural contexts. This approach may bring about AI that is not only more contextually aware but also more trustworthy to diverse groups of users.

In other AI news, PYMNTS wrote recently about research showing that conversational AI is increasingly supplanting, and not complementing, traditional search in a variety of categories, including shopping and financial workflows.

For example, data from PYMNTS Intelligence showed that so-called “power users” — the people who perform the most AI-driven tasks — were almost a third more likely in December than in November to say they had replaced old search methods with AI-driven alternatives.

“Among mainstream users, replacement rose from 22% to 30% month over month. Light users remained largely unchanged at roughly 11%,” the report added. “The same dynamic appears in purchasing behavior. Nearly half of power users, or 48%, report supplanting old approaches with AI-driven alternatives. Mainstream users saw replacement climb from 22% to 32%.”